Conference Agenda

The European Ground Motion Service (EGMS) constitutes the first application of high-resolution monitoring of ground deformation for the Copernicus Participating States. It provides valuable information on geohazards and human-induced deformation thanks to the interferometric analysis of Sentinel-1 radar images. This challenging initiative constitutes the first ground motion public dataset, open and available for various applications and studies.
The subject of this abstract is to validate all EGMS products (Basic, Calibrated and Ortho) in terms of spatial coverage and density of measurement points. A total of twelve sites have been selected for this activity, covering various areas of Europe, as well as representing equally the EGMS data processing entities. To measure the quality of the point density we employ open land cover data to evaluate the density per class. Furthermore, we propose statistical parameters associated with the data processing and timeseries estimation to ensure they are consistent.
The usability criteria to be evaluated concern the completeness of the product, its consistency, and the pointwise quality measures. Ensuring the completeness and consistency of the EGMS product is essential to its effective use. To achieve completeness, it is important to ensure that the data gaps and density measurements are consistent with the land cover classes that are prone to landscape variation. Consistency is also vital for point density across the same land cover class for different regions. For instance, urban classes will have higher density than farming grounds, and this density should be consistent between the ascending and descending products. Pointwise quality measures are critical in assessing the quality of the EGMS PSI results. For example, the temporal coherence is expected to be higher in urban classes, and the root-mean-square error should be lower. Overall, these measures and standards are crucial in ensuring the usefulness and reliability of the EGMS product for a wide range of applications, including environmental management, urban planning, and disaster response.
For the validation of point density, a dataset of 12 selected sites across Europe is used, representing the four processing entities (TRE Altamira, GAF, e-GEOSS, NORCE). The aim of the point density validation activity is to ensure consistency across the EU territories by comparing the point density at three sites for each algorithm, one of which is in a rural mountainous area and the other two are urban. The dataset is obtained directly from the Copernicus Land – Urban Atlas 2018 and contains validated Urban Atlas data with the different land cover classes polygons, along with metadata and quality information. We have extensive Urban Atlas (version 2018) verified datasets on the cities of Barcelona/Bucharest (covered by TRE Altamira), Bologna/Sofia (covered by e-GEOSS), Stockholm/Warsaw (covered by NORCE) and Brussels/Bratislava (covered by GAF). In parallel we select four different rural and mountainous areas to analyse more challenging scenarios as well for the four processing chains of the providers.
There are 27 different land cover classes defined in Urban Atlas. To facilitate the analysis and the interpretation of the results, we aggregate and present our findings for each of the main CLC groups: Artificial Surfaces, Forest and seminatural areas, Agricultural areas, Wetlands and Water bodies.
For the validation measures, key performance indices (KPI) are calculated, with values between 0 and 1. We normalise the estimated density values for each service provider with respect to the highest value for Artificial surfaces, Agricultural areas and Forest and seminatural areas. Users expect consistent and good densities in these classes, specifically in the Artificial surfaces. And the lowest value for Wetlands and Water bodies. This will enable outlier detection since the applied algorithms should barely produce any measurement points on these surfaces.
Regarding the pre-processing of the data from EGMS, one of the challenges was the overlapping of bursts from different Sentinel-1 satellite tracks. If all bursts were included in the analysis, areas with more track overlaps would result in a higher point density, creating a bias in the data. To address this issue, a custom algorithm was designed to identify and extract the unique, non-overlapping polygon for each burst. This iterative algorithm was specifically designed to ensure a fair comparison among different areas, and to eliminate any biases that could impact the results of the analysis.
In conclusion, as an open and freely available dataset, the EGMS will provide valuable resources for a wide range of applications and studies, including those that leverage free and open-source software for geospatial analysis. The validation results presented here will help to ensure the accuracy and reliability of the EGMS product, thereby enabling further research and applications in areas such as geohazards, environmental monitoring, and infrastructure management.

References

Costantini, M., Minati, F., Trillo, F., Ferretti, A., Novali, F., Passera, E., Dehls, J., Larsen, Y., Marinkovic, P., Eineder, M. and Brcic, R., 2021, July. European ground motion service (EGMS). In 2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS (pp. 3293-3296). IEEE.

Urban Atlas, 2018. Copernicus Land Monitoring Service. European Environment Agency: Copenhagen, Denmark.

Synthetic Aperture Radar Interferometry (InSAR) is a space geodesy technique which is systematically used for measuring ground displacements produced by earthquakes, volcanic activity and other geophysical processes. A limiting factor to this technique is the effect of the troposphere, as spatial and temporal variations in temperature, pressure, and humidity introduce significant phase delays in the microwave signal propagation, which contributes with a false deformation component. This component can be discriminated as a) the stratified part, linked with the propagation column thickness and is a function of the Digital Elevation Model and b) the turbulent part, which is due to local weather conditions, like clouds, rainfall, etc. and needs more sophisticated handling. Numerical Weather Prediction (NWP) models are being increasingly used as a tropospheric correction method in InSAR, as they can potentially overcome several of the problems faced with other predictive correction techniques (such as timing, spatial coverage and data availability issues). Here, we investigate the extent to which a high-resolution Weather Research Forecasting (WRF) 1-km re-analysis can produce detailed tropospheric delay maps of the required accuracy. Our study focuses on an area of approximately 150 × 90 km² in the region of the Western Gulf of Corinth (GoC), Greece, where prominent topography makes the removal of both the stratified and turbulent atmospheric phase screens a challenging task. Micro-climatic and topographical characteristics in the Gulf of Corinth are highly variable, meaning that the high-resolution numerical weather modeling will need to capture near-surface atmospheric processes which are related to complex topography, such as sea breezes, orographic flows, turbulent boundary layer interactions etc. This is particularly useful when it comes to estimating the highly variable water vapour signals which contribute to the noise signal.

The model is locally configured and its parameterization includes numerous complex schemes, which are tested in order to demonstrate the optimal configuration at the specific location. WRF output is validated with the use of GNSS tropospheric data retrieved from a dense array of stations covering the selected study area. Model validation with vertical column data (GNSS zenithal delays) instead of ground measurements offers the capability of evaluating the model’s forecasting skill over the entire 3-D field. Having identified the optimal model parameterization, we correct sixteen Sentinel-1A interferograms with differential delay maps at the line-of-sight (LOS) produced by WRF re-analysis. In most cases, corrections lead to a decrease of the phase gradient, with average root-mean square (RMS) and standard deviation (SD) reductions of the wrapped phase of 6.0% and 19.3% respectively. Results suggest a high potential of the model to re-produce both the long-wavelength stratified atmospheric signal and the short-wave turbulent atmospheric component which are evident in the interferograms.

The tropospheric correction of InSAR interferograms and subsequent improvements in the detection of co-seismic, post-seismic and other types of ground deformation, following our methodology, have applicability on a global scale, reflecting the strong impact of our research on the study of geophysical processes with the use of remote sensing techniques. In a framework of the need of rapid response for the determination of a sudden geohazard event from space, the need of an operational (routinely or automated) tropospheric corrections provision based on the proposed methodology is among the aims of the group. As part of multi-temporal interferometry products, our correction method could be exploited either by routine services, such as Copernicus Land Monitoring Service (CLMS) operated by the European Environment Agency (EEA) or on-demand services, such as the Geohazards Exploitation Platform (GEP) operated by ESA.

As magma moves within a volcanic system it alters the distribution of pressure throughout and can cause spatially and temporally complex deformation patterns at the surface. These patterns can be studied to obtain insights into the orientation of magma migration, and the potential volume of the mobilized magma body. The array of variable parameters in magmatic systems, such as temperature, composition and melt lens geometry, are key in controlling the presentation of surface deformation and potential eruptive styles during active periods.

Inferences from volcano geodesy are guided by analysis of the system's rheological and physical properties, which can vary widely throughout a single system following the conception of a Trans-Crustal-Magmatic-System (TCMS). For TCMS, the most classical and simple model of a liquid magma chamber surrounded by an elastic crust has been redeveloped to incorporate potentially numerous melt-rich pockets throughout a widespread mushy, partially molten region of the crust.

Accounting for the presence of a mushy texture implies that a complex mixture of crystals and melt must be considered in the system and therefore viscous and porous behaviour must be accounted for alongside elasticity. This difference in rheological behavior implies an alteration in the appearance and evolution of surface deformation. At present, the influence of porous and viscous parameters have been tested in some models and volcanoes, e.g., Newman et al. (2005), Reverso et al. (2014), Hickey & Gottsmann (2014), Segall (2016). As InSAR resolution continues to increase, the study of more subtle geodetic patterns due to magmatic movement remains simplified. More detailed geodetic measurements may hold more information for reconstruction of subsurface processes. Here, we determine the most influential parameters within a magmatic system, from structural geometry to rheological properties of the crystals and melt and their interdependent relationships, via sensitivity testing.

Using a finite-element method we simulate an intrusion of magma into a mush zone’s structure, by assuming an overpressurized source surrounded by a crystalline mush. Then, we extract a series of potential deformation patterns at the surface due to a variety of subsurface conditions and pressure changes in order to be compared against InSAR images of surface deformation patterns above active volcanic areas. The volcanic systems used for this comparison are selected based upon the level of active or recorded deformation, alongside the likelihood of TCMS presence. The latter must be supported by extensive observational datasets such as geochemical analysis and geophysical mapping of the plumbing system.

The InSAR results for deformation above mush zones will be inverted to assign the most likely deformation sources based upon simulated deformation sequences with known internal parameters. This incorporates a range of pressure changes, structural geometries and rheological parameters, as well as allowing for variable magmatic compositions. The pathways of the inversion model results will contribute towards a training dataset for a deep learning tool being developed to detect, confirm and classify the presence and cause of surface deformation at volcanoes.

References:

Hickey, J. and Gottsmann, J., 2014. Benchmarking and developing numerical Finite Element models of volcanic deformation. Journal of Volcanology and Geothermal Research, 280, pp.126-130.

Newman, A.V., Dixon, T.H. and Gourmelen, N., 2006. A four-dimensional viscoelastic deformation model for Long Valley Caldera, California, between 1995 and 2000. Journal of Volcanology and Geothermal Research, 150(1-3), pp.244-269.

Reverso, T., Vandemeulebrouck, J., Jouanne, F., Pinel, V., Villemin, T., Sturkell, E. and Bascou, P., 2014. A two‐magma chamber model as a source of deformation at Grímsvötn Volcano, Iceland. Journal of Geophysical Research: Solid Earth, 119(6), pp.4666-4683.

Segall, P., 2016. Repressurization following eruption from a magma chamber with a viscoelastic aureole. Journal of Geophysical Research: Solid Earth, 121(12), pp.8501-8522.

The European Ground Motion Service (EGMS) is the first operational service providing ground-motion measurements based on SAR-interferometry (InSAR) at a continental level [1]. It is part of the Copernicus Land Monitoring Service managed by the European Environment Agency (EEA). The EGMS is based on the full resolution InSAR processing of ESA Sentinel-1 radar data acquisitions and covers almost all European landmasses (i.e. all Copernicus Participating states) [2]. The first Baseline release includes ground motion timeseries from 2015 to 2020. Yearly updates of this open dataset will be released every 12 months, in Q3 of each year, except for the first one that was released in February 2023. Funds are ensured to continue the Service beyond 2024.

The EGMS employs persistent scatterers and distributed scatterers in combination with a Global Navigation Satellite System model to calibrate the ground motion products. This public dataset consists of three products levels (Basic, Calibrated and Ortho). The Basic and Calibrated product levels are full resolution (20 x 5 m) Line of sight velocity maps coming from ascending/descending orbits. The Ortho product offers horizontal (East-West) and vertical (Up-Down) velocities, anchored to the reference geodetic model resampled at 100 x 100 m. Since InSAR data production involves the application of thresholds and filters to remove unwanted phase artefacts, the results may contain systematic effects, outliers or simply measurement noise.

Independent validation is being carried out by a consortium composed of six partners to assess the quality and usability of the EGMS products. The validation is divided into seven separate validation activities: Point density check; Comparison with other ground motion services; Comparison with inventories of phenomena; Consistency check with ancillary geo-information; Comparison with GNSS; Comparison with in-situ monitoring; Evaluation XYZ and displacements with Corner Reflectors.

The subject of this abstract is to describe the comparison with other ground motion services. A total of nine validation sites have been selected for this validation activity using data from the national ground motion services of Norway, Sweden, Denmark, the Netherlands and Germany, the regional services for the Italian regions of Tuscany, Valle d'Aosta and Veneto, and data for Mount Etna, Sicily, specifically processed for the validation by IREA. Due to its volcanic activity, Mount Etna provides a particularly interesting validation site with areas showing strong subsidence and others experiencing strong heave and with displacement time-series that have a strong non-linear component. Therefore, the technical approach for the comparison with other GMS data is presented using the Mount Etna validation site as example.

The comparison of two different InSAR datasets is based on the approach published by [3]. Both datasets are first resampled spatially (to a common regular grid) and temporally (to common acquisition dates) to make a direct comparison possible, including recalculating velocities to the temporally resampled data. A key aspect of the validation is the identification of Active Displacement Areas (ADAs) which is carried out using an automated procedure. All identified ADAs are compared regarding their (a) spatial overlap; (b) velocity and (c) time-series development. A comparison of the overall point density is also carried out.

For the most important validation measures, normalized key performance indices (KPI) are calculated, which are then reduced to a single KPI for each validation site using a weighted average. The weights are chosen based on the relevance of the respective validation measure for the respective validation site. KPIs as well as an expert's visual inspection of the comparison will finally provide the basis for the validation.

References

[1] Crosetto, M.; Solari, L.; Mróz, M.; Balasis-Levinsen, J.; Casagli, N.; Frei, M.; Oyen, A.; Moldestad, D.A.; Bateson, L.; Guerrieri, L.; Comerci, V.; Andersen, H.S. The Evolution of Wide-Area DInSAR: From Regional and National Services to the European Ground Motion Service. Remote Sens. 2020, 12, 2043. https://doi.org/10.3390/rs12122043

[2] Costantini, Mario & Minati, F. & Trillo, Fritz & Ferretti, Alessandro & Novali, Fabrizio & Passera, Emanuele & Dehls, John & Larsen, Yngvar & Marinkovic, Petar & Eineder, Michael & Brcic, Ramon & Siegmund, Robert & Kotzerke, Paul & Probeck, Markus & Kenyeres, Ambrus & Proietti, Sergio & Solari, Lorenzo & Andersen, Henrik. (2021). European Ground Motion Service (EGMS). 10.1109/IGARSS47720.2021.9553562.

[3] Sadeghi, Z., Wright, T.J., Hooper, A.J., Jordan, C., Novellino, A., Bateson, L., Biggs, J. (2021). Benchmarking and Inter-Comparison of Sentinel -1 InSAR velocities and time series. Remote Sensing of Environment. 256. 112306. 10.1016/j.rse.2021.112306.

Understanding geophysical phenomena, such as volcanic eruptions and their associated processes, plays an essential role in disaster risk management (Harris, 2015). In particular, effusion rates, extent, and volume of lava flows are key eruption parameters necessary for evaluating hazards posed by effusive eruptions (Pedersen et al., 2022a). To monitor the development and progression of volcanic processes, it is necessary to utilise high-temporal resolution data that regularly document and track such events. Both optical and synthetic aperture radar (SAR) Earth observation (EO) data can be used to map and monitor lava flows. Although the use of optical imagery is limited by clouds or volcanic plums after volcanic eruptions (Boccardo et al., 2015), SAR systems can provide data on a regular basis owing to the weather independence and day and night capabilities, making them extremely useful for monitoring lava flows (Pinel et al., 2014).

In the Fagradalsfjall volcanic system in southwestern Iceland, an eruption occurred from March to September 2021, followed by another event in 2022 after a quiescence period of 6000 years. The eruption presents a unique opportunity to observe the flow dynamics and characteristics of lava flows, such as their extent, volume, runout, and thickness. Based on aerial photogrammetric surveys and derived orthophotos, Pléiades stereo images, digital elevation models (DEMs), and thickness and thickness change maps, Pedersen et al., (2022a) manually mapped the lava flows and calculated the lava volume and effusion.

In this study, we explore the applicability of Sentinel-1 (C-band) SAR backscatter information for mapping the lava flows of the recent Fagradalsfjall eruptions. Lava flow mapping using freely available EO data is less time-consuming and cost-effective than field measurements. Moreover, Sentinel-1 data can be used to generate multi-temporal DEMs using interferometric SAR (InSAR) techniques, which can be applied for regular monitoring of land surface elevation changes (Dabiri et al., 2020) and for the characterisation of lava flows, if the quality of the generated DEMs is sufficient. The main objectives of this study are (1) to semi-automatically map the lava flow extent for the 2021 and 2022 Fagradalsfjall eruptions using object-based image analysis (OBIA) and Sentinel-1 data backscatter information, and (2) to assess the suitability and applicability of Sentinel-1 derived DEMs for lava flow volume estimation.

We used pre-, syn-, and post-event Sentinel-1 A & B dual-polarisation Interferometric Wide Swath (IWS) Level-1 high-resolution Ground Range Detected (GRD) products to map the extent and evolution of the Fagradalsfjall lava flows in 2021 and 2022, and Single Look Complex (SLC) products for interferometry and DEM generation. Several layers were used for the segmentation and delineation of the lava flow outlines, including terrain-corrected gamma backscatter information, different polarisation ratio layers, and textural layers based on the grey-level co-occurrence matrix (GLCM), such as contrast, dissimilarity, and entropy. The multiresolution segmentation algorithm was used to generate homogenous objects, which served as the basis for classifying lava flows using backscatter, textural, and spatial information. The accuracy of the mapping results was estimated by considering the overlapping area between the OBIA results and lava outlines created by Pedersen et al., (2022b). The lava flows were generally well depicted by OBIA; however, the creation of suitable image objects is challenging because the backscatter signals can vary between different acquisitions, for example, due to changes in soil moisture. Moreover, the side-looking geometry of SAR in steep topography causes foreshortening and shadow effects. Hence, some parts of the lava flows were not fully captured using the descending flight direction. Utilisation of ascending and descending orbits may overcome this constraint to some extent. Future studies should further explore the potential and transferability of object-based change detection analysis for lava flow mapping using time-series Sentinel-1 data.

The lava flow delineations were then used as inputs for the volume estimation. Therefore, we created pre- and post-event DEMs for the eruptions for both ascending and descending flight paths using Sentinel-1 image pairs and InSAR algorithms, and compared the resulting DEMs. We used an open-source Python package for DEM generation and volume estimation (Abad et al., 2022). Additionally, we performed post-processing steps, such as co-registration, to align the generated DEMs in the vertical direction using the ArcticDEM (2 m resolution) as a reference, prior to the volume estimation based on the DEMs of Difference (DoDs). The quality assessment of the generated DEMs consisted of the computation of several statistical error measures, such as the normalised median absolute deviation (NMAD), with respect to the reference DEM, and based on topographical derivatives, such as slope and aspect. The estimated volumes were then compared to those from the literature and published repositories (Pedersen et al., 2022b). Although the quality of the generated DEMs is generally promising, the results differ depending on the image pair used for DEM generation. The DoDs reflect the spatial distribution of lava flows to some extent; however, lava flow distinction from the surroundings is ambiguous in areas close to steep slopes. Consequently, the lava flow volume estimations vary, with some estimations close to the reference, and others that significantly over- or underestimate the volume. Thus, further research is needed to increase the DEM accuracy and identify the sources of errors. This can include a detailed assessment of the influence of the image parameters (e.g. perpendicular and temporal baselines), improving post-processing methods, such as combining different co-registration techniques to reduce the bias between the generated DEMs, and the fusion of the DEMs generated from descending and ascending flight directions. Multi-temporal DEMs are rarely available; thus, DEMs derived from freely available Sentinel-1 data can be of great value for studying geomorphological landscape volume changes caused by lava flows. However, a requirement is that a sufficient quality of the generated DEMs can be achieved.

Abad, L., Hölbling, D., Dabiri, Z., & Robson, B. A. (2022). AN OPEN-SOURCE-BASED WORKFLOW FOR DEM GENERATION FROM SENTINEL-1 FOR LANDSLIDE VOLUME ESTIMATION. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XLVIII-4/W1-2022, 5–11. https://doi.org/10.5194/isprs-archives-XLVIII-4-W1-2022-5-2022

Boccardo, P., Gentile, V., Tonolo, F. G., Grandoni, D., & Vassileva, M. (2015). Multitemporal SAR coherence analysis: Lava flow monitoring case study. 2015 IEEE International Geoscience and Remote Sensing Symposium (IGARSS), 2699–2702. https://doi.org/10.1109/IGARSS.2015.7326370

Dabiri, Z., Hölbling, D., Abad, L., Helgason, J. K., Sæmundsson, Þ., & Tiede, D. (2020). Assessment of Landslide-Induced Geomorphological Changes in Hítardalur Valley, Iceland, Using Sentinel-1 and Sentinel-2 Data. Applied Sciences, 10(17), 5848. https://doi.org/10.3390/app10175848

Harris, A. J. L. (2015). Chapter 2 - Basaltic Lava Flow Hazard. In J. F. Shroder & P. Papale (Eds.), Volcanic Hazards, Risks and Disasters (pp. 17–46). Elsevier. https://doi.org/10.1016/B978-0-12-396453-3.00002-2

Pedersen, G. B. M., Belart, J. M. C., Óskarsson, B. V., Gudmundsson, M. T., Gies, N., Högnadóttir, T., Hjartardóttir, Á. R., Pinel, V., Berthier, E., Dürig, T., Reynolds, H. I., Hamilton, C. W., Valsson, G., Einarsson, P., Ben‐Yehosua, D., Gunnarsson, A., & Oddsson, B. (2022a). Volume, Effusion Rate, and Lava Transport During the 2021 Fagradalsfjall Eruption: Results From Near Real‐Time Photogrammetric Monitoring. Geophysical Research Letters, 49(13), 1–11. https://doi.org/10.1029/2021GL097125

Pedersen, G. B. M., Belart, J. M. C., Óskarsson, B. v., Gudmundsson, M. T., Gies, N., Högnadóttir, T., Hjartadótti, Á. R., Pinel, V., Berthier, E., Dürig, T., Reynolds, H. I., Hamilton, C. W., Valsson, G., Einarsson, P., Ben-Yehosua, D., Gunnarsson, A., & Oddsson, B. (2022b). Digital Elevation Models, orthoimages and lava outlines of the 2021 Fagradalsfjall eruption: Results from near real-time photogrammetric monitoring (v1.1) [Data set]. https://doi.org/10.5281/ZENODO.6598466

Pinel, V., Poland, M. P., & Hooper, A. (2014). Volcanology: Lessons learned from Synthetic Aperture Radar imagery. Journal of Volcanology and Geothermal Research, 289, 81–113. https://doi.org/10.1016/j.jvolgeores.2014.10.010

Deep learning (DL) for volcanic deformation detection is commonly done using the classification model to flag volcanic deformation in Interferometric Synthetic Aperture Radar (InSAR) data. This approach generally focused on faster, larger deformations because of higher data availability and associated challenges with detecting subtle deformations. To detect subtle deformations, InSAR data needs atmospheric and solid earth tide corrections, and persistent and distributed scatterer (PS/DS), which is work-intensive. On the other hand, DL is known to be data-intensive, needing a training set significant in the amount and quality of samples.

To overcome the limited training data, we propose using generative adversarial networks (GANs) to generate more extensive realistic synthetic training data. GANs consist of two components, a generator, and a discriminator. The generator tries to create realistic-looking images, while the discriminator tries to distinguish the generated image from a real one. Trained together, the model learns to generate realistic images. In addition, GANs can generate infinite synthetic data containing regional deformation patterns and can be replicated for other regions.

We employ PS/DS techniques to generate high deformation accuracy InSAR data covering Central Volcanic Zone in South America from 2014-2020. This region is sparsely populated and dense with volcanoes. The data are corrected for the tropospheric and ionospheric delay and solid earth tide, to achieve 1 mm/year accuracy. From the data, we cut out the 102.4 km by 102.4 km frames over existing volcanoes, which we use to test our DL model for detecting volcanic deformations.

A classification model is used to show which data set teaches the model better to distinguish volcanic deformations. The model is trained to output 1 if volcanic deformation is present in the image or 0 otherwise. We create two training sets using synthetic data. The positive class uses synthetic volumetric volcano deformation simulations combined with background noise, while the negative only has background noise. Two different sets are based on differently generated background noise sets. First, traditionally created synthetic noise, consisting of stratified and turbulent noise, and second, data generated using the GAN. We use StarGAN v2, a multi-domain and bidirectional state-of-the-art image-to-image translation model. We use it to learn the transformation from synthetic background data to real background data and apply it to the synthetic training set to make the data more realistic.

To train GANs, we use the data surrounding the test region. This same data is used as a fine-tuning set for the classification model trained on completely synthetic data. We compare the four models based on InceptionResNet v2 architecture: a model trained on synthetic data, fine-tuned model, a model trained on GAN-generated data, and a model trained on synthetic data and fine-tuned on GAN-generated data. The model metrics and explainability are analyzed using grad-CAM and t-SNE feature visualization.

The Gamma Portable Radar Interferometer (GPRI) is a versatile ground-based real aperture radar instrument (FMCW) with a multitude of operation modes. In the standard acquisition mode a rotation of the antennas is used for image generation (17.2 GHz central frequency, 200 MHz bandwidth, 17.4mm wavelength). Rotation of the antennas requires a few tens of seconds and this defines the lower limit of the revisit time interval of the acquired time series. Interferometric analysis is therefore limited to surfaces that remain coherent for at least the revisit time interval. In this contribution, we present a processing method that permits observing fast movements of surfaces, e.g. water, that typically decorrelate within milliseconds. The real-aperture nature of the GPRI makes imaging of surfaces with such a short coherence time possible because each radar pulse images only a radial line in the final image. No aperture synthesis requiring multiple coherent radar pulses is done. The high gain of the antennas used by the GPRI results in an excellent noise equivalent sigma zero of around -35 dB at 2 km, so that even with a grazing incidence angle mapping of capillary waves on water surfaces could be achieved up to distances of several kilometers.

The new processing method applies to data acquired with a real aperture radar operated in a rotational acquisition mode. The angular rotation angle between successive pulses has to be smaller than the angular width of the antenna beam pattern, so that the area covered by successive beams includes a common section. While sweeping over the common section, this common section is mapped at slightly different time. The differential phase between the successive, focused echos relates directly to the line-of-sight (LOS) displacement for this time interval so that LOS velocity values can be calculated for the common beam section. Due to the overlapping beams during a rotational scan a 2D map of the line-of-sight velocity can be computed over the observation area.

To demonstrate the method, the GPRI was deployed 65 meters above Lake Thun, Switzerland, on 24 June 2022. With an area of 48 km² Lake Thun is a relatively small water body that showed small waves during the experiment (states 0 and 1 , (glassy/small ripples), gradually increasing to state 2 (small wavelets; crests did not break), according to the WMO Sea State Code Table 3700.

Data were acquired with an antenna rotation rate of 2.5 deg/s, a beam width of 0.4 deg, and a chirp length of 2 ms. Interferograms were formed between successive echoes with a time delays between 4 and 16 ms. Under the present conditions of experiment, it was possible to observe the line-of-sight component of the wave velocity within a 180° sector up to a distance of more than 2 km. Surface velocities up to 0.7 m/s were observed and interpreted as the phase velocity of capillary waves. For these instrument parameters, the maximum time-delay for interferogram formation was limited to approximately 80 ms which is when the antenna pattern has 50% overlap. This maximum time delay is significantly longer than the observed decorrelation time of the water surface of 10-20 ms. Surprisingly, for a few pixels on the water surface, we observed decorrelation times significantly longer than 20 ms. Photographic evidence suggests that these targets are floating debris or birds indicating that pulse-to-pulse interferometry can also be used to detect coherent targets with very low backscatter on the surface of the water. Contrarily, the loss of coherence of consecutive echoes can be used to mask surfaces (and shadow) where the physical echo is below the noise-equivalent-sigma zero so that the measured data contains only uncorrelated noise.

Flexible chirp length, pulse repetition frequency and rotation rates of the GPRI provide a wide range of observable velocities. High PRFs permit studying very fast phenomena as long as the observed objects or surface remain coherent and within the antenna beam pattern for at least two pulses. With an adjustable chirp length PRFs are possible that range from 100 Hz [0.75m resolution, 20 km range] to approximately 100 kHz [3m resolution, range limited to 100 m]. With this range of PRFs maximum line-of-sight velocities of v_max = λ/4*PRF = 0.43 ... 430 m/s can be measured unambiguously.

The range of observable velocity in the pulse-to-pulse interferometry mode extends the existing upper limit for the unambiguously measurable velocity in acquisition-to-acquisition interferograms almost seamlessly. For acquisition-to-acquisition interferograms with temporal baselines of Δt = 30 s the upper velocity limit is λ/4 / Δt = 0.14 mm/s. For pulse-to-pulse interferograms, the minimum measurable velocities is given by the precision of phase estimation and by the antenna rotation speed. The slower the antennas rotate, the more independent echos are measured and the better the estimation of the displacement phase. With a nominal rotation rate of 10°/s and a beam width of 0.4° several hundred independent echoes are measured for each common beam section and can be used for velocity analysis. Assuming a phase noise of 5° results in a lower limit for the measurable velocities of 6 mm/s. Reducing the rotation rate to 0.25°/s connects directly to the velocity limit of the the acquisition-to-acquisition method.

With this new range of observable velocities, the new pulse-to-pulse processing method extends the capability of the GPRI for velocity measurements by six orders of magnitude.

The formation of short-time interferograms over a large sector of interest is a quite unique capability of the GPRI instrument. Operating two radar systems at two different locations can potentially determine two components of the surface velocity vector field. Larger waves causing stronger backscatter are expected over the ocean that permit operation of the GPRI out to significantly greater distances compared to the calm conditions of Lake Thun.

Peat areas in the Netherlands exhibit extremely dynamic vertical motion, including both reversible and irreversible components. Yet the exact behaviour is spatially variable, and difficult to estimate. This results in a poorly known estimation of greenhouse gas emissions and impact to existing infrastructure, and consequently limited ability to design and deploy mitigating or adaptive measures. To monitor the full peat areas, InSAR has the necessary combination of resolution, temporal sampling, and coverage. Due to the vegetation, however, it suffers from temporal decorrelation, while the noise combined with rapid vertical motion makes phase ambiguity estimation extremely difficult.

We have deployed four "free-floating" radar transponders (FFTs) into peat parcels around the Netherlands. A radar transponder is an electronic corner reflector, amplifying and returning the radar wave emitted by the satellite. Since most motion originates from the uppermost layers of peat, the FFT needs to be directly connected with the surface, i.e. with a very shallow foundation.

Using the FFT as the reference point for arcs to distributed scatterers in the surrounding parcels would result in most motion being removed from the estimated time series, since the parcels are expected to respond in a similar way to environmental input such as precipitation and temperature. This would result in a more robust and reliable phase ambiguity estimation procedure. The motion of the FFT itself can, due to its high phase precision, easily be estimated with respect to a reference point of which the motion is known, such as an Integrated Geodetic Reference Station. Nevertheless, even with this high phase precision we need to employ context-guided phase unwrapping as proposed by P. Conroy et al. (2022) due to the extremely dynamic vertical motion.

We designed a frame to support the radar transponder a few centimeters below the surface, where weight dissipation was the main driver for the design to prevent the radar transponder subsiding autonomously with respect to the surface. The soft soils are also the reason we opted for light-weight transponders, as passive corner reflectors with a similar radar cross-section require a weighty and large support frame.

We installed the four FFTs in areas with ground truth provided by an extensometer installed a few meters away, allowing validation of the InSAR displacement estimates. Three FFTs were installed between December 2021 and March 2022. A fourth one was installed in February 2023, but is not included yet in this study. Each transponder is programmed to respond to two ascending and two descending SAR acquisitions. Regular leveling campaigns were held at all four sites to monitor possible autonomous subsidence with respect to the surface. We did not find evidence of autonomous motion in any of the FFTs.

Using only the acquisitions in which the FFTs were visible, we analyzed the phase response and displacement estimates with respect to the extensometers. For FFTs Aldeboarn and Assendelft, we chose the reference point for InSAR to be on a pile-supported building belonging to a farm about 280 m and 220 m away, respectively. For FFT Zegveld the reference point is a founded Integrated Geodetic Reference Station, including corner reflectors and GNSS, about 170 m away.

For two FFTs (Aldeboarn and Assendelft) we observe good agreement with the extensometer time series, where the RMSE of the relative vertical position projected onto the vertical with respect to the extensometer varies between 3 mm and 6 mm per track. For FFT Zegveld the RMSE varies between 7 mm and 10 mm per track. All FFTs behave as intended: as a coherent point scatterer moving with the surface. For the first time we can see the actual highly dynamic movement of the peat soils from InSAR without the need for multilooking, hereby providing a coherent reference point that can be used to expand the InSAR analysis into other parcels.

While yielding reliable results, several FFTs experienced missed acquisitions during the year. For FFTs Aldeboarn and Assendelft the rate of success is 82% (110 Success/24 Failed) and 87% (103 Success/16 Failed), respectively. For FFT Zegveld the rate of success was 49% (45 Success/47 Failed) between December 2021 and September 2022. We replaced the radar transponder in Zegveld with an updated model, and have not missed acquisitions since (58 Success/0 Failed).

These results show that the concept of free-floating transponders is a very useful addition to the InSAR toolkit. Apart from serving as a 'moving reference point', we apply the concept for rapid site characterization, which helps in the tuning and optimization of location-dependent InSAR distributed scatterer processing, and for deployment at locations where reliable opportunistic point scatterers cannot be found.

[1] P. Conroy, S.A.N. van Diepen, S. van Asselen, G. Erkens, F.J. van Leijen, and R.F. Hanssen, Probabilistic Estimation of InSAR Displacement Phase Guided by Contextual Information and Artificial Intelligence. IEEE Transactions on Geoscience and Remote Sensing, vol. 60, Sept. 2022.

Askja Volcano is located at the divergent plate boundary in Iceland, in the Northern Volcanic Zone. It was characterised by subsidence for four decades until a period of uplift began in 2021 and still going on. The cause of the subsidence is not yet well understood, with proposed mechanisms including magma cooling, contraction, and magma drainage from shallow to deeper magma chambers. In this work, we will present surface deformation time series from 2015 to 2020 and examine the role of plate spreading and the rheology of the underlying magmatic system in the subsidence signal, through modelling.

Askja Volcano compromises three calderas in an area of 45 km² and is spatially related to a fissure swarm produced by the divergence between the North American plate and the Eurasian plate. A rifting episode occurred in this volcano from 1874 to 1876, followed by two eruptive periods during 1921-1929 and 1961.

We used Synthetic Aperture Radar Interferometry (InSAR) data acquired from Sentinel-1 between 2015 and 2020. We have analysed 4 frames (2 ascending and 2 descending) to generate a network including longer timespan (summer to summer of 1 year long) connections and avoiding low coherence interferograms influenced by snow during winter, using LiCSBAS (Morishita et al., 2020). Atmospheric noise was reduced using GACOS (Yu, Li, Penna, & Crippa, 2018). We estimated the line-of-sight velocity for each frame and tied the results to the ITRF reference frame (Altamimi, Métivier, & Collilieux, 2012) using Global Navigation Satellite System (GNSS) data from 35 stations around the volcano. Then, we subtract glacial isostatic effects produced by the ongoing retreat of the nearby Vatnajokull icecap, using a scaled version of the model of Auriac et al., (2014). We consider the remaining signal as deformation produced by processes in the magmatic system below the volcano, and the effects of plate movements. A 3D finite element model using COMSOL Multiphysics is used to explain the observed surface deformation.

References:

Altamimi, Z., Métivier, L., & Collilieux, X. (2012). ITRF2008 plate motion model. Journal of Geophysical Research: Solid Earth, 117(B7). https://doi.org/https://doi.org/10.1029/2011JB008930

Auriac, A., Sigmundsson, F., Hooper, A., Spaans, K. H., Björnsson, H., Pálsson, F., … Feigl, K. L. (2014). InSAR observations and models of crustal deformation due to a glacial surge in Iceland. Geophysical Journal International, 198(3), 1329–1341. https://doi.org/10.1093/gji/ggu205

Morishita, Y., Lazecky, M., Wright, T. J., Weiss, J. R., Elliott, J. R., & Hooper, A. (2020). LiCSBAS: an open-source InSAR time series analysis package integrated with the LiCSAR automated Sentinel-1 InSAR processor. Remote Sensing, 12(3), 424.

Yu, C., Li, Z., Penna, N. T., & Crippa, P. (2018). Generic atmospheric correction model for interferometric synthetic aperture radar observations. Journal of Geophysical Research: Solid Earth, 123(10), 9202–9222.

Freeze-thaw cycles in Arctic permafrost regions can lead to considerable ground displacements. Surface subsidence caused by thawing in summer can be substantial especially for areas of ice-rich permafrost and may be countered by frost heave in winter. These displacements can reach up to decimetre-scale and are caused by phase changes from ground ice to liquid water and vice versa. InSAR has proven to be a valuable tool to monitor displacements in these often remote locations. In this study, we detect ground displacements using Sentinel-1 data, which provides 12-days repeat time intervals for most Arctic regions. Due to generally low coherence values during longer time intervals, however, the number of usable interferograms for displacement calculations in the study area is restricted. In order to achieve correct InSAR displacement timeseries with this limited number of interferograms, it is essential to correct for atmospheric effects that can significantly distort results, especially during the thawing periods. We therefore processed interferograms in series and compared these unfiltered timeseries with results of applied spatial filtering (linear least-squares method, filter radius 6 km) as well as results corrected with the Generic Atmospheric Correction Service (GACOS), which utilises the ECMWF weather model data as well as DEM data to provide tropospheric delay maps. Comparisons of methods have been performed for selected regions throughout the Arctic, in order to determine a best practice for an easily applied correction method suitable for a circumpolar implementation that would allow an extensive study of permafrost degradation and disturbance zones. Results show in most cases improvements for GACOS corrected results. For the spatially filtered results displacement timeseries get smoothed out, but also the magnitude of overall displacements is often greatly reduced. Furthermore, large scale displacements are filtered out. Results have been compared to mechanically measured in situ data of yearly subsidence and to borehole temperature measurements. Comparisons to in situ data of yearly subsidence at one of the study regions revealed that, while InSAR results are mostly lower than in situ data, GACOS corrected results delivered the closest match and spatially filtered results performed worst. Highest agreement with thaw progression in boreholes was also found for GACOS corrected results. Moreover, an improvement in error statistics could be derived for the filtering methods in most regions.

High strain areas are regions of the Earth's crust, associated with tectonic plate boundaries, where the rates of ground deformation are particularly high. These areas are characterized by high seismic activity, making them of significant concern. The ability to estimate ground deformation in these regions is critical for understanding the underlying geological processes and for assessing the potential risk of future seismic events.

The motivation for this study is to help providing a better understanding of the behavior of the earth's crust in high strain areas. Interferometric Synthetic Aperture Radar (InSAR) has shown great promise in delivering millimetre-scale ground displacement information over long distances across plate boundaries. In this project, we aim to globally measure ground deformation using the InSAR Persistent and Distributed Scatterer (PS/DS) technique, focusing on the regions where the second invariant of the strain is higher than 3 nanostrain per year.

Due to the large amount of data that has to be processed, we use the high-performance data analytics platform made available by the framework of the Terra_Byte project, a cooperation between the German Aerospace Center (DLR) and the Leibniz Computer Centre (LRZ). This enables us to process large volumes of data efficiently. We use the IWAP processor to apply the PS/DS technique to time-series of seven years of SAR images acquired by the Sentinel-1 mission. To improve the accuracy of our analysis and reduce the influence of ionospheric variations we use CODE total electron contents maps. The impact of solid earth tides (SETs) is limited by using the IERS 2010 convention, which provides a standard reference for the modelling of SETs. Most important, we use ECMWF reanalysis data to correct for tropospheric delays, which are the biggest error source and limiting factor for the interferometric performance at large distances. The influence of soil moisture and vegetation growth on distributed scatterers is limited by the full covariance matrix approach used in the interferograms generation. Finally, we calibrate and compare our results with GNSS measurements to show a detailed picture of ground deformation.

The results of this project will be publicly available on a global scale, including: velocity maps, timeseries, line-of-sight projection vectors. The product palette will allow custom calibration or 2D decomposition by the user. Possible applications are: the large coverage and homogeneous processing characteristics of the data could serve as a baseline reference or comparison for other studies. Geoscientists will be able to use the deformation measurements to gain a better understanding of geological processes, with the dense PS/DS measurements filling in the gaps between existing GNSS survey data, possibly finding new strain areas, contributing to the advancement of scientific knowledge in this field.

In the presentation we will show first products of selected areas generated by our processing chain, such as Turkey and other well known regions.

An automatic soil moisture retrieval algorithm from Synthetic Aperture Radar (SAR) over agricultural bare and vegetated fields is investigated. Soil moisture retrieval is based on (i) multi-frequency and polarimetric SAR data in L- (SAOCOM), X- (COSMO-SkyMed both first and second generation) and C-band (Sentinel-1) integration [1][2]; (ii) bare and vegetated soil scattering models inversion [3][4][5]; (iii) Bayesian minimization and machine learning techniques; (iv) biomass estimation from hyper-spectral and multi-spectral electro-optical data [6][7]. The work is carried out by a consortium composed by e-GEOS S.p.A., “La Sapienza” University, Tor Vergata University, Tuscia University and IBF Servizi S.p.A. in the framework of the CLEXIDRA project funded by the Italian Space Agency (ASI). The activity is supported by in-situ data collected over crop fields located in Argentina (Monte Buey) and in Northern Italy (Jolanda di Savoia).

Preliminary results show that co-polar L-band backscattering is sensitive to soil water content. SAR L-band dataset collected in the Argentinian test site - corrected for vegetation effects by using a semi-empirical vegetation contribution model (WCM) - well agree with data simulated by using a semi-empirical electromagnetic model (SEM) of bare soil for low NDVI values. For high NDVI values, both HH and VV co-polarized SAR backscattering coefficients exceed values estimated by SEM thus indicating a significant contribution due to vegetation. When the vegetation contribution is subtracted by WCM, the corrected backscattering coefficients get closer to the SEM estimation. This approach can be used to tune the semi-empirical WCM in order to have a manageable model function, as example exploiting information coming from other SAR bands. In addition, the performances offered by other scattering models for bare soil surfaces will be evaluated.

In Northern Italy site, land parcels have been selected basing on their homogeneity and regular size for comparison with satellite data. The parcels have been split into homogeneous zones - Management Unit Zones (MUZ) - based on a soil geophysical survey; then Elementary Sampling Units (ESU) have been selected to collect both soil roughness and soil moisture data along with some estimates of the water content of plants. Ancillary data and in-situ measurements acquired in coincidence with satellite images include boundaries of agricultural fields, crop type and sowing dates which are fundamental for calibration and validation.

Ongoing activities include two main tasks: first, the exploitation of the COSMO–SkyMed X-band time series of radar imagery collected over the Northern Italy test site aiming at improving the estimation of the contribution of the vegetation to backscattering coefficient in L-band; second, to set up a SAR model inversion based on advanced artificial intelligence techniques. The final ambitious objective of the project is the generation of soil moisture maps for pre-operational use as a tool to support irrigation management activities.

References

[1] Brogioni M., S. Pettinato, G. Macelloni, S. Paloscia, P. Pampaloni, N. Pierdicca & F. Ticconi, "Sensitivity of bistatic scattering to soil moisture and surface roughness of bare soils", International Journal of Remote Sensing, 31:15, 4227-4255, 2010.

[2] Y. Oh, “Quantitative Retrieval of Soil Moisture Content and Surface Roughness From Multipolarized Radar Observations of Bare Soil Surfaces”, IEEE Trans. Geosci. Remote Sensing, vol. 42, 596-601, 2004.

[3] Oh Y., K. Sarabandi, F. T. Ulaby, “Semi-empirical model of the ensemble-averaged differential Mueller matrix for microwave backscattering from bare soil surfaces”, IEEE Trans. Geosci. Remote Sens., vol. 40, no. 6, pp. 1348-1355, June 2002.

[4] E. P. W. Attema and F. T. Ulaby, “Vegetation modeled as a water cloud,” Radio Sci., vol. 13, pp. 357-364, 1978.

[5] M. Bracaglia, P. Ferrazzoli, L. Guerriero, “A fully polarimetric multiple scattering model for crops”, Remote Sensing Environ., vol. 54, pp. 170-179, 1995.

[6] Wocher, M., Berger, K., Verrelst, J., Hank, T., 2022. Retrieval of carbon content and biomass from hyperspectral imagery over cultivated areas. ISPRS Journal of Photogrammetry and Remote Sensing 193, pp. 104-114.

[7] Mzid, N., Casa, R., Pascucci, S., Tolomio, M., Pignatti, S., 2022. Assessment of the Potential of PRISMA Hyperspectral Data to Estimate Soil Moisture. International Geoscience and Remote Sensing Symposium (IGARSS) 2022-July, pp. 5606-5609.

This contribution describes the procedure followed for validating EGMS products with GNSS data. This work is performed within the framework of the Services supporting the European Environment Agency’s (EEA) implementation of the Copernicus European Ground Motion Service – product validation.

The main objective of this activity is the comparison of deformation mean velocities and time series from the EGMS products (2a, 2b and 3) against GNSS data. For this we will apply test statistics, to judge whether the differences are significant, see e.g. [1].

Because GNSS time series are sampled at different times than InSAR and their stations are usually not collocated with InSAR observations, the data needs first to be pre-processed. The pre-processing steps are as follows:

1. Temporal interpolation: Interpolate GNSS time series to match InSAR acquisition dates using a 12-day window.
2. Time reference: Use the same reference date for both GNSS and InSAR time series.
3. Projection of GNSS time series to radar line-of-sight (LOS): Transform GNSS displacement to radar LOS for level 2a and 2b data.
4. GNSS spatial referencing: Select one GNSS station as reference station per thematic area for level 2a data and calculate velocity differences between reference frames for level 2b and 3 products.
5. InSAR MP selection: Select InSAR MPs based on distance and height w.r.t. ground.
6. Spatial interpolation: Interpolate selected InSAR MP time series spatially to GNSS location and estimate interpolation errors.
7. Double differences: Only needed for L2a products.
8. GNSS-InSAR comparison: Compare data sets through time series and deformation model using BLUE.

The workflow is generally the same for all data products (L2a, L2b, L3), but there are some differences. Double differences in space and time are calculated for comparing L2a products to GNSS, while this is unnecessary for L2b and L3 products, which are spatially relative to ETRF 2000. Additionally, when compared to L3 products, GNSS time series are not projected to LOS since L3 products already provide vertical and horizontal components.

Furthermore, we select only those GNSS stations that are considered by the provider to be reliable.

We apply the procedure to different test sites around Europe. This contribution presents the outcomes of the validation process applied to the island of Gran Canaria in Spain and in Jutland, west Denmark.

Gran Canaria is a volcanic island located in the Canary Islands, Spain. The volcano is Gran Canaria is dormant. The last eruption occurred around 2000 years ago.

Jutland is a large peninsula that contains the mainland regions of Denmark. While the country as a whole is experiencing uplift due to post-glacial processes, some areas along the coast of Jutland are undergoing subsidence caused by local phenomena.

References:

[1] Teunissen, P. J. G. (2000b). Testing theory; an introduction (1 ed.). Delft: Delft University Press.

As the accessibility of polar regions increases due to global warming, the development of plant technology in permafrost regions rich in oil and gas is required. To develop resource plant technology suitable for the permafrost regions, it is necessary to select optimal locations for plant construction by analyzing various geospatial information. In permafrost regions, surface displacements occur due to freezing and thawing of the active layer, which can cause instability of the structure. However, there are few cases in which surface displacement is considered in the selection of optimal locations for resource plant construction in the permafrost regions. In this study, the importance of surface displacements in selecting a location of a resource plant in the permafrost regions was evaluated in Athabasca, Alberta in Canada, one of the largest oil sands deposits in the world. To this end, various geospatial information and Analytic Hierarchy Process (AHP), which has been widely used to solve the problems of optimal location selection, were integrated. Air temperature, surface temperature, and subsurface temperature derived from ERA5 reanalysis data provided by the European Center for Medium-Range Weather Forecasts (ECMWF), land cover, elevation, slope, distance from transportation infrastructure (roads, railways, pipelines, and airports), and the surface displacement were used as the geospatial information for the optimal location selection. All geospatial data, except transportation infrastructure, are pre-2011. The surface displacement was derived from the Small BAseline Subset Interferometric Synthetic Aperture Radar (SBAS-InSAR) of 17 ALOS PALSAR images acquired from February 2007 to March 2011. The attributes of each geospatial information for the study area were analyzed and scored, and the goodness of the locations was calculated by applying it to the AHP. The location of oil sands plants constructed after 2011 was used to evaluate whether the optimal sites determined by the AHP are reliable. We could confirm that the oil sands plants built after 2011 were located in the area with high suitability class. The results of the sensitivity analysis on the geospatial information applied to the AHP showed that the surface displacement should be considered important in the optimal location selection of resource plants in the permafrost regions.

Multi-temporal Synthetic Aperture Radar Interferometry (MT-InSAR) is the only geodetic technique allowing to measure ground deformation down to mm/yr over continuous areas. Vegetation cover in Equatorial regions favors the use of L-band SAR data to improve interferometric coherence. However, the electron content of ionosphere, affecting the propagation of the SAR signal, show particularly strong spatio-temporal variations near the Equator, while the dispersive nature of the ionosphere makes its effect stronger on low-frequencies, such as L-band signals. To tackle this problem, range split-spectrum method can be implemented to compensate the ionospheric phase contribution.

Here, we propose a procedure of ionospheric correction for time-series of ALOS-PALSAR data, based on the split-spectrum method and optimized for low-coherence areas. We pay particular attention to the phase unwrapping of sub-band interferograms, to the filtering of the estimated ionospheric phase screens and to the time-series inversion of these phase screens. To evaluate the efficacy of this method to retrieve subtle deformation rates in Equatorial regions, we compute time-series using four ALOS-PALSAR datasets in contexts of low to medium coherence, showing slow (mm/yr to cm/yr) deformation rates of tectonic or volcanic origin. The processed tracks are located in Ecuador, Trinidad and Sumatra, with datasets typical of ALOS-PALSAR archive, including 15 to 19 acquisitions. They include very high, dominating ionospheric noise, corresponding to equivalent displacements of up to 2 m.

The correction method performs well and allows to reduce drastically the noise level due to ionosphere, with significant improvement compared with a simple ramp fitting method. This is due to frequent highly non-linear patterns of perturbation, characterizing Equatorial TEC distribution. From a geostatistical analysis, we derive an empirical accuracy of the LOS velocity derived from the corrected time-series. We design a statistical tool to quantify the uncertainty of the corrected time-series, highlighting its dependence on spatial distance. Thus, from the typical ALOS-PALSAR archive, using our ionospheric mitigation procedure, one can expect to be able to detect deformation rates of ~6 mm/yr at large distances (> 50 km), typical of interseismic strain accumulation. Looking at smaller wavelength deformation patterns (< 10 km), typical of fault creep, one can expect a detection threshold of around 3 mm/yr. These values are consistent with the accuracy derived from the comparison of velocities between two tracks in their overlapping area. In the case studies that we processed, time-series corrected from ionosphere allow to retrieve accurately fault creep and volcanic signal but it is still too noisy for retrieving tiny long-wavelength signals such as slow interseismic strain accumulation.

Shiveluch volcano is the northernmost volcano of the Kamchatka Peninsula, located 45 km from the village of Klyuchi. On the peninsula, the volcano is one of the most active and extremely dangerous. It has been erupting almost constantly since the beginning of the XX century. Its eruptions are characterized as paroxysmal explosive, they can be catastrophic, often accompanied by powerful ash emissions and, as a rule, pyroclastic flows. After a powerful explosive eruption on August 29, 2019, the dome collapsed and a pyroclastic flow descended. The mixture, consisting of volcanic gas, ash and stones, thrown into the air at the time of the explosion, settled on the southeastern slope of the volcano.

We used series of SAR images of the European Space Agency Sentinel-1A satellite for the period from May to October of 2020 and 2021 years. The maps of displacement rate of the volcano surface revealed an area with large subsidence, which coincides with location of pyroclastic flow on the southeast slope. The maximum average displacement rates on 2020 and 2021 were 385 and 257 mm/yr respectively.

We investigate possible causes of the subsidence of the pyroclastic flow surface, which formed during the eruption volcano Shiveluch on 29 August 2019. First, we estimated thickness of the pyroclastic deposits with SAR radar images for 2020 year. Subsidence rate has sufficiently high correlation coefficient (-0.69) with pyroclastic flow thickness, but shows a substantial dispersion.

Then we developed a thermo-mechanical model, which takes into account compaction of deposits due to changes of porosity and density over time. The model explains the dependence of the subsidence rate of the flow surface on the pyroclastic layer thickness when assuming flow cooling and a little decrease of porosity. The decrease of porosity depending on the initial pyroclastic flow temperature ranges from 1.5 to 1.7% during 2 years from 2019 to 2021. Dispersion of data around dependence "subsidence rate – flow thickness" explained by processes of erosion of pyroclastic deposits.

We present results from analysis of full-resolution, multi-year SLC stacks in an arid region impacted by a range of soil moisture conditions. The study region, along the southern coast of the Arabian peninsula, experienced three large rain events during the time period 2017-2020, some of which resulted in widespread flooding, loss of life, and damage to infrastructure. The region does not contain any large-scale deformation signals and includes broad areas of low topographic relief and fairly constant land cover/soil type, making it a good location for a study that aims to separate out the effects of soil moisture from other factors that affect InSAR data. We show the results of a correction approach that reduces the impact of large rain events on coherence and phase closure. We also illustrate how we can see the effects of soil mositure on both VV and VH observations (Sentinel-1 imagery). On a pixel-by-pixel basis, in regions where coherence is low for pairs that include a wet and dry date, but is high for interferograms between two dry dates over even longer time intervals, we find that there is often a near-linear relationship between coherence and phase at a given pixel. The slope of this relationship varies from pixel to pixel, where present. Some pixels to not appear to experience any significant changes relative to their near neighbors when soil moisture changes, others have a large phase difference from their neighbors, to a similar degree, each time it rains. We model this effect with an exponential distribuion of "soil moisture sensitivity", with most pixels exhibiting little effect but a few pixels having a strong dependence on soil moisture. This simple model can reproduce the observed trends in coherence magnitude and phase closure that we see in the real data. We show how we can build our model of "soil moisture sensitivity" for each pixel with as few as two storms, and use this model to reduce the impact of soil moisture change on a third, independent rain event. We also present synthetic data using our model that reproduces this result, and predicts the sorts of biases to the long-term inferred displacement rate that other workers have observed when they use the shortest-timescale interferometric pairs compared with set of longer-timespan pairs.

Tectonic deformation in northern Central America results from the interaction between the Cocos, Caribbean, and North America plates. This deformation is mostly accommodated by the sub-parallel Motagua and Polochic left-lateral faults, north-south-trending grabens south of the Motagua Fault, the Middle America subduction zone, and right-lateral faults along the Middle America volcanic arc (including the El Salvador fault zone and Jalpatagua faults in El Salvador and Guatemala, respectively). Large earthquakes associated with these faults include the destructive 1976 Mw 7.5 earthquake along the Motagua fault and the 2012 Mw 7.5 Champerico subduction thrust earthquake.

We show the potential of permanent scatterers and distributed scatterers (PSDS) InSAR techniques applied to a Sentinel-1 (S1) archive, to retrieve current deformation at large scale in this complex tectonic context. We analyze a time series of S1 radar images spanning from 2014 to 2022, along two ascending and two descending tracks covering most of Guatemala, El Salvador and western Honduras. The wide area PSDS interferometry approach (based on Adam et al., 2013, Ansari et al., 2018, Parizzi et al., 2020) includes corrections for tropospheric and ionospheric phase delays and solid earth tides. The resulting displacement time series are referenced to GNSS data (only one constant is adjusted per independently-processed frame) and decomposed into one linear and two seasonal terms. We present the InSAR-based velocity field for this region corresponding to the linear term dominated by tectonics, and analyze its spatial variations in map and along key profiles across the main faults.

Our results show a good first order agreement with GNSS data and with the most recent GNSS-based elastic-kinematic block models for the region (Ellis et al., 2019; Garnier et al., 2021; 2022). They highlight the North America and Caribbean plates' relative motion, accommodated mainly on the Motagua fault as well as on the Polochic fault. They also evidence significant internal east-west extension of the Caribbean plate between Honduras and western Guatemala, and show right-lateral slip across the Mid-America arc, with a clear velocity contrast across the El Salvador fault zone. The unprecedented high spatial density of our InSAR results allows to reveal a 40 km-long creeping section along the Motagua fault; we extract the along-strike variations of the creep and discuss them in regards of the local geology and of the co- and post-seismic slip distribution of the 1976 earthquake. Due to their sensitivity to vertical motion, our InSAR measurements also allow more refined estimates of lateral coupling variations along the subduction interface. We illustrate such sensitivity through forward block models with varying coupling values and depths along the subduction.

Finally, we also explore the non-tectonic signal and seasonal terms of the observed deformation, which include residual atmospheric signal, anthropogenic deformation (e.g. subsidence related to groundwater extraction) and hydrology-related seasonal variations.

Adam, N. et al. (2013), doi: 1857-1860. 10.1109/IGARSS.2013.6723164

Ansari, H. et al. (2018), doi: 10.1109/TGRS.2018.2826045

Ellis, A. et al. (2019), https://doi.org/10.1093/gji/ggz173

Parizzi, P. et al. (2020), doi: 10.1109/TGRS.2020.3039006

Garnier et al. (2021), https://doi.org/10.1130/GES02243.1

Garnier et al. (2022), https://doi.org/10.1029/2021TC006739

Synthetic-Aperture Radar (SAR) images are becoming more and more popular due to their resilience
against adverse weather conditions and clouds. However, the rapid growth of SAR data places
a significant burden on its storage and transmission. Consequently, efficient SAR data compression
algorithms are needed, particularly to optimize bandwidth and downlink time after spaceborne acquisitions.
In the last decade, numerous compression algorithms for SAR images have been proposed, some of
them being based on optical image compression standards, such as JPEG, JPEG2000 or SPIHT [1].
In order to perform compression, these algorithms rely on transformations such as the Discrete Cosine
Transform (DCT) or the Discrete Wavelet Transform (DWT) to achieve spatial decorrelation. Subsequently,
in case of lossy compression, the generated decorrelated coefficients are quantized before
being encoded in a bit-stream to be downloaded to the ground.
With the rise of Machine Learning methods to tackle remote sensing image processing problems,
researchers have proposed various Convolutional Neural Network (CNN) architectures to perform SAR
data compression [2, 3]. The structure of autoencoders, with their latent space, naturally complies to
the spatial decorrelation step necessary to compress the images.
The SAR image compression can be performed on-board, with a forward pass through the Encoder
followed by the quantization and encoding of the latent space to further reduce the bit-rate. The
generated bitstream is then transmitted to the ground, where the original image is reconstructed with
the Decoder.
While these models demonstrate promising performance, they are designed for ground-based processing
with millions of parameters and resource-intensive operations. On the other hand, on-board data
compression must meet the limited hardware resource constraints, be real-time and should minimize
energy consumption.
With this regard, this work presents a benchmark of an autoencoder for SAR data compression.
The model is constrained to fit in space-qualified hardware, especially FPGA boards that are commonly
deployed on-board satellites [4]. Comparison is made with traditional compression methods,
such as JPEG, JPEG2000 or SPIHT, using several image quality metrics and taking into account
the particularities of SAR signal. In future work, this light-weighted autoencoder will be tested on
Commercial-Off-The-Shelf (COTS) components suitable for space application.
References
[1] G. Yu, T. Vladimirova, and M. N. Sweeting, “Image compression systems on board satellites,”
Acta Astronautica, vol. 64, pp. 988–1005, May 2009.
[2] Q. Xu, Y. Xiang, Z. Di, Y. Fan, Q. Feng, Q. Wu, and J. Shi, “Synthetic Aperture Radar Image
Compression Based on a Variational Autoencoder,” IEEE Geoscience and Remote Sensing Letters,
vol. 19, pp. 1–5, 2022. Conference Name: IEEE Geoscience and Remote Sensing Letters.
[3] C. Fu, B. Du, and L. Zhang, “SAR Image Compression Based on Multi-Resblock and Global
Context,” IEEE Geoscience and Remote Sensing Letters, vol. 20, pp. 1–5, 2023. Conference Name:
IEEE Geoscience and Remote Sensing Letters.
1
[4] M. Caon, P. M. Ros, M. Martina, T. Bianchi, E. Magli, F. Membibre, A. Ramos, A. Latorre,
M. Kerr, S. Wiehle, H. Breit, D. G¨unzel, S. Mandapati, U. Balss, and B. Tings, “Very Low
Latency Architecture for Earth Observation Satellite Onboard Data Handling, Compression, and
Encryption,” in 2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS,
pp. 7791–7794, July 2021. ISSN: 2153-7003.
2

Understanding the mechanisms that control the activity of an eruption is one of the most important aspects of volcanic hazard forecasting. Multiple studies have identified the factors that appear to control their explosiveness, among which the most critical are related to the ascent and decompression rates of magma during the eruption, and then the magma overpressure. These processes in turn depend on internal factors associated with the magma itself, as well as external factors that can modify the conditions of the system and therefore its eruptive activity. Due to the complex interaction between the chemical and mechanical processes that take place in the magmatic system, these processes remain unclear. The Nevados de Chillán volcanic complex (NdCVC) in the Southern Volcanic Zone (SVZ) of Chile has experienced multiple explosive and effusive transitions during its last eruption that began in January 2016, and which extended over six and a half years. Through the analysis of long deformation time series from InSAR and GNSS, we have identified three episodes of surface deformation with a similar spatial pattern occurring between 2019 and 2023. These episodes correlate with effusive activity linked to a predominant magmatic phase of the eruption, whereas no deformation was observed during the first 3.5 years of the eruption when phreatic activity dominated. Petrological studies have concluded that the volcanic system underneath NdCVC is vertically zoned, composed of a shallower dacitic reservoir fed by less evolved magmas coming from a deeper reservoir, consistent with the widely accepted theory that vertically distributed mush zones are maintained by episodic recharge by deep magma into the upper crust. Based on these recent results, we implemented a numerical model consisting of a simplified plumbing system where two elastically deformable magma chambers are connected. A computational technique for approximate inference in state-space model is combined with this model and makes it possible to explore how the feeding process from a deeper reservoir to a shallow one can change the mechanical properties of the upper part of the plumbing system. This two-reservoir model well explains the temporal behavior of the displacement recorded by InSAR and GNSS at NdCVC. We present here the results of the satellite- and ground-based observations and discuss their implications for the understanding the dynamics of the plumbing system beneath the volcano and its eruptive activity.

National ground motion services, and more recently the products provided by the European Ground Motion Service, provide comprehensive deformation maps for stable radar scatterers, which typically correspond to man-made structures and terrain-types which are sparsely vegetated year-round, such as heathlands or bare-rock areas. However, in Denmark, as in many other countries, there is both a research and a commercial interest in monitoring also the ground deformations of other landscapes, such as cultivated peatlands, or rural areas where gas storage or extraction sites are located. The latter typically loose interferometric coherence at C-band, in the crop growth season, which typically spans from late-spring to early autumn, and are therefore void of measurements in the products provided by nation-wide PSInSAR-based monitoring services.

InSAR methods based on the inversion of networks of multi-looked interferograms, target distributed scatterers, rather than persistent ones, and can therefore be successful in observing the seasonal deformations of rural landscapes. However, care must be taken, to ensure that the resulting time-series are not affected by significant measurement biases. Several studies in recent years have shown that the latter may be introduced for instance by soil- and tree-moisture variability, and that these effects can be flagged by non-zero closure phases, formed between triplets of adjacent acquisitions.

In this study we consider different peatland areas in Jutland, Denmark, where corner reflector networks have been deployed by Geopartner Inspections since December 2021, within the ReWet project (https://projects.au.dk/rewet), which aims at providing a research platform for studies on peatlands under different management practices. These areas exhibit seasonal uplift and subsidence deformation patterns, which can reach up to 30 mm, and which show a strong spatial variability. We process the available Sentinel-1 data over these areas, which consist in general of two ascending and two descending radar tracks, using both a PSInSAR approach and a distributed scatterer (SBAS-like) approach. The former provides the relative motion between the radar reflectors year-round. The multi-looked InSAR measurements provide instead a more comprehensive mapping of the spatial pattern and variability of the seasonal deformations, which is however temporally confined to the autumn-winter seasons. We compare the time-series obtained from the inversion of different networks of multilooked interferograms against the PSInSAR results, to quantify the biases associated to the multi-looked measurements, and their relation to non-zero closure phases.

The European Ground Motion Service (EGMS) is the first operational service providing ground-motion measurements based on SAR-interferometry (InSAR) at a continental level [1]. It is part of the Copernicus Land Monitoring Service managed by the European Environment Agency (EEA). The EGMS is based on the full resolution InSAR processing of ESA Sentinel-1 radar data acquisitions and covers almost all European landmasses (i.e. all Copernicus Participating states) [2]. The first Baseline release includes ground motion time series from 2015 to 2020. Yearly updates of this open dataset will be released every 12 months, in Q3 of each year, except for the first one that was released in February 2023. Funds are ensured to continue the Service beyond 2024.

The EGMS employs persistent scatterers and distributed scatterers in combination with a Global Navigation Satellite System model to calibrate the ground motion products. This public dataset consists of three products levels (Basic, Calibrated and Ortho). The Basic and Calibrated product levels are full resolution (20 x 5 m) Line of sight velocity maps coming from ascending/descending orbits. The Ortho product offers horizontal (East-West) and vertical (Up-Down) velocities, anchored to the reference geodetic model resampled at 100 x 100 m. Since InSAR data production involves the application of thresholds and filters to remove unwanted phase artefacts, the results may contain systematic effects, outliers or simply measurement noise.

Independent validation is being carried out by a consortium composed of six partners to assess the quality and usability of the EGMS products. The validation is divided into seven separate validation activities: Point density check; Comparison with other ground motion services; Comparison with inventories of phenomena; Consistency check with ancillary geo-information; Comparison with GNSS; Comparison with in-situ monitoring; Evaluation XYZ and displacements with Corner Reflectors.

The subject of this abstract is to describe the comparison with ancillary geoinformation, which assesses the consistency of EGMS results with geological, geomorphological, and geotechnical data based on the concept of "radar-interpretation" described in [3]. The approach consists of an integration of InSAR measurements along with other ancillary data (land cover maps, geological maps, satellite images/aerial photos, topographic maps, fault systems, etc.) to obtain an accurate analysis of the studied phenomenon. Here, we use this approach to assess the general consistence of the EGMS products (Basic, Calibrated and Ortho) with the available ancillary geoinformation.

The validation sites for this validation activity have been chosen to cover a broad range of ground motion phenomena including urban subsidence, oil/gas or water extraction, mining, waste disposal site, and active faults. Depending on the validation site's characteristics and the ancillary datasets available, a selection of the following validation measures is applied: (a) the co-location of active deformation areas with spatial features in, e.g., geological units, topographic features, or spatial features in bedrock depth assessed; (b) the amplitude of the ground motion signal will be compared with geological structures, e.g., type of overburden or depth to bedrock; and (c) the consistency of the temporal evolution of the ground motion is compared to, e.g., mining activity or oil/gas production.

This consistency check will rely on statistical values calculated for certain areas/units depending in the ancillary geoinformation, as well as visual inspection by an expert. As the main objective for this validation activity is to provide a measure of plausibility of the EGMS products with the available ancillary geoinformation, the interpretation of the results by an expert is most important. Subsequently, key performance indices (KPI) are not directly calculated from statistical measures. Instead, the statistical measures are intended to help the expert in his interpretation of the data.

The comparison of EGMS products with ancillary geoinformation has been carried out in some sites in Norway, Spain, the Netherlands, Czechia, and Portugal and examples from these sites will be used to demonstrate the validation approach.

References

[1] Crosetto, M.; Solari, L.; Mróz, M.; Balasis-Levinsen, J.; Casagli, N.; Frei, M.; Oyen, A.; Moldestad, D.A.; Bateson, L.; Guerrieri, L.; Comerci, V.; Andersen, H.S. The Evolution of Wide-Area DInSAR: From Regional and National Services to the European Ground Motion Service. Remote Sens. 2020, 12, 2043. https://doi.org/10.3390/rs12122043

[2] Costantini, Mario & Minati, F. & Trillo, Fritz & Ferretti, Alessandro & Novali, Fabrizio & Passera, Emanuele & Dehls, John & Larsen, Yngvar & Marinkovic, Petar & Eineder, Michael & Brcic, Ramon & Siegmund, Robert & Kotzerke, Paul & Probeck, Markus & Kenyeres, Ambrus & Proietti, Sergio & Solari, Lorenzo & Andersen, Henrik. (2021). European Ground Motion Service (EGMS). 10.1109/IGARSS47720.2021.9553562.

[3] Farina, P., Casagli, N., Ferretti, A. (2008). Radar-interpretation of InSAR measurements for landslide investigations in civil protection practices. Proceedings of the 1st North American Landslide Conference. 272-283.

The main objective of the National Service of Observation ISDeform is to assist scientists in the monitoring of ground deformation related to natural hazards: earthquakes, landslides, volcanic activity, using optical and radar satellite imagery. The SNO actions include (a) the development of a database for images and products; (b) the evolution and operational maintenance of processing and visualization softwares; (c) the maintenance of on-line processing services and of systematic processing on the french territory; (d) the promotion of outreach activities related to remote sensing such astraining, short courses and MOOC.

Recent developments include the release of online services dedicated to on-demand processing: GDM-SAR for radar interferometry using Sentinel-1 images and GDM-OPT for cross-correlation using Sentinel-2 optical images. These services aim to provide high added-value satellite products: displacement fields, velocity maps, time series and Digital Surface Models (DSM) to support the use of satellite data by the scientific French community as well as internationalpartners in the South. The ISDeform service will also deliver standardized metadata to facilitate database searches and to ensure reproducibility of processing and interoperability at European level.

In addition, one of the missions of ISDeform is to routinely monitor the ground deformation for a selection of instrumented sites causing potential hazards that threaten the population. On these sites, the ISDeform service collect and process satellite data from various radar (Sentinel-1, TerraSAR-X, ALOS) or optical (Sentinel-2, Pleiades) missions. The targets are:

- active volcanoes located in overseas French territory: Piton de la Fournaise, Soufrière of Guadeloupe, Montagne Pelée and Mayotte

- the Indonesian volcano, Merapi, chosen as an analogue for French West Indies volcanoes

- active landslides located in mountainous regions in France: Harmalière, La Clapière, Avignonet, Super Sauze

For monitoring these sites, an adapted flowchart based on the InSAR processing chain NSBAS, called FAST-SAR (for Fully Automated processing for Small Targets using SAR images) is under development. The main objective of FAST-SAR is to routinely process radar images to obtain InSAR products on small areas as soon as new Sentinel-1 acquisitions are available. Such products will be available to the scientific community as well as to volcano and landslides observatories.

For the large-scale applications, the service ISDeform will deliver a velocity map of the ground deformation over France using the FLATSIM processing chain (ForM@Ter LArge-scale multi-Temporal Sentinel-1 InterferoMetry processing chain) to assess the impact of long-term geological or anthropological processes (e.g., seismic activity, hydrological loading, geothermal exploitation, clay swelling, tectonic loading).

The advent of the European Ground Motion Service (EGMS) offers chances and opportunities to EU Member States practitioners and researchers into the geohazard and infrastructure monitoring. As part of the EGMS validation team, under the lead of SIXENSE, Geosphere Austria carried out the in-situ validation activity for five test sites spread over Europe. The focus of this paper is the inter-comparison of different in-situ monitoring systems (geodetic tracking systems, GNSS, piezometers, levelling) in four different countries (Austria, Czech Republic, France and Spain) against the main products of the EGMS:

• Basic - Ascending and descending
• Calibrated - Ascending and descending
• Ortho – East-West and Up-Down

The comparison was performed in a JupiterHub environment created ad hoc for the validation project by our partner Terrasigna, which also developed a web-based validation data upload interface and a data catalogue (which follows the OGC and CSW standards). The workflow was developed in R language and validates error, precision and accuracy of the in-situ velocities and time series (TS) against the correspondent MT-InSAR values of the EGMS.

The workflow, made of several highly customisable modules, is reproducible and delivers directly tables and figures. More in detail, the R scripts:

• read and visualise the two datasets;
• perform a series of analysis such as smoothing (simplification), outliers search and trends extraction for both TS;
• inter-compare all the combinations of derived TS datasets and calculate for each couple RMSE, Coefficient of Determination (R²) and index of agreement;
• plot the TS and bar diagrams of the best scores in terms of minimum errors, maximum accuracy and maximum precision;
• deliver a Quality Index (QI) between 0-1 for each EGMS product;

The results of the in-situ validation activity for the EGMS product (2015-2021) will be presented and in depth analysed.

The type of ground motion phenomena took into account varies:

• deep seated landslide (Vögelsberg and Navis, Austria),
• subsidence due to active coal mine activity (Turow, Poland/Czech Republic),
• uplift due to abandoned mine activity (Forbach, France)
• subsidence due to water extraction (Lorca, Spain)

This validation activity provides a good example for discussing strengths and weaknesses of the EGMS products if compared to state-of-the art in-situ monitoring systems.

Deformation patterns at individual volcanoes are usually treated as isolated cases and interpreted on the basis of the individual characteristics of each volcano. Through a global analysis of deformation time series from InSAR and other geodetic techniques, we have identified a common temporal pattern during uplift episodes for all the volcanoes studied. We test the ability of common mechanical models to explain this pattern and conclude that fluid flow from a magma-intruded region to the adjacent porous rock is likely an important process in all cases. This has significant implications for our understanding of the mechanical controls acting beneath volcanoes and our ability to forecast volcanic activity. We also use this result, together with other temporal and spatial patterns of volcano deformation that we have identified, to develop a large database of simulated volcano deformation for machine learning applications.

Uplift signals have been observed worldwide and have classically been interpreted as the result of a magma reservoir filling at depth, and have therefore been identified as a possible precursor signal to volcanic eruptions. Although uplift of a few tens of centimeters has preceded several volcanic eruptions, large calderas have shown metric and long-lasting episodes of uplift without erupting, questioning then the magmatic origin of these episodes. At present, the processes behind volcanic uplift episodes are unclear and the classical models used to interpret them have become controversial due to their inherent assumptions that are not consistent with the expected mechanical behavior of a volcanic system. In the first part of this work, we compile time series of multiple volcanoes extracted from the literature, computed from InSAR, GNSS and tilt measurements. By comparing them, we identify a transitional time from which all uplift episodes follow the same temporal pattern of evolution, regardless of the volcano’s location and composition, etc, suggesting a common mechanism. By analyzing and comparing different mechanical models incorporating elasticity, viscoelasticity and poroviscoelasticity, our results suggest that the common post-transitional pattern is driven by fluid transport between the injected magma and the adjacent rock. It is then the adjacent rock acting as poroviscoelastic material, which will accommodate these fluids causing the increase in surface displacements for a time. Although all volcanoes appear to evolve in a similar way after this critical point, we show that the parameters describing this evolution vary from system to system, and it is these properties that control the time it takes for each volcano to reach a state where uplift ends.

In the second part of this work, we focus on the identification of typical spatial patterns associated with volcanic deformation. Through the development of an approach to automatically calculate surface displacement time series from Sentinel-1, we compare interferograms at different volcanoes globally and classify significantly similar deformation patterns. Together with the temporal patterns of deformation already characterized, we then explore different models to simulate volcanic deformation observed globally. In addition to considering magmatic sources interacting with the host-rock, we also consider non-magmatic sources as possible candidates to explain deformation at volcanoes, accounting for processes such as slow landslides, changes in hydrothermal systems, geothermal activity and slip on faults. Finally, we use these models to simulate thousands of interferograms, to which realistic noise is added, to train deep learning networks developed to detect and forecast deformation.

Differential Synthetic Aperture Radar Interferometry (DInSAR) is a microwave remote sensing technique that has been originally developed to investigate single events characterized by the surface displacements and is nowadays successfully exploited in different scenarios, such as those relevant to earthquakes, volcano eruptions and landslides, as well as deformation of anthropic structures like buildings, bridges and roads [1]. We further remark that a relevant extension of the original DInSAR technique, often referred to as advanced DInSAR, has been developed to investigate the temporal evolution of the detected deformations through the retrieval of the displacement time series of the investigated scenario. This is effectively achieved through the inversion of an appropriate set of multi-temporal interferograms produced from a sequence of SAR acquired images of the area of interest. Among several advanced DInSAR techniques, the Small BAseline Subset (SBAS) is a well-established approach which has been widely used for the analysis of several deformation phenomena [2].

For what concerns the advanced DInSAR methods, effective and robust Phase Unwrapping (PhU) algorithms have to be typically implemented and exploited in order to accurately retrieve the ground deformation signals. This operation represents a rather critical step for the retrieval of the displacement information because of the intrinsically ill-posed nature of the problem which may lead to solutions that, despite being mathematically correct, do not reproduce the actual unwrapped phase profile [3].

A common indicator for the quality of the PhU solution within advanced DInSAR methods like the SBAS technique [2] is the temporal coherence [4]. This is a point-like parameter available for methods where the displacement time-series are retrieved through the inversion of an overdetermined linear equation system [M,N] with M > N, where M is the number of the generated (redundant) interferograms and N represents the exploited SAR images, whose solution can be obtained in the LS sense.

We present in the following a simple solution to identify and correct possible PhU errors, based on a different and innovative use of the temporal coherence parameter as defined in [4].

In principle, the higher is the value of the temporal coherence, the better is the quality of the PhU solution for the analysed point. Unfortunately, the temporal coherence loses its sensitivity when the number of interferograms increases. Accordingly, to overcome this issue we propose to compute for each point a time series of local temporal coherences, i.e. computed by exploiting a limited number of interferograms. To do this, starting from the first acquisition date of the analysed dataset, we define a time window range, say Δw, and a time sampling, say t_i, where the step size Δt = t_i+1-t_i is selected in agreement with the satellite revisiting time. Accordingly, for the generic i-th step, we consider the time window centred around the t_i value and we calculate the temporal coherence on a limited subset of interferograms whose master and slave image pairs are included in the selected time window [t_i - Δw/2, t_i + Δw/2].

This solution is computationally efficient and allows us to regain sensitivity on possible PhU errors. Indeed, by doing so, the number of interferograms to be analysed in order to identify those characterized by PhU errors has been drastically reduced, making the local temporal coherence more sensitive to small variations in a single interferogram. A subsequent algorithm of PhU errors correction can be then applied only to the involved interferograms, strongly reducing the time computing and increasing the ability to spot and correct the wrong interferogram. In our case, to identify and subsequently correct the PhU errors we use a combined L₁-norm inversion and a genetic algorithm whose process is described in [5].

A more detailed description of the implemented procedure and an extended experimental analysis, based on Sentinel-1 datasets, will be provided in the final paper and at the conference time.

REFERENCES

[1] P. A. Rosen et al., "Synthetic aperture radar interferometry," in Proceedings of the IEEE, vol. 88, no. 3, pp. 333-382, March 2000.

[2] Manunta, M. et al., “The Parallel SBAS Approach for Sentinel-1 Interferometric Wide Swath Deformation Time-Series Generation: Algorithm Description and Products Quality Assessment”, IEEE Trans. Geosci. Remote Sens., 2019.

[3] H. A. Zebker and J. Villasenor. “Decorrelation in interferometric radar echoes”, IEEE Transactions on Geoscience and Remote Sensing, vol 30, no. 5, pp: 950- 959, September 1992.

[4] A. Pepe and R. Lanari, "On the Extension of the Minimum Cost Flow Algorithm for Phase Unwrapping of Multitemporal Differential SAR Interferograms," in IEEE Transactions on Geoscience and Remote Sensing, vol. 44, no. 9, pp. 2374-2383, Sept. 2006, doi: 10.1109/TGRS.2006.873207.

[5] De Luca C. et al. "A genetic algorithm for phase unwrapping errors correction in the SBAS-DInSAR approach." IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2019.

Several PSInSAR (Persistent Scatterer InSAR) approaches currently in use, are based on the analysis of phase differences between PSs connected in a sparse network, which are referred to as phase arcs. These approaches typically require a subsequent spatial integration step, often computed as a weighted least squares inversion, to yield the phase difference with respect to a common reference PS [1]. This spatial integration step can be highly sensitive to the weighting scheme chosen for the inversion, in particular when the spatial distribution of the PSs exhibits gaps due to decorrelating surfaces (e.g. due to vegetation, water, snow/ice, etc.).

In our work we adapt the concept of connectivity, first proposed to characterize the reliability of phase unwrapping in a DInSAR (Differential InSAR) context [2], to a PSInSAR processing scenario.
Connectivity, in its original formulation, represents a quality parameter for the ensemble of possible paths connecting any two interferogram pixels, where each path consists of a sequence of wrapped phase differences. Once a quality metric, such as the magnitude of interferometric coherence in the DInSAR case, is assigned to each phase arc, connectivity represents the worst link on the best path connecting two pixels, and it can be calculated using a modified version of Dijkstra’s algorithm [3].
In our adaptation, connectivity is computed between the reference PS and every other measurement point on the sparse PS network, using temporal coherence as a quality metric, instead of interferometric coherence.
The assumption behind this approach is that while temporal coherence provides insight into the quality of each phase arc, connectivity provides insight into the full integration path needed to reach each PS. Thus, the connectivity concept provides a more holistic view of the PS network, while also considering the placement of the reference PS.

The aim of this study is twofold: to investigate if connectivity can reduce the sensitivity to some critical processing parameters, which affect the aforementioned spatial integration step; to investigate to what extent this parameter can be used for error characterization.

To quantify the impact of connectivity we simulate a realistic ground deformation pattern with spatially correlated noise to account for atmospheric delays, and spatially uncorrelated Gaussian noise to account for phase changes related to decorrelation. We consider a real PS network, based on a TerraSAR-X dataset covering the greater Copenhagen area, comprising urban areas with a high PS density, as well as lakes and forests void of PSs.

We analyze the phase integration errors arising from the choice of different processing parameters, and the effect of connectivity thresholding to reduce the inconsistencies between processing results. For a given choice of processing parameters, we then investigate whether the connectivity of a given PS is a good predictor of the phase integration errors affecting it.

Connectivity is found to provide complementary information compared to temporal coherence, regarding the quality of phase inversion carried out in a PSInSAR context.

[1] A. Ferretti, C. Prati, and F. Rocca, “Non-linear subsidence rate estimation using permanent scatterers in differential SAR interferometry,” IEEE Trans. Geosci. Remote Sensing, vol. 38, pp. 2202–2212, Sept. 2000

[2] L. Galli, "A new approach based on network theory to locate phase unwrapping unreliable results," IGARSS 2001. Scanning the Present and Resolving the Future. Proceedings. IEEE 2001 International Geoscience and Remote Sensing Symposium (Cat. No.01CH37217), Sydney, NSW, Australia, 2001, pp. 118-120 vol.1, doi: 10.1109/IGARSS.2001.976075.

[3] E. W. Dijkstra, “A note on two problems in connexion with graphs,” Numer. Math., vol. 1, no 1, pp. 269-271, Dec. 1959.

Phase Unwrapping (PU) has long been a tricky problem for InSAR data processing. With the wide application of the InSAR time series, PU has become an even more pressing issue given the fact that PU errors can propagate from the point where they occurred and affect all subsequent acquisitions. Phase consistency (or closure phase), the sum of phase gradients around a loop of three or more interferograms, can be used to detect and further correct PU errors. It is based on the assumption that even after multilooking and spatial filtering, the closure phase will still be within [-pi, pi], and any values beyond this range will be treated as PU errors, which can potentially be corrected by adding modulo 2pi to one or more of the interferograms. The 3-D Minimum Cost Flow (MCF) algorithm, instead of applying phase consistency check after the PU, utilizes the phase consistency as an additional constraint during the PU. This constraint is based on the prior knowledge that the sums of unwrapped phase gradients, that belonging to different interferograms in the closure phase, can be obtained before PU. The 3-D MCF algorithm does not require a temporal deformation model nor preliminary atmospheric phase calibrations, and can reduce the chance of phase aliasing by combining the MCF model and phase consistency. However, the size of the design matrix expands rapidly with the increased volume of the dataset, making the 3-D MCF approach very memory and time consuming; thus it is very difficult or even impossible to apply to a large dataset (e.g., hundreds of interferograms, each with millions of pixels). To overcome the efficiency issue, here we present a divide and conquer approach for quick 3-D MCF PU. Instead of solving all the pixels on all interferograms simultaneously, we first pick out the pixels that are inconsistent in space or time, which are identified using the wrapped phase gradients. We then divide these pixels, also including the ‘good’ pixels that are connected to them in space or time, into several patches based on their spatial and temporal relationship. Finally, we set up the design matrix for each patch, whose size now is significantly smaller compared to the original method that uses all the pixels, and solve these equations independently. Our preliminary tests run successfully and achieve good results on a laptop using a medium size dataset (30 Sentinel-1 interferograms with ~300, 000 pixels on each). We will also present results on much larger datasets to evaluate the performance of the algorithm. In summary, our improved algorithm, which can also easily be parallelised, greatly enhances the performance of the 3-D MCF algorithm, which is essential for processing the large InSAR datasets that are now routinely acquired.

Tidal flats are active transition zones between land and ocean. Their dynamics and morphological evolution are driven and affected by oceanic and fluvial processes such as tides, waves and river-flow, and anthropogenic activities such as land subsidence, land reclamation, and dredging. Consecutive monitoring of the tidal flat dynamics, particularly tidal flat DEM dynamics, is of significance to recognize coastal erosion and changes in natural ecosystems. Yet, as tidal flats can fluctuate dramatically, even on a daily basis, this requires wide-area, high-density, frequent and long term monitoring. Consequently, in-situ point-based techniques like GPS over wide areas are sub-optimal and extremely expensive. Therefore, in this study, we resort to both radar and optical satellite observations from space, as they cover the entire Earth with high-frequent updates and up to meter-level spatial resolution. We treat radar and optical images as the main input to develop a method for tidal flat dynamic DEM generation. Within this method, we propose a way to exclude noisy SAR observation based on the analysis of its polarimetric features, and a way to align both radar and optical images in a common reference system, and we use Object-based image segmentation (OBIS) to determine waterlines and delineate tidal flats, sub- and supra- tidal regions. The water level is estimated by the Delft3D model, which is then used for tidal flat rim’s height interpolation at every satellite acquisition time. To test and demonstrate this method, we used 132 Radarsat-2 SAR, 199 Sentinel-1 SAR and 157 Landsat images acquired between 1986 and 2020, covering the Dutch Wadden Sea tidal flat regions. We extracted the coastline and sandbank information over the past 34 years and 10 DEM instances from 2011 to 2020. The generated DEMs match well with high-resolution Lidar and sediment measurements. The mean absolute error is about 20 cm. We found that the area of coastlines and sandbanks expanded at a rate of 0.1074-0.3241 km^2 yr^−1 and 0.010-0.073 km^2 yr^−1, respectively, while the net volume of tidal flats increased by approximately 8.6 x 10^7 m^3. We conclude that our method demonstrates the potential of using space-borne radar and optical images for generating tidal flat DEM dynamics for more than three decades with relative high accuracy, and our method is suitable for large scale tidal flat mapping and change detection.

The karst hydrosystem of Fontaine in Vaucluse is located in the Cretaceous limestone massif in southeastern France. With a 1162 km² impluvium, this karst is a multi-instrumented site for measuring the spatio-temporal evolution of water flow, surface deformation (GNSS, inclinometers), seismic and gravimetric signatures. The SAR Sentinel-1 image archive is an exceptional database for the construction of high resolution time series of surface deformation over the whole region. We use the InSAR time series calculated with the NSBAS processing chain (Doin et al., 2011; Grandin et al., 2015) in the framework of the Flatsim project (CNES/ForM@Ter; Thollard et al., 2021) and the French ISDeform National Observation Service. The objective of this study is to extract the low amplitude deformation associated with the evolution of the water stock in the karst and the hydrological processes at depth (constraints on lateral flows, flow networks, system response to loading, etc.). One of the main challenges is to separate the atmospheric signal and the deformation signal which are both affected by seasonal variations. First, we test “blind methods”, such as PCA or ICA, in order to evaluate the temporal behavior of the surface deformation. This analysis helps to identify distinct areas affected by various behaviors that could be related to the 3D spatial distribution of the water reservoir(s) which is not fully known for the whole karst. In particular, we aim to track the respective role of the porous matrice and the karstic conduits within the 800 m thick unsaturated zone on the circulation of water from the surface to the saturated zone. The combinaison of data acquired along ascending and descending tracks will make it possible to separate horizontal and vertical components and thus help to define the origin of the deformations. Second, the time series will be analyzed taking into account external geophysical inputs such as the water flow of the Vaucluse Fountain and precipitation which is mainly due to storms resulting from air streams coming from the Mediterranean Sea. We interpret the extracted signals in relation to the observables acquired on the karst. The delays and threshold effects between rainfall loading and deformation will be highlighted in order to provide constraints on the dynamics of hydrological networks under the ground, and more specifically the buffer stock of water and the non-linear effects in the non-saturated zone.

The western Galapagos volcanoes are a geologically active region and have experienced over 10 eruptions since 1991, by the time after the launch of the ERS-1 SAR system. Among them, 6 eruptions have occurred since the operation time of ALOS PALSAR. Active volcanoes often exhibit long-term deformation behaviors due to the reservoir’s pressurization [1], and accurate monitoring of its deformation pattern is essential for hazard assessment and process understanding. Synthetic aperture radar interferometry (InSAR) is a remote sensing technique widely used for monitoring surface deformation with geophysical processes in millimeter to centimeter precision. However, the ionosphere is one of the primary error sources in InSAR measurements, particularly in low-latitude regions [2], i.e., the Galapagos archipelago, where the ionosphere varies in different spatial scales and ionospheric scintillation is prevalent. In addition, the low-frequency SAR systems, i.e., ALOS PALSAR in L-band, are more sensitive to ionospheric variations. Hence, mitigating the anisotropic ionospheric artifacts in the multi-temporal ALOS PALSAR data is essential for a better understanding of the magnetic deformation over the western Galapagos volcanoes.

In this study, a total of 22 ALOS PALSAR images obtained between January 2007 to March 2010 over the western Galapagos were used to investigate the anisotropic ionospheric artifacts and to extract the precise surface deformation. We processed the data using the small baseline subset (SBAS) algorithm [3] to obtain the time series of surface deformation, and 152 interferograms were generated with given spatial and temporal baselines. To evaluate the influence of the ionospheric variations on these interferograms, we first derived the azimuth deformation using the multi-aperture InSAR (MAI) algorithm [4]. The results indicate that 57.3% and 23% of the analyzed interferograms were affected by the background changes and anomalies in the ionosphere, respectively, while 19.7% of them were influenced by strong ionospheric scintillation. Subsequently, we adopted the range split-spectrum method [5], aided by MAI interferograms, to effectively mitigate the anisotropic ionospheric artifacts. Finally, the time-series analysis revealed that an uplift of up to 34.10 cm/year was observed in the caldera of Sierra Negra volcano, and subsidence of up to 15.18 cm/year was detected in the lava flow region of the 2008 eruption of Cerro Azul volcano. These findings provide valuable insights into the deformation and geodynamic processes of the western Galapagos volcanoes.

REFERENCES:

[1] E. Chaussard, F. Amelung, and Y. Aoki, "Characterization of open and closed volcanic systems in Indonesia and Mexico using InSAR time series," Journal of Geophysical Research: Solid Earth, vol. 118, no. 8, pp. 3957-3969, 2013.

[2] F. J. Meyer, K. Chotoo, S. D. Chotoo, B. D. Huxtable, and C. S. Carrano, "The influence of equatorial scintillation on L-band SAR image quality and phase," IEEE Transactions on Geoscience and Remote Sensing, vol. 54, no. 2, pp. 869-880, 2016.

[3] P. Berardino, G. Fornaro, R. Lanari, and E. Sansosti, "A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms," (in English), IEEE Transactions on Geoscience and Remote Sensing, vol. 40, no. 11, pp. 2375-2383, Nov 2002, doi: Doi 10.1109/Tgrs.2002.803792.

[4] N. B. D. Bechor and H. A. Zebker, "Measuring two-dimensional movements using a single InSAR pair," Geophysical Research Letters, vol. 33, no. 16, p. L16311, 2006.

[5] G. Gomba, A. Parizzi, F. De Zan, M. Eineder, and R. Bamler, "Toward operational compensation of ionospheric effects in SAR interferograms: the split-spectrum method," IEEE Transactions on Geoscience and Remote Sensing, vol. 54, no. 3, pp. 1446-1461, 2016.

The Dead Sea fault, the ~1000-km-long left-lateral transform plate boundary in the eastern Mediterranean between the Sinai and Arabian plates, has been extensively studied since the 1950s. Geological studies, GPS observations, and plate motion models show that the slip rate of most of the fault is about 4-5 mm/year, with the Arabian plate to the east moving north, with respect to the Sinai plate to the west. InSAR observations, on the other hand, have not provided useful information about the present-day strain accumulation on the Dead Sea fault, due to the north-south orientation of the fault and the insensitivity of InSAR to north-south displacements. To overcome this, we used time-series analysis of along-track burst-overlap interferometric (BOI) observations along the entire Dead Sea fault, from both ascending and descending orbit Sentinel-1 data from 2014-2021, to retrieve the horizontal along-track displacements in burst-overlap areas. To improve the results, we applied a point-selection method and spatial filtering, as well as stacking of several adjacent BOI areas, yielding a clear picture of the interseismic deformation at the different sections of the Dead Sea fault. Elastic modeling based on the BOI observations indicates the Dead Sea fault slip rate gradually decreases from south to north. In the south, in the Gulf of Aqaba and Wadi Araba, we find a slip rate of 5 mm/year and 4.7 mm/year, respectively. North of the Dead Sea and the Carmel splay fault, in Jordan Valley, a lower rate of 3.8 mm/year is found. Further north, the Yammouneh fault, a part of the Dead Sea fault, cuts across the Lebanon restraining bend and here we find a slip rate of 3.4 mm/year. At the northern Dead Sea fault in Syria, we find an even lower rate of 2.8 mm/year, indicating the slip rate in the north is significantly lower than for the southern Dead Sea fault. Our results are in accord with GPS observations, where they are available, and also demonstrate that low rates of a few millimeters per year can be resolved by BOI time-series analysis, even in areas with medium-to-low coherence. These findings contribute to a more comprehensive understanding of plate kinematics in the eastern Mediterranean and show that the earthquake hazard of the Dead Sea fault decreases towards the north.

I. INTRODUCTION

For many geophysical applications, the use of radar and optical images is very complementary and gives valuable information, e.g. for earthquake induced surface displacement measurement, landslide monitoring, change detection, flood or more generally damage mapping. For multi-modality analysis, it is necessary to properly register these images. However, the automatic registration of radar and optical images still remains a difficult task. This is due to the different nature of the sensors used to acquire these data, leading to different geometries and to various intensities for a common acquisition area. Many algorithms have been developed for the automatic registration of optical and SAR data, based on various techniques, such as mutual information, primitive extraction, descriptors (e.g. SIFT, BRISK) and more recently DL (Deep Learning) based methods. These last methods often use radar to optical or optical to radar translation in order to help the registration step. In this abstract, a method for automatic registration of SAR and optical images is presented. Our algorithm, called OSCAR (Optical and SAR Correlation-based Automatic Registration), generates fake SAR images as many DL based methods. However, in our case, the fake SAR images are obtained by usual image processing filters. If available, Digital Elevation Models are used to project the optical images into SAR geometry and to enhance the fake SAR images. The algorithm was applied to several datasets acquired by sensors of various resolutions (optical Pléiades Neo, SAR Sentinel-1 and TerraSAR-X). The results show that the proposed algorithm gives robust results and reduces the RMSE (Root Mean Square Error) from several tens of pixels to only a few pixels. First, the principle of the proposed algorithm is described. Then, the data and an experiment realized for precise quantitative evaluation are presented. Finally, registration results are qualitatively and quantitatively evaluated.

II. PRINCIPLE

Our algorithm can be applied either to images projected in SAR geometry or to orthorectified images. There are two variants of our algorithm:

- The first one can be used if the topography is almost flat. In this case, native or orthorectified geometry images can be processed. This version is called OSCAR.

- The second one is recommended when the topography is not flat (urban areas, montaineous areas). In this case, an accurate Digital Surface Model (DSM) is required as input to the algorithm. The optical image is then projected into the SAR geometry using this DSM. This second version is called OSCAR-topo.

A. Generation of fake SAR images

The principle of OSCAR is to produce fake SAR images from the optical image. For both versions of OSCAR, five fake SAR images are simulated. For the OSCAR-topo version, these fake SAR images are enhanced by taking into account the geometry. Optical and SAR images are very different. The aim is to simulate fake SAR images from optical images using simple filters and simple physical observations. Flat areas generally appear homogeneous in optical areas as there is no shadow, unless there is a change in color or texture. These areas are generally dark in SAR images, in particular very flat areas like water or roads. On the contrary, when an area is not flat, they generally appear less homogeneous on optical images as there is some shadowed and lightened pixels. On such areas, the SAR image is generally quite bright because there may be double-bounce signals returning towards the satellite or simply some surfaces oriented towards the satellite. In between, surface like non flat vegetation generally appears with medium amplitude both in optical and SAR images. Of course, there are many examples where this over simplified model does not hold. For example, terrain relief may imply radar shadows where there is no signal. Reciprocally, some shadowed areas on optical images appear homogeneous but may be bright in SAR images. Five filters have been applied to optical images.

- Standard deviation on a square window of dimensions WxW pixels (W is set to 3 by default).

- Minimum of standard deviation on a square window of dimensions WxW pixels. Indeed, one of the drawbacks of the standard deviation is that it tends to thicken the edges by producing a high standard deviation for all variants of our algorithm: pixels closer than W/2 pixels to an edge. The calculation of the minimum of the standard deviation on a square window of the same dimension thus allows to better locate the edges on the filtered image.

- Sobel filter

- Morphological gradient

- Absolute value of Laplacian

They globally highlight the edges on the optical image. For the OSCAR-topo variant, the radar geometry is also taken into account to simulate the fake SAR images. We use a VHR DSM derived by photogrammetry applied on stereo optical images that is perfectly superimposed with the optical images following a method described in [1]. The amplitude of a SAR image is proportional to the product of the square root of the pixel area and the cosine of the local incidence angle i.e. the angular difference between the wave direction and the local normal to the surface. The algorithm also identifies SAR shadow areas and computes a binary mask set to 0 for all shadowed pixels. The FakeGeom image is the product of the three different geometric contributions (area, incidence and shadow mask). Then, each fake image is multiplied by the FakeGeom image in order to obtain the five final fake images noted Fake1 to Fake5.

B. Correlation step and multi-scale processing

The images are downsampled for coarse registration before full resolution fine registration. At each scale, the SAR image highest values are thresholded. Then, each of the five fake images Fake1 to Fake5 is correlated with the true SAR image by a Fourier phase correlation. We then obtain five disparity maps. It is well known that such maps often contain outliers. RANSAC (RANdom Sample Consensus) method [2] is used here and models the disparities by a similarity transformation, i.e. translation, rotation and scale. At each scale, RANSAC estimates a transformation which is applied to the SAR image at the next finer scale to help correlation. The final estimate of the transformation is the sum of the transformations estimated at each scale. Finally, the SAR image is resampled and registered onto the optical image.

III. TEST AREAS, DATA AND EXPERIMENT

In practice, it remains difficult to compare SAR-optical registration algorithms of the litterature because there is no common dataset. The existing open datasets [3], [4] are composed of very little images which are not representative of real cases and would be difficult to register for many state-of-the-art algorithms. In particular, many algorithms using multiscale strategies as ours would not be adapted to such very little images. The best qualitative assessment would be to compare the RMSE before and after registration. However, precise manual pointing of Ground Control Points (GCP) is often a highly tedious task due to the difference between SAR and optical images. This has justified the need for an experiment with colocalized SAR corner reflectors and optical “reflectors”. Corner reflectors have been installed between 06/07/2022 and 04/08/2022 on Brétigny-sur-Orge former aerodrome, in southern Paris suburbs. These corner reflectors are in fact a couple of corner reflectors such that they remain visible on ascending and descending right-looking acquisitions and correspond to the same phase center. They were installed right on the middle of tarpaulins that can be easily identified on optical images. It enables the colocation of tie points on radar and optical images. Tarpaulins are 4 m by 4 m blue squares. Brétigny area is globally flat and includes an aerodrome, agricultural areas, urban areas and little forested areas. Three radar images with medium and very high resolutions have been used for our test. TerraSAR-X Spotlight images with about 1 m resolution have been acquired on ascending and descending orbits with a right-looking view. Sentinel-1 (S1) image is a dual-polarization 10 m resolution orthorectified TOPSAR image acquired on a descending path. The optical image is a Pléiades Neo image (PNEO) acquired on 08/07/2022 with a resolution of 32 cm. It has been orthorectified with a 50 cm resolution.

IV. RESULTS AND CONCLUSION

OSCAR has been tested with its two versions. The first one consists in registering optical and radar images projected in orthorectified geometry. It has been applied to the S1 data and to the PNEO image resampled to the S1 resolution. The second one consists in registering the optical image with the radar image acquired on the ascending (resp. descending) orbit directly in radar geometry. The projection in SAR geometry has been done with internal processing chain and use of VHR DSM computed by photogrammetry using PNEO stereo acquisition. In this case, the images were registered by OSCAR-topo. Corner reflectors and other tie points have been manually marked on PNEO image and on TerraSAR-X radar images before and after registration. For TerraSAR-X descending image and PNEO, the results show that the RMSE decreases from about 265 m (208 pixels) to about 2.8 m (2.2 pixels) for OSCAR and 1.6 m (1.2 pixel) for OSCAR-topo. For TerraSAR-X ascending image and PNEO, the results show that the RMSE decreases from about 212 m (167 pixels) to about 1.9 m (1.5 pixel) for OSCAR and for OSCAR-topo. This suggests that even even for this semi-urban flat area, OSCAR-topo may help registration. For Sentinel-1 and PNEO, it is difficult to find tie points to measure the RMSE due to coarse resolution. Visually, the images are very well registered and the initial offset is estimated to about 224 m (45 pixels).

As a conclusion, this experiment shows on our test site that OSCAR is able to achieve very precise registration between SAR and optical images. Further tests on other areas (denser urban areas, agricultural landscapes, mountainous areas) using different data (Cosmo-SkyMed, Sentinel-2, Pleiades, Ikonos) will be made to better qualify the performance of OSCAR.

ACKNOWLEDGMENTS

We thank Arnaud Bazin (Drone Center) and our colleagues for their help during the experiment.

REFERENCES

[1] C. Guerin, R. Binet, and M. Pierrot-Deseilligny, “Automatic detection of elevation changes by differential dsm analysis: Application to urban areas,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 7, no. 10, pp. 4020–4037, 2014.
[2] M. A. Fischler and R. C. Bolles, “Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography,” Commun. ACM, vol. 24, pp. 381–395, 1981.

[3] Y. Wang and X. X. Zhu, “The SARptical dataset for joint analysis of SAR and optical image in dense urban area,” 2018. [Online]. Available: https://arxiv.org/abs/1801.07532

[4] M. Schmitt, L. H. Hughes, and X. X. Zhu, “The sen1-2 dataset for deep learning in sar-optical data fusion,” 2018. [Online]. Available: https://arxiv.org/abs/1807.01569

In nation-wide radar satellite time series data of Germany provided by the German Ground Motion Service based on Sentinel-1 data (bodenbewegungsdienst.bgr.de), a linear subsidence motion of several kilometer spatial wavelength shows up south-east of Kiel, northern Germany. The center region of this signal, showing line-of-sight displacement velocities of about 2 mm/yr only, coincides with the facilities of a gas storage site managing two in-service and one out-of-service caverns in the salt dome beneath. The original cavity sizes of the two larger caverns exceed 400.000 m³ each, comparable to the volume of a large Gothic cathedral like the Cologne Cathedral. The salt dome beneath Kiel reaches up to depths of around 1000 m and the surrounding structure is well known from boreholes and other geophysical analyses. The roof layers above the dome consist of thick and competent deposits, mainly chalk, silk and claystone below layers of clays, silts, sands and glacial marls. The Kiel storage site is the oldest of Germany, one of the deepest and also smallest regarding the volumes in Germany.

Despite a thick and competent cover layer, the long-term ductile behavior of halite, which evidently causes shrinking of the cavern volumes through time, results in the observed continuous surface subsidence across several square kilometers. This set-up, surface displacement above a known source, presents a good opportunity for a controlled experiment. We can test geophysical modeling abilities as used in many geoscientific fields like volcanology, with small displacement signals and on a large scale. For the inverse modeling we use the Grond module of the seismological open-source software toolbox Pyrocko (pyrocko.org). We present a Bayesian optimization of an isotropic volume point source embedded in a viscoelastic host medium below a horizontally layered elastic roof medium to fit the surface subsidence signal. We use InSAR time series data from two ascending and two descending look directions. This model setup simplifies the actual and quite heterogeneous host rock structure considerably and the source problem with just one source model for three closely spaced caverns (within 500 m horizontal distances). Furthermore, the signal-to-noise ratio of the satellite data is rather small and they show considerable spatial gaps, where areas of agriculture and forests dominate.

Nevertheless this controlled experiment was very successful and provides confidence to our geophysical modeling approaches. The results show a cavern position that is within several meters to one of the large shrinking caverns. The estimated depth corresponds very well to the top of the caverns. Also the estimated volume loss of about 21.000 m³ per year also well matches repeated volume measurements inside the actual caverns pointing to 24.000 m³ per year.

The Tien Shan Mountain range in Central Asia plays a vital role in absorbing north-south convergence caused by the Indo-Eurasian collision. However, the region is still prone to large-magnitude earthquakes, posing a significant hazard to the local population. To monitor seismic activity in the region, GNSS and InSAR measurements have been used. However, sparse GNSS benchmarks are not sufficient for the large application. In addition, conventional (across-track) InSAR typically only offer precise data on the horizontal displacement in an east-west direction due to the near-polar orbit. We present along-track velocity results from Burst Overlaps Along-Track (BOAT) Interferometry, which is a technique possible with Sentinel-1 due to the TOPS acquisition mode that allows for precise measurements also in the north-south direction.

Although precision of measurements in burst overlaps is expected down to 1 mm, this technique is affected by additional error sources, decreasing the precision. We incorporate basic corrections for the solid Earth tides and ionosphere. Ionospheric influence affecting primarily data from ascending tracks that are acquired in dusk time when the ionosphere is more active. For this, we will compare ionospheric models of IRI2016 and CODE, and apply for the correction.

Our BOAT offsets are estimated from data resampled during their coregistration to a reference scene. The coregistration incorporated intensity cross-correlation and average offsets from spectral diversity over large number of burst overlaps, describing along-track shift of the scenes with regards to their expected footprints given precise orbits of the satellite. By adding the overall azimuth offsets to the BOAT offsets, we obtain along-track velocity estimates in a global reference frame of ITRF2014.

We apply this approach to the Tien Shan and surrounding regions where GNSS measurements are sparse in order to reveal along-track displacement in BOAT InSAR time series at a large scale. We integrate our along-track velocities with conventional across-track InSAR, that are relative measurements, and GNSS velocities to produce a 3D velocity field for the entire area that will be tied to the global reference frame. We will characterise the strain accumulation across Tien Shan and discuss the implications for earthquake cycle deformation and seismic hazard in the Tien Shan.

Persistent Scatterer Interferometry uses a stack of at least 20 SAR images to measure ground deformations with millimetric precision. An adequate interferogram network, with a well distributed connection between pairs of images and the appropriate combination of temporal and perpendicular baselines is essential to derive robust measurements. Using a high degree of redundancy of interferograms per image usually makes the InSAR processing more robust, but, as a result, it can be computationally expensive. Therefore, generating an interferogram network is necessary, especially when time is a constraint such as in crisis management. Here, we describe the strategy to constitute an optimal interferogram network.

When forming interferograms, different connections between images can influence the measurement of the deformation: magnitude, precision, accuracy, etc. On the one hand, interferograms with short temporal and perpendicular baselines are used typically selected to measure strong deformations. On the other hand, interferograms with large perpendicular baselines are also necessary to better estimate the topography and obtain a precise geocoding of the results. A priori, the more interferograms there are, the better the atmospheric terms can be estimated. Thus, the choice of a proper interferogram network on each case is important on InSAR studies.

Generally, interferogram pairs are generated by connecting the available images considering the user’s predefined choice of maximum and minimum temporal and perpendicular baselines (default method). With respect to those parameters, the pairs of interferograms can be optimally formed by applying a weighting on the available connections, thus not necessarily connecting all the available images with all the possible connections. Sixense Satellite has developed an algorithm to efficiently generate the interferogram network based on the Kruskal tree algorithm. The core of this code is the computation of the decorrelation matrix based on temporal and perpendicular baselines, as well as on doppler polynomial parameters. A weighting factor on these matrices is then applied. This code can also flexibly densify the interferogram network by adding more connections such as including large perpendicular baseline to increase the sensitivity of small height differences, hence a better estimation of the topographic phase. This algorithm also considers the degree of redundancy of interferograms per image, which is also useful in the multi-reference technique to maintain the optimal size of the interferogram network. The algorithm considers the connections per image to estimate the optimal combination of interferograms with a balanced contribution of temporal and perpendicular baselines, but also the contribution of each of the images in the network.

In this poster, we will show examples of InSAR results obtained with different interferogram networks generated with the algorithm explained above. The data processing will be performed with ATLAS InSAR, Sixense’s processing chain that has been developed around the core software GAMMA. Two stacks of images over London will be used: a stack of 178 TerraSAR-X images covering a 10-year period from May 2011 to April 2021, and a stack of 225 Sentinel-1 images from November 2015 to September 2021.

Subsidence measurement is inevitable for ensuring the sustainability of buildings in urban areas, especially in residential zones. Monitoring land surface deformation is easily accomplished using time series analysis of Interferometric Synthetic Aperture Radar (InSAR). Since the last decade, a wide area located in Sirjan has experienced a significant rate of subsidence due to the overexploitation of groundwater from an aquifer in Sirjan Basin. In this research, the Small Baseline Subset (SBAS) time series analysis of ENVISAT ASAR radar images is used for monitoring land surface subsidence in Sirjan Plain induced by excessive extraction of groundwater. Although the SBAS algorithm has reduced the effect of the decorrelation phase due to loss of coherency, we are not able to estimate the time series of deformation and mean velocity map in some locations over the area as a result of changes in backscattering behavior with time which is mainly happened in the densely vegetated surface. Due to the failure of SBAS time series analysis and inherent limitations of Persistent Scatterer Interferometry in estimating high-rate deformation, methods based on Artificial Intelligence (AI) can be a substitutive approach for estimating the subsidence in the decorrelated areas. In this study, we have created an Artificial Neural Networks (ANN) to address the problem of decorrelated pixels over the Sirjan Plain. Input variables of the model contain the geological and hydrogeological parameters of the aquifer system. These parameters either have been extracted from field observations including clay thickness, clay frequency, water decline, and water depth, or have been estimated from groundwater modeling including hydraulic conductivity and storage coefficient. First, the SBAS algorithm is applied on 12 descending ENVISAT ASAR images from track 206 spanning from 1 June 2004 to 28 September 2010. Those areas affected by decorrelation are filtered out from the time series analysis results. The subsidence rate in these areas is further estimated using the generated network. The network is trained by coherent pixels whose deformation rates were extracted from SBAS. Due to the complex behavior of subsidence in the study area, a single network is not able to model the subsidence over the whole area. Consequently, the study area is split into several parts each of which is modeled by a separate network. The results obtained from all networks show that the subsidence rate calculated from the trained network agrees well with those measured from SBAS time series analysis. The trained networks are further employed to simulate the subsidence rate in the incoherent pixels.

Heavy precipitation, such as snowfall, in mountainous areas or high-latitude regions during wintertime, poses a challenge for Synthetic Aperture Radar interferometry (InSAR) applications. The presence of a snow layer on the surface of the scatterers (natural or artificial) can cause temporal decorrelation and loss of coherence, making it difficult to make accurate measurements during snowy periods. This can create discontinuities in the displacement time series of measurement points, resulting in gaps of several months in the time series of persistent scatterers observed in the products of the European Ground Motion Service (EGMS). However, properly designed and installed artificial corner reflectors, act as coherent targets, enable continuous measurements at desired locations, and facilitate geodetic or deformation monitoring applications in these challenging regions.

Since 2021, Lantmäteriet, the Swedish mapping, cadastral and land registration authority, has installed various types and sizes of corner reflectors in multiple locations, with the aim of enhancing the national geodetic infrastructure of Sweden. We have installed triangular trihedral, double backflipped squared and trimmed trihedral squared types and equipped most of them with a cover made of radar-transparent polycarbonate material to protect against snow. These corner reflectors are designed for C-band Sentinel-1 SAR imaging and are co-located with permanent GNSS stations, with both the GNSS and corner reflectors installed on bedrock. Co-locating the corner reflectors with GNSS stations has the potential to contribute to the development of national and European ground motion services in future updates. Additionally, co-locating the reflectors with GNSS stations helps to transform the relative ground motions estimated with InSAR into an absolute geodetic reference frame with higher accuracy.

In this presentation, we will mainly report on our progress in designing and installing corner reflectors in Sweden. We will also compare the performance of different types and sizes of corner reflectors in different seasons including the temporal variations of the radar cross-section. Furthermore, we will analyse two trihedral triangular corner reflectors, made of aluminium plates with a one-meter inner leg size, located approximately 100 meters apart, in a test field at the Mårtsbo observatory. These reflectors have been set up in this location since September 202, and both are oriented for ascending Sentinel-1 tracks. One reflector was installed on a 1.2 m high mast and has a snow cover protector, while the other one is on the ground and without any snow cover protection. We have carried out various analyses on these two nearby reflectors, such as comparing the temporal variations of the backscattered radar intensities and the radar cross sections (RCS). Our analysis shows clear differences between the performance of these two reflectors, particularly during the snowfall periods from November 2021 to April 2022 and from November 2022 to March 2023. These results highlight again the importance of snow cover protection for corner reflectors in snowy regions and have implications for the use of reflectors in geodetic and deformation monitoring applications.

West Antarctic ice streams have thinned and accelerated over the last 50 years, significantly contributing to global sea level rise. Pine Island Glacier (PIG) is the fastest flowing and one of the top contributors to sea level rise in this area. Since 1970, PIG’s grounding line has retreated ~10km across most of its centre while its shelf has accelerated up to 75% and thinned by about 100m. Modelling and observational evidence indicate that the increased rate of ice loss has been driven by increased delivery of relatively warm Circumpolar Deep Water onto the continental shelf and the associated increase in ocean melt. While large-scale spatial patterns have been tracked over large temporal resolutions, the details of the ice shelf geometric evolution remain poorly constrained. This is especially the case at sub-kilometre scales, where elongated, channelised features carved by and directing oceanic melt have been observed over various time windows using in situ and remote sensing methods. At present, channel features have only been analysed for a single time step. Here, we make use of a full decade of observation (2011 - onwards) from CryoSat-2’s Interferometric Synthetic Aperture Radar (SARIn) mode to investigate the complex temporal and spatial evolution of channelised melt, from the channels’ birth at the grounding line to their disappearance at the calving front.

We deploy a Lagrangian methodology combing CryoSat-2 SARIn swath surface elevation data with high resolution, time-varying, velocity data taken from a combination of TerraSAR-X (2011 - 2013) and Sentinel-1 (2014 – onwards) products, to create high-resolution basal melting maps between 2011 and 2021 over PIG ice shelf. These melt maps are used to track and compare how the melt and ice geometry develop through space and time. We highlight the role of channels in modulating and directing melt across an ice shelf and investigate how these relationships develop as the channels are advected down the ice shelf, as well as investigating their impact on the ice shelf stability. These sub-kilometre scale patterns seem to be essential components in the ice-ocean interaction, highlighting the need for their effects to be incorporated into future sea level rise projections.

Groundwater overexploitation and its resulting surface subsidence have been critical issues in the North China Plain (NCP) for the last half-century. This problem, however, is being alleviated by the implementation of the South-to-North Water Diversion (SNWD) Project since 2015. Here, we monitor surface deformation and investigate aquifer physical properties in NCP by combining Interferometric Synthetic Aperture Radar (InSAR), Global Positioning System (GPS), and hydraulic head data observed during 2015-2019.

We process data from the ascending track 142 of the Sentinel-1A/1B satellites, with a total of 92 acquisitions among 5 consecutive frames during 4 years. The InSAR time series are generated using the StaMPS software package, and all of the interferograms are formed with respect to one reference image. By dividing the study area into overlapping patches, we use parallel computing algorithms and cluster job management system to reduce the computational overburden. With this method, we effectively reduce computation time and successfully obtain the InSAR time series in NCP with full resolution for the first time. The atmospheric phase screen (APS) is estimated and reduced using a combined method, in which the first-order APS is estimated using the ERA5 global atmosphere model, and the residual APS is estimated using the Common Scene Stacking method.

Geodetic observations reveal widespread and remarkable subsidence in the NCP, with an average rate of ~30 mm/yr, and ~100 mm/yr for the maximum. We successfully extract seasonal and long-term deformation components caused by different hydrogeological processes. By joint analysis of the seasonal deformation and hydraulic head changes, we estimate the storativity of 0.07~12.04*10^-3 and the thickness of clay lenses of 0.08~2.00 m for the confined aquifer system, and attribute their spatial distribution patterns to the alluvial and lacustrine sediments of the subsystem layers. Our study also reveals fulfilment of the SNWD Project in alleviating the groundwater shortage. About 57% of the NCP is found to have experienced subsidence deacceleration, mostly along the SNWD aqueduct lines, by a total of 37.0 mm on average during 2015-2019. The subsidence was reduced by 4.1 mm on average for the entire NCP, suggesting that although subsidence was still ongoing, the trend was reversed, particularly for some major cities along the routes of the SNWD Project. A distinct difference in subsidence rates is found across the borderline between the Hebei and Shandong Provinces, resulting from differences in groundwater use management. Our study demonstrates that the integration of geodetic and hydrological data can be effectively used for the assessment of groundwater circulation and to assist groundwater management and policy formulation.

Landslides are natural hazards that could lead to long-lasting risk in fatalities, infrastructure damage, and economic losses. It is critical to monitor landslide evolution, understand the mechanics of landslides, and further assess the risk of further instability during the post-failure stage. In June 2020, the ancient Aniangzhai (ANZ) landslide in Danba County, Sichuan Province, China was reactivated by following a series of complex hazard events. From that time until June 2021, emergency engineering work was undertaken to prevent further failure of the reactivated landslide. In this work, we examine the joint use of time-series Interferometric Synthetic Aperture Radar (TS-InSAR) and Optical Pixel Offset Tracking (POT) to explore deformation characteristics and spatial-temporal evolution of the reactivated ANZ landslide during the post-failure stage. The line-of-sight (LoS) surface displacements over the landslide body were derived by the TS-InSAR processing with both ascending and descending Sentinel-1 SAR datasets acquired between July 2020 and June 2021. Additionally, using 11 high-resolution optical images (3 m spatial resolution) between May 2020 and June 2021 acquired from the PlanetScope satellite, the large horizontal displacements over the ANZ slope were retrieved by the POT processing. The relationships between sun illumination differences, temporal baseline of correlation pairs and the uncertainties were deeply explored. A maximum LoS displacement rate of approximately 190 mm/year over the slope from July 2020 to June 2021 was obtained from the TS-InSAR results. The time series analysis based on InSAR results also suggested that the reactivated ANZ landslide experienced a gradual decrease in surface displacement and has transitioned into a steady deformation state. A slight acceleration between 22 May 2021 and 3 June 2021 was detected from the descending observation due to increased rainfall in May 2021. It is worth noting that the sun illumination parameter is the most significant factor to control the quality of POT results. The uncertainties in the North/South direction showed a higher degree of correlation with the sun illumination differences than in the East/West direction. The POT result revealed a significant increase of about 24 m in horizontal displacement between 24 June 2020 and 11 June 2021. Most importantly, the time series analysis of POT results also revealed that the horizontal displacements over the ANZ slope slowed down significantly until May 2021. Which is consistent with the linear trend status detected from the TS-InSAR results. The joint analysis of TS-InSAR and optical POT results demonstrated the effectiveness of preventive engineering work in slowing down the movement of the reactivated ANZ landslide.

Interferometric Synthetic Aperture Radar (InSAR) stacking analysis provides very powerful remote sensing tools to measure deformation of the Earth’s surface very effectively and accurately, over large areas.

The deformation analysis can be divided into two main categories based on surface backscatter: Persistent Scatterers (PS) and Distributed Scatterers (DS). On the one hand, PSs are objects characterized by a high signal-to-noise ratio and mainly appear as very bright and continuously stable points in time, typically man-made features. DS, on the other hand, have an average or low signal-to-noise ratio and can be exploited only if they form homogeneous groups of pixels large enough to allow statistical analysis and which can remain coherent over time even if discontinuously, typically rural areas.

Historical approaches that can measure separately DS or PS are the Small Baseline subset (SBAS) and Persistent Scatterers Interferometry (PSI) respectively. Since the last decade, research has made many advances in this domain, providing new methods capable of simultaneously extracting measurements form both PS and DS.

What we propose here is an exhaustive comparison of the original SBAS and PSI techniques according to Ferretti et al. (2001) and Berardino et al. (2002) algorithms, with two new derived processing chains, named Enhanced SBAS (E-SBAS) and Enhanced PSI (E-PSI). Both derived methods provide measurement of PS and DS backscatter displacements simultaneously, but following different processing philosophies. Each of the two techniques offers different characteristics in terms of absolute precision, ability to manage non-continuous or non-linear historical time series and coverage.

For the statistical and visual comparison, we use the software SARscape COTS, which provides the four processing chains. SARscape is an established commercial software tool developed by the sarmap team for processing remote sensing data for the generation of standard and customized products. Among the numerous tools dedicated to SAR data processing, all the tools related to differential interferometry and stacking InSAR are also implemented, providing cutting-edge algorithms to perform multi-temporal Interferometric analyzes. Specifically, in its new version 5.7, the spectrum of stacking tools is further expanded providing also E-SBAS and E-PSI. SARscape software is capable of ingesting any kind of SAR data acquired as part of national and international SAR missions and allowing us to run a fair comparison as exhaustive as possible.

The proposed approach for E-SBAS is inspired by (Lanari, 2014). The deformation products will be obtained exploiting a combination of both Small Baseline subset (SBAS) and Persistent Scatterers Interferometry (PSI) methods, in order to estimate the temporal deformation at both DS and point-like PS. The low-pass (LP) and high-pass (HP) terms are used to name the low spatial resolution and residual high spatial frequency components of signals related to both deformation and topography.

The role of the SBAS technique is twofold: on the one hand, it will provide the LP deformation time series in correspondence of DS points and the LP DEM-residual topography; on the other hand, the SBAS will estimate the residual atmospheric phase delay still affecting the interferometric data after the preliminary correction carried out by leveraging GACOS products and ionospheric propagation models.

The temporal displacement associated to PS points will be obtained applying the PSI method to interferograms previously calibrated removing the LP topography, deformation and residual atmosphere estimated by the SBAS technique. This strategy “connects” the PSI and SBAS methods ensuring consistency of deformation results obtained at point-like and DS targets and, therefore, provides better results with respect to the approach of executing the two methods independently from each other. The proposed hybrid approach is not just the simple application of the two techniques independently, indeed, the proposed approach is able to analyze both strong reflectors and distributed targets, delivering lower resolution DS results combined with higher resolution PS for even non-linear trends in an integrated continuous spatial solution.

The proposed approach for E-PSI is inspired by Ferretti, 2011 and Fornaro, 2015. The joint processing of PS and DS can be carried out independently, without the need for significant changes in the standard PS processing chain. Such approach is aimed to extend the standard PS analysis on rural areas and in this regard, two main steps are needed: first, the identification of ensamples of pixels which are similar from a statistical point of view must be performed. The Kolmogorov-Smirnov (KS) and Anderson–Darling(AD) tests are both based on the amplitude of coregistered and calibrated stack of SAR data. KS algorithm is simple and effective, but it can present poor sensitivity to deviations of the pixels under test. Indeed, AD compared to KS, puts more weight on the tails of the distributions but at the cost of a more expensive computation. Second, for all of the DS identified by statistical tests, the covariance matrix taking advantage of the ensemble of similar pixels, is estimated. SLC phases in correspondence of DS are weighted in an optimal way, either by the maximum likelihood estimator (MLE) under assumption of Gaussianity, or exploiting the largest principal component of the covariance matrix. DS exhibiting a coherence higher than a certain threshold are jointly processed with the PS for the final estimation of the displacement time series.

To assess the performance of the different processing chains, a test site is chosen and regularly monitored by Sentinel-1 data. The test site is heterogeneous, showing both urban and rural areas in order to observe the behavior of different DS types. Our evaluation is aimed at assessing both the processing times and the final quality of the results in terms of spatial coverage increase with the desired information as well as the capability of estimating different deformation temporal evolutions.

A. Ferretti, C. Prati and F. Rocca, 2001. Permanent scatterers in SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, vol. 39, no. 1, pp. 8-20, doi: 10.1109/36.898661.

P. Berardino, G. Fornaro, R. Lanari, E. Sansosti, 2002. A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms. IEEE Transactions on geoscience and remote sensing 40 (11), 2375-2383.

F. Casu, S. Elefante, P. Imperatore, I. Zinno, M. Manunta, C. De Luca, R. Lanari, 2014. SBAS-DInSAR Parallel Processing for Deformation Time-Series Computation. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 7, no. 8, pp. 3285-3296, doi: 10.1109/JSTARS.2014.2322671.

A. Ferretti, A. Fumagalli, F. Novali, C. Prati, F. Rocca and A. Rucci, 2011. A New Algorithm for Processing Interferometric Data-Stacks: SqueeSAR. IEEE Transactions on Geoscience and Remote Sensing, vol. 49, no. 9, pp. 3460-3470, doi: 10.1109/TGRS.2011.2124465.

G. Fornaro, S. Verde, D. Reale and A. Pauciullo, 2015. CAESAR: An Approach Based on Covariance Matrix Decomposition to Improve Multibaseline–Multitemporal Interferometric SAR Processing. IEEE Transactions on Geoscience and Remote Sensing, vol. 53, no. 4, pp. 2050-2065, doi: 10.1109/TGRS.2014.2352853.

Eruptions at long-inactive volcanoes are usually preceded by days to months of unrest as magma migrates gradually to shallower depths. This is built into plans by civil protection agencies for societal response. On 19th March 2022, at São Jorge, Azores Islands, after 60 years of repose, magma reached almost the surface in a vertical dyke intrusion within a few hours of the seismicity onset with no previous precursory signals. Recent eruptions at São Jorge have produced pyroclastic density currents, and the potential for an eruption to occur with little warning poses a great hazard to the population.

Comment

We captured the surface deformation due to the dyke intrusion using Sentinel-1 InSAR and GNSS and monitored the post-event dynamics closely with additional instruments but the intrusion did not continue to the surface. We established a model based on measurements of seismicity and land surface deformation that attempts to explain this volcanic unrest.

Deformation was high in the first day of activity (>5 cm of uplift) and significantly decreased afterwards. It reached other neighboring islands over a distance of at least 45 km away from São Jorge, expanding the region with approximately north-south displacements in magnitude of up to 2 cm, partly captured by both GNSS measurements and spectral diversity in burst overlap regions of Sentinel-1 data.

Although unrest continued for weeks, subsequent magma intrusion after the first day was below 4 km deep. São Jorge lies in a rift zone where extensional stress is expected to be built over time to accommodate magma at depth. We interpret the cause of the initial shallow injection to be due to the deviatoric stress there being so high that the suction due to opening was greater than the force required to reach a greater height. After relaxing the stress field at shallow levels, the next most energetically favourable location for magma injection was deeper. This implies that an eruption was unlikely during the first hours, despite reaching such shallow depth.

The unrest at Azores however did not conclude by this event, as since the end of June 2022, an increase in seismic activity started to appear in Terceira Island below the Santa Barbara volcano. Since then, seismic activity has remained persistent, sometimes with a few dozen events per day. Last eruptions related with this volcano occurred in 1761 and in 1867, being this last a submarine one.

We observed surface deformation of Terceira. The current results of Sentinel-1 InSAR processing using updated LiCSAR processing chain, GACOS atmospheric correction data and modified LiCSBAS time series approach are not conclusive at the moment but we will discuss a possibility and implications of increased uplift rate of Santa Barbara by around 4 mm/year.

The São Jorge event indicates that elastic strain accumulated from longterm periods of tectonic spreading at dormant volcanoes can be released by sudden, episodic shallow dyking events triggering the activity of deeper magmatic processes. Increased seismicity below Terceira is not considered directly connected to the São Jorge event, as the magma migrated in opposite direction. This contribution shows significance of using satellite InSAR to support observation of volcanic areas and importance of covering volcanoes considered inactive.

SAR images benefit from excellent geometric accuracy due to accurate time measurements in range and precise orbit determination in azimuth [1]. Moreover, the interferometric phase of each single pixel can be exploited to achieve differential range measurements for the reconstruction of topography and the observation of Earth surface deformation. But these measurements are influenced by the spatial and temporal variability of the atmospheric conditions, by solid Earth dynamics, and by SAR processor approximations, which may lead to spurious displacements shifts of up to several meters [1,2]. These effects become visible in various SAR applications including the retrieval of surface velocities using offset tracking or InSAR processing, which might require several post-processing steps and external information for correction.

To facilitate straightforward correction of the perturbing signals in the Sentinel-1 (S-1) SAR data, the Extended Timing Annotation Dataset (ETAD) was developed in a joint effort by ESA and DLR [3][4]. ETAD is a novel and flexible product for correcting the SAR range and azimuth time annotations in standard S-1 interferometric wide-swath and stripmap products. Generated on an image by image basis, it accounts for the most relevant perturbation effects, including tropospheric delays based on 3D ECMWF operational analysis data, ionospheric delays based on total electron content (TEC) maps inferred from GNSS, solid Earth tides calculated following geodetic conventions, and corrections of SAR processor approximations. The effects are converted to range and azimuth time corrections with an accuracy at a global level of at least 0.2 m, and are provided as 200m resolution grids matching the swath and burst structure of S-1 SAR data. First successful usage of ETAD corrections could be demonstrated in ice velocity tracking and InSAR applications [4].

The ETAD is planned to become an operational Sentinel-1 product by Spring 2023. Currently, the processing software is undergoing integration to ground segment production service. In parallel to establishing operational production, DLR and ESA are also evaluating possible future evolutions of the product, studying inter alia better tailoring for InSAR application, the inclusion of additional solid Earth effects, and possibilities of near real time provision. This evaluation is supported by the feedback of the S1 ETAD pilot study set up by ESA between January and September 2022 aimed to provide early access to ETAD products to expert users, promoting independent validation and supporting the definition of eventual improvements of the product. The SETAP Processor was hosted in the Geohazard Exploitation Platform to allow for processing by the pilot participants and the hosting was supported by ESA Network of Resources Initiative.

Our presentation will summarize the ETAD product and report on the status of operational integration. Moreover, we will give insight to the ongoing study of future product evolution.

Acknowledgement

The S1-ETAD scientific evolution study, contract No. 4000126567/19/I-BG, is financed by the Copernicus Programme of the European Union implemented by ESA.

The authors thank all the research groups that participated in the ETAD pilot study for their valuable feedback on the product when applying it in SAR applications such as offset tracking, InSAR processing, data geolocation and geocoding, and stack co-registration. List of participating institutions in alphabetical order: Caltech, DIAN srl, DLR, ENVEO, IREA-CNR, JPL, Joanneum Research , NORCE, PPO.labs, TRE ALTAMIRA, University of Jena, University of Leeds, University of Strasbourg.

Views and opinion expressed are however those of the author(s) only and the European Commission and/or ESA cannot be held responsible for any use which may be made of the information contained therein.

[1] Gisinger, C., Schubert, A., Breit, H., Garthwaite, M., Balss, U., Willberg, M., Small, D., Eineder, M., Miranda, N.: In-Depth Verification of Sentinel-1 and TerraSAR-X Geolocation Accuracy using the Australian Corner Reflector Array. IEEE Transactions on Geoscience and Remote Sensing, vol. 59, no. 2, pp. 1154-1181, 2021. doi: 10.1109/TGRS.2019.2961248

[2] Yunjun, Z., Fattahi, H., Pi, X., Rosen, P., Simons, M., Agram, P., Aoki, Y.: Range Geolocation Accuracy of C-/L-Band SAR and its Implications for Operational Stack Coregistration. IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1-19, 2022. doi: 10.1109/TGRS.2022.3168509.

[3] ESA: Sentinel-1 Extended Timing Annotation Dataset (ETAD). Data product website on Sentinel-1 webpage, accessed 2/22/2023. https://sentinel.esa.int/web/sentinel/missions/sentinel-1/data-products/etad-dataset

[4] Gisinger, C., Libert, L., Marinkovic, P., Krieger, L., Larsen, Y., Valentino, A., Breit, H., Balss, U., Suchandt, S., Nagler, T., Eineder, M., Miranda, N.: The Extended Timing Annotation Dataset for Sentinel-1 - Product Description and First Evaluation Results. IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1-22, 2022. doi: 10.1109/TGRS.2022.3194216

Flash floods in arid zones are responsbile for the transport of large volumes of sediments downstream up to >70 kms of the entrainment zones to populated areas. In the Atacama Desert in northern Chile, this happened in March 2015 and in May 2017, disrupting the lives of the inhabitants of the Atacama valleys for several months and resulted in a high death toll, large urban areas flooded, large volumes of sediment deposited in urban area, etc.

We have analysed a 2014-2023 time series of SAR amplitude and coherence in the valley floors of the Atacama Desert where we know from previous field work that the passage of flash floods has caused deposition, incission or both, permanently changing the surface of the valleys floors. It is not possible to dechipher mass gain or loss in with SAR amplitude or coherence but we can indirectly assess, based on characteristic grain-sizes, what type of sedimentary flow (and processes) was responsible for the surface change. We can do this at local scales, but thanks to amplitude and coherence time series we can jump to regional scales and assist understand this threat to the people living in valley floors of arid areas.

Thus, we have tested the utility of Synthetic Aperture Radar (SAR) C-band (Sentinel-1) backscatter intensity (amplitude onwards) and coherence to track surface changes in ephemeral valley floors of the Atacama Desert (~27ºS) and identify changes during extreme flood events. SAR amplitude, when used as an indirect measurement of grain-size on unvegetated surfaces, assists to interpretet grain-size at gentle valley floors chracteristic of arid landscapes. Then, we have calibrated the results with up to >200 grain-size stations measured in the field from which we have extracted the main statistical parameters (D50, D84, interquartile range, etc.). In this way, we can relate the shifts in amplitude and coherence to particular grain-size distributions after understanding the response of these surfaces to moisture and continuous ‘reworking’ processes (e.g., aeolian sediment transport). We have extracted from the characteristic trend of amplitude and coherence variations in the 2014-2023: (i) the characteristic ‘drying-period’ (time of maximum amplitude and coherence drop) after removing the moisture effect, (ii) extract the characteristic ‘reworking’ time (time during which the surface has been subject to reworking processes such as aeolian sediment removal, small runoff from snow melt, etc.).

We also have explored how topographic metrics (valley width, gradient, others) and the contribution of upstream area control the relative location of diverse sedimentary processes based on high-resolution topography produced by means of structure-from-motion photogrammetry techniques.

In conclusion, this work have focused on long-time series of ephemeral channels to extract the main parameters controlling amplitude and coherence change (amplitude and coherence drop, moisture increase, drying and reworking of the surface). From this, characteristic values of SAR amplitud ‘drop’ (in dB) allowed us to identify surface types, which has helped us to map at regional scales the flash floods that have impacted the region. The latter allows us to use SAR backscatter intensity maps, complemented with coherence, as a proxy to predict flow types (e.g., flow rheologies) within ephemeral drainages in arid zones such as the Atacama Desert during flash floods, and thus assist mitigation strategies and understanding the response of arid landscapes to extreme precipitation events.

To date, studies of Antarctic bedrock deformation have focused on velocities obtained from a sparse network of continues Global Navigation Satellite System (GNSS) stations. Recent studies (e.g., [1-3]) highlight that the GNSS rates indicate different subsidence and uplift patterns in either the Northern or the Southern parts of the Antarctic Peninsula region and these patterns cannot yet be explained by viscoelastic models. Accordingly, to capture deformation anomalies at small spatial scales and hence better constrain glacial isostatic adjustment (GIA) models, we take advantage of time-series analysis of interferometric SAR (InSAR) data to densify measurements between the sparse GNSS points in the area.

Determining accurate estimates of the solid Earth response to the change in surface loading and Antarctica’s current contribution to sea level is only possible when the signal due to past change is isolated. This signal is estimated using GIA models [4]. An estimate of the GIA signal can be provided by GNSS observations and remote sensor measurements. Although, our understanding of the ice-loss associated bedrock deformation in Antarctica has evolved rapidly in recent years, thanks to GNSS observations, the installed GNSS stations on Antarctic are far apart from each other often far from the glaciers losing most mass.

In this study, we apply InSAR to the Antarctic Peninsula to increase the spatial sampling of deformation measurements and further understand both spatiotemporal ice mass change and the rheology of the solid Earth in the region. We create InSAR relative line-of-sight (LOS) bedrock-displacement time series and velocities over 2015-2022 (spring and summer seasons), and construct the interferograms using the “Looking inside the Continents from Space SAR” (LiCSAR) processor [5]. We carefully examine the effect of different medium- and high-resolution Digital Elevation Models (DEMs) on the accuracy of InSAR phase measurements and remove the topographic contribution using the high-resolution DEM data. InSAR analysis of the Sentinel-1 data is performed using the Stanford Method for Persistent Scatterers (StaMPS) software [6-7] and a refinement process is applied to remove spatiotemporally unstable pixels from the images.

We make our measurements on individual rocky outcrops and apply the Vienna Mapping Function 3 (VMF3) tropospheric correction and the latest Ionospheric correction methodologies/data (i.e., the split-spectrum and the Centre for Orbit Determination in Europe (CODE)) to mitigate atmospheric artifacts. We then use GPS rates derived from nearby stations to validate our InSAR velocities.

References:

[1] Nield, G. A., Barletta, V. R., Bordoni, A., King, M. A., Whitehouse, P. L., Clarke, P. J., et al. (2014). Rapid bedrock uplift in the Antarctic Peninsula explained by viscoelastic response to recent ice unloading. Earth and Planetary Science Letters, 397, 32–41. https://doi.org/10.1016/j. epsl.2014.04.019

[2] Samrat, N. H., King, M. A., Watson, C., Hooper, A., Chen, X., Barletta, V. R., & Bordoni, A. (2020). Reduced ice mass loss and three-dimensional viscoelastic deformation in northern Antarctic Peninsula inferred from GPS. Geophysical Journal International, 222(2), 1013–1022. https:// doi.org/10.1093/gji/ggaa229

[3] Martín-Español, A., Zammit-Mangion, A., Clarke, P. J., Flament, T., Helm, V., & King, M. A. (2016). Spatial and temporal antarctic ice sheet mass trends, glacio-isostatic adjustment, and surface processes from a joint inversion of satellite altimeter, gravity, and GPS data. Journal of Geophysical Research: Earth Surface, 121(2), 182–200.

[4] Whitehouse, P. L., Bentley, M. J., Milne, G. A., King, M. A., & Thomas, I. D. (2012). A new glacial isostatic adjustment model for Antarctica: Calibrated and tested using observations of relative sea-level change and present-day uplift rates. Geophysical Journal International, 190(3), 1464–1482. https://doi.org/10.1111/j.1365-246x.2012.05557.

[5] Lazecký, M.; Spaans, K.; González, P.J.; Maghsoudi, Y.; Morishita, Y.; Albino, F.; Elliott, J.; Greenall, N.; Hatton, E.; Hooper, A.; Juncu, D.; McDougall, A.; Walters, R.J.; Watson, C.S.; Weiss, J.R.; Wright, T.J. LiCSAR: An Automatic InSAR Tool for Measuring and Monitoring Tectonic and Volcanic Activity. Remote Sens. 2020, 12, 2430. https://doi.org/10.3390/rs12152430

[6] Hooper, A. 2008. A multi‐temporal InSAR method incorporating both persistent scatterer and small baseline approaches. Geophysical Research Letters, 35.

[7] Hooper, A., Spaans, K., Bekaert, D., Cuenca, M., Arıkan, M. & Oyen, A. 2010. StaMPS/MTI Manual, Delft: Institute of Earth Observation and Space Systems. Delft University of Technology, http://radar. tudelft. nl/~ ahooper/stamps/StaMPS_ Manual_v3, 2.

The Copernicus POD (Precise Orbit Determination) Service is part of the Copernicus Processing Data Ground Segment (PDGS) of the Copernicus Sentinel-1, -2, -3 and -6 missions. A GMV-led consortium is operating the Copernicus POD (CPOD) Service since the launch of Sentinel-1A in 2014. The CPOD Service is in charge of generating precise orbital products and auxiliary data files for their use as part of the processing chains of the respective Sentinel PDGS.

Since the launches of Sentinel-1A in April 2014 and of Sentinel-1B in April 2016 the CPOD Service is providing POD products for the satellites based on the dual frequency high precision GPS data from the on-board receivers. Three different orbit products were provided for both satellites until the decommissioning of Sentinel-1B in mid 2022. Now, this is done for Sentinel-1A only and preparations for the upcoming Sentinel-1C satellite have started.

The PREORB product contains a prediction of 4 orbital revolutions to the future. It has a maximum latency of 30 minutes from the reception of GPS data, and an accuracy requirement better than 1 m in 2D for the first revolution. The near real-time (NRT) orbit product has a latency of maximum 45 minutes and an accuracy requirement of 10 cm in 2D. The non-time critical (NTC) orbit product has a latency requirement of less than 20 days and a very high accuracy requirement of 5 cm in 3D.

The orbit accuracy validation is mainly done by cross-comparing the CPOD orbits with independent orbit solutions provided by the Copernicus POD Quality Working Group. This is essential to monitor and to even improve the orbit accuracy, because for Sentinel-1 this is the only possibility to externally assess the quality of the orbits.

Since the beginning of 2023 the CPOD Service has switched to FocusPOD, a new in-house GMV developed POD software. Excellent preparations and planning guaranteed a smooth transition and the continuity of the high performance of all Sentinel POD products in terms of availability, latency and accuracy.

We present the Copernicus POD Service in terms of operations and orbital accuracy achieved for all orbital products for Sentinel-1A and -1B. Focus is led to the validation of all orbit product lines, recent improvements and the impact of the switch to FocusPOD. Brief outlook to the new Sentinel-1C satellite carrying a multi-GNSS receiver tracking GPS and Galileo is given.

We present an estimate of Antarctic Ice Sheet grounding line discharge from 1985 to present. This dataset draws on new monthly velocity mosaics derived from intensity tracking of Sentinel-1 image pairs as well as publicly-available estimates of ice velocity from 1985 to 2015. We present several discharge estimates using time-varying ice thickness and a range of publicly-available bed topographic datasets, including a new merged product where problematic bed topography has been corrected. As new velocity estimates are acquired, this product will be updated automatically each month, using an estimate of the current ice thickness based on recent thinning rates. We provide discharge estimates and their uncertainties at the pixel-scale, basin-scale and ice-sheet scale as well as tools to extract discharge time-series for any region of interest or to make retrospective corrections to the discharge estimates as new thickness and firn density data become available. As part of our goal to make an operational product, all input data, code and output will be made available and updated as new input data are available and as new features are added.

SARlab at Simon Fraser University owns and operates multiple SAR sensors including an airborne multichannel X-, L-, and C-band system [1], high-bandwidth mmWave sensors (80, 144, and 250 GHz bands) for in-lab experiments [2-4], and a multi-aperture electronically scanned V-band radar for ground- or drone-based use in the lab or the field. [5]. Each of these radars produces data in a different format. Applications research demands consistent, reliable, and fast focusing of the data produced by these sensors, while methods researchers want to continuously advance SAR focusing algorithms. To meet the needs of both of these groups, we have devised a universal modular SAR processing architecture which enables ingest, focusing, postprocessing, and export stages to be developed independently and in different programming languages. This allows new SAR/InSAR processing methods to be developed for multiple systems and easily swapped out for and compared with existing processing code.

The majority of SAR focusing packages described in the literature are restricted to using specific sensors and algorithms [6, 7]. The architecture presented here takes inspiration from open software like ESA’s SNAP [8] with its modular flowgraph-based approach and multi-sensor support. However, unlike SNAP’s focus on postprocessing of focused data, we aim to apply the ideas that made SNAP successful to the task of focusing the raw data produced by our various sensors.

In this architecture, the overall task of processing SAR data has been split into four main stages: ingest, focusing, postprocessing, and export. The ingest stage transforms sensor-specific raw data files into a standardized format to be fed to the focusing stage. The focusing stage takes this standard raw data and focuses it into a collection of SLC image rows, which are passed to the postprocessing stage. The postprocessing stage can be used for filtering or other transformations of the SLC data before the export stage generates products for viewing by analysts or to be used with other software. The architecture presented here focuses on the interfaces between each of these stages. This allows the focusing of data from different sensors by swapping out only the ingest stage or the comparison of different focusing algorithms while using the same ingest and export stages. The output of the processor is defined by the export stage, which can be customized to suit the need of the end-user of the imagery. For example, export stages can be created to generate SAR images with different geometries (range-Doppler, geographic), and different filetypes (GeoTIFF, Gamma format, SNAP compatible, etc).

The processing elements of the system interact with each other through interfaces called buffers. The buffers linking each pair of stages will be specialized in terms of what data and metadata it contains, but all provide common semantics like that of a FIFO queue. Elements are pushed into the queue by the producing stage and popped in the same order by the consuming stage. The contents of a buffer are split into three categories: data, dynamic metadata, and static metadata. These categories differ in the frequency with which they are updated. Data changes with every push (e.g., the samples in a pulse or the timestamp of a position measurement). Dynamic metadata may change as often as every push but is likely to change less frequently. Static metadata is set once at the initialization of the buffer and never changes.

The underlying implementation of the different buffers can be provided by many different data transfer methods such as in-memory queues, sockets, pipes, or files. Each implementation favors a particular computing scheme, like in-memory processing, distributed computing, and disk caching. All implementations, however, can communicate between the C, C++, and Python programming languages and the Linux, Windows, and MacOS operating systems to allow processor stages to be written in different languages and run on different computers. Future implementations could interface with other platforms such as FPGA co-processors. Hardware acceleration would enable real-time focusing for InSAR applications such as the in-flight generation of interferograms or coherent change detection (CCD) maps over top of a previously acquired reference set.

Development of ingest and processing stages to fit into this framework is ongoing. Demonstrations and results showing the processing of data from multiple sensors using this architecture will be presented. Examples processed with the system we intend to present include repeat pass InSAR from data acquired with the gantry-operated 80GHz SAR in the lab and from the SFU airborne L-band system over a rock glacier target as well as multi-frequency (X- and C-band) single-pass InSAR from a recent snow penetration experiment with an optical structure-from-motion snow surface reference.

REFERENCES:

[1] Stacey, J., Gronnemose, W., Eppler, J., & Rabus, B. (2022, July). En Route to Operational Repeat-Pass InSAR with SFU’s SAR-Optical Airborne System. In EUSAR 2022; 14th European Conference on Synthetic Aperture Radar (pp. 1-5). VDE.

[2] Pohl, N., Jaeschke, T., & Aufinger, K. (2012). An ultra-wideband 80 GHz FMCW radar system using a SiGe bipolar transceiver chip stabilized by a fractional-N PLL synthesizer. IEEE Transactions on Microwave Theory and Techniques, 60(3), 757-765.

[3] Jaeschke, T., Bredendiek, C., Küppers, S., & Pohl, N. (2014). High-precision D-band FMCW-radar sensor based on a wideband SiGe-transceiver MMIC. IEEE Transactions on Microwave Theory and Techniques, 62(12), 3582-3597.

[4] Thomas, S., Bredendiek, C., Jaeschke, T., Vogelsang, F., & Pohl, N. (2016, March). A compact, energy-efficient 240 GHz FMCW radar sensor with high modulation bandwidth. In 2016 German Microwave Conference (GeMiC) (pp. 397-400). IEEE.

[5] Fox, P., & Ojefors, E. (2022). Advanced Multi-Mode Multi-Mission Software-Defined mmWave Radar for Low Size, Weight, Power and Cost. Microwave Journal, 65(9), 18-31.

[6] Batra, A., Wiemeler, M., Kreul, T., Goehringer, D., & Kaiser, T. (2019). SAR Signal Processing Architecture and Effects of Motion Errors for mmWave and THz Frequencies. 2019 Second International Workshop on Mobile Terahertz Systems (IWMTS), 1–6.

[7] Hersey, R. K., & Culpepper, E. (2016). Radar processing architecture for simultaneous SAR, GMTI, ATR, and tracking. 2016 IEEE Radar Conference (RadarConf), 1–5.

[8] Zuhlke, M., Fomferra, N., Brockmann, C., Peters, M., Veci, L., Malik, J., & Regner, P. (2015). SNAP (Sentinel Application Platform) and the ESA Sentinel 3 Toolbox. 734, 21.

Sentinel-1 satellite provides free access to dual-polarization (VV and VH) images. The integration of information from both VV and VH channels in polarimetric persistent scatterer interferometry (PolPSI) techniques is expected to enhance the accuracy of ground deformation monitoring as compared to conventional PSI techniques, which utilize only the VV channel for Sentinel-1.

Persistent scatterer (PS) and distributed scatterer (DS) points play a crucial role in the PSI techniques. PSs with high phase qualities are commonly found in urban areas. As a complementary for PSs, DS points whose phase is affected by noise are commonly present in rural areas.

In this study, the identification and selection of PS and DS is based on an optimal channel created by combining the two polarimetric channels. PS candidates are selected through the amplitude dispersion (DA) criterion. To jointly utilize both PS and DS points, an adaptive speckle filtering based on the selection of homogeneous pixels (HP) was applied to the coherency matrix. Then, DS candidates were identified by using the average coherence criterion. Finally, using both PS and DS points, the Coherent Pixels Technique (CPT) was employed as the Persistent Scatterer Interferometry (PSI) processing method.

In order to analyze how the introduction of the VH channel helps improve the deformation measurement results, an experiment over Barcelona in Spain was carried out. The dataset consists of 189 dual-polarization SAR images acquired between December 2016 and January 2021. A wide variety of scenarios are present in this region, i.e., airport, harbor, and urban areas which exhibit diverse orientations of streets and buildings with respect to the acquisition geometry. Additionally, ground deformation is expected over some areas due to settlement of recent constructions and in the harbor.

Regarding PS, there are two cases in which the VH data contribute to improve the PS density. The first corresponds to scatterers that are oriented with respect to the incidence plane. The VH amplitude value of those scatterers are higher than VV channel. The second case appears more frequently than the first case and corresponds to pixels in which the VH amplitude is low but stable. Through the application of PolPSI technique, the VH channel can contribute to the selection of high-quality pixels by reducing the presence of peaks and fluctuations present in the VV channel, thus enabling the selection of pixels with good quality which would not have been identified if only VV data were processed (Luo, et al., 2022).

Instead of increasing the density, the contribution of VH channel for the identification of DS points is associated with a more accurate selection of HP. The polarimetric information enables the differentiation of pixels that belong to different targets but have similar amplitude values in the VV channel. This results in a more reliable deformation measurement, as the HP group becomes more accurate.

A comparison with experimental data and all cases (single- and dual-pol) serves to illustrate and evaluate the performance of PolPSI in this domain.

Reference:

Luo, J., Lopez-Sanchez, J. M., De Zan, F., Mallorqui, J. J., & Tomás, R. (2022). Assessment of the Contribution of Polarimetric Persistent Scatterer Interferometry on Sentinel-1 Data. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 15, 7997-8009.

Depletion of Iran’s non-renewable groundwater has contributed to land-surface deformation across the country (Motagh et al., 2008). Such depletion has been enhanced by regional droughts, but basin-scale depletions are driven mainly by extensive human groundwater extraction (Ashraf et al., 2021). Continued unsustainable groundwater management in Iran could lead to irreversible environmental impacts that threaten the country’s water, food, and thus socio-economic security.

We use Sentinel-1 Interferometric Synthetic Aperture Radar (InSAR) to analyse the locations, rates, and patterns of land surface subsidence across Iran. We use the Centre for Observing and Monitoring Earthquakes and Tectonics (COMET) “Looking into Continents from Space” (LiCSAR) automated processing system to process eight years (2014-2022) of Sentinel-1 SAR acquisitions (Lazecký et al., 2020) for InSAR analysis across Iran. The system generates short baseline networks of interferograms. We correct for atmospheric noise interferogram-wise using the GACOS system (Yu et al., 2018) and perform line-of-sight time series analysis using open-source LiCSBAS software (Morishita et al., 2020). Line-of-sight velocities are decomposed to construct vertical and horizontal (east-west) surface velocity fields across Iran (Watson et al., 2022).

Ninety-nine subsiding regions across Iran are documented on the COMET-LiCS Subsidence Portal (Payne et al., 2022; https://comet-subsidencedb.org/). The portal presents automatically processed LiCSAR Sentinel-1 interferograms and LiCSBAS velocity time series for these regions. Interactive tools allow stakeholders to make quick, critical assessments related to extents and rates of subsidence. However, regions experiencing subsidence in Iran often have high vegetation cover. Additionally, LiCSBAS uses a short baseline network strategy. For these two reasons, fading bias (e.g. De Zan et al., 2015) may be introduced to calculated InSAR velocities. Additionally, as velocity gradients are steep at the centres of subsiding regions, interferogram unwrapping errors may be incorporated into InSAR velocities. Validating portal data and velocities is therefore essential before expanding the portal to have a global focus.

We use other Earth Observation (EO) datasets to validate vertical subsidence rates in a one-hundred-kilometre squared region of south-west Tehran, Iran’s capital city. Here, preliminary InSAR results indicate that vertical surface subsidence rates exceed 100 mm/year (Foroughnia et al., 2019, Dehghani et al., 2013, Haghshena Haghighi & Motagh, 2019), some of the fastest measured subsidence rates in the world. This region has high vegetation cover and InSAR time series are calculated using small baseline interferogram networks. Comparison of InSAR velocities and validation datasets may therefore constrain the magnitudes of fading bias, unwrapping errors, and other biases. Our validation datasets include very high-resolution Pléiades optical stereo imagery; ICESat and ICESat-2 laser altimetry; and GEDI lidar data. By comparing subsidence rates calculated using all four EO datasets we aim to validate InSAR velocities whilst investigating and constraining the benefits, drawbacks, and biases associated with each technique.

Mahdi Motagh, Thomas R. Walter, Mohammad Ali Sharifi, Eric Fielding, Andreas Schenk, Jan Anderssohn, Jochen Zschau. Land subsidence in Iran caused by widespread water reservoir overexploitation. Geophysical Research Letters 35 American Geophysical Union (AGU), 2008.

Samaneh Ashraf, Ali Nazemi, Amir AghaKouchak. Anthropogenic drought dominates groundwater depletion in Iran. Scientific Reports 11, 9135 Nature, 2021.

Milan Lazecký, Karsten Spaans, Pablo J. González, Yasser Maghsoudi, Yu Morishita, Fabien Albino, John Elliott, Nicholas Greenall, Emma Hatton, Andrew Hooper, Daniel Juncu, Alistair McDougall, Richard J. Walters, C. Scott Watson, Jonathan R. Weiss, Tim J. Wright. LiCSAR: An Automatic InSAR Tool for Measuring and Monitoring Tectonic and Volcanic Activity. Remote Sensing 12, 2430 MDPI AG, 2020.

Chen Yu, Zhenhong Li, Nigel T. Penna, Paola Crippa. Generic Atmospheric Correction Model for Interferometric Synthetic Aperture Radar Observations. Journal of Geophysical Research: Solid Earth 123, 9202–9222 American Geophysical Union (AGU), 2018.

Yu Morishita, Milan Lazecky, Tim Wright, Jonathan Weiss, John Elliott, Andy Hooper. LiCSBAS: An Open-Source InSAR Time Series Analysis Package Integrated with the LiCSAR Automated Sentinel-1 InSAR Processor. Remote Sensing 12, 424 MDPI AG, 2020.

Andrew R. Watson, John R. Elliott, Richard J. Walters. Interseismic Strain Accumulation Across the Main Recent Fault SW Iran, From Sentinel-1 InSAR Observations. Journal of Geophysical Research: Solid Earth 127, American Geophysical Union (AGU), 2022.

Francesco De Zan, Mariantonietta Zonno, Paco Lopez-Dekker. Phase Inconsistencies and Multiple Scattering in SAR Interferometry. IEEE Transactions on Geoscience and Remote Sensing 53, 6608–6616 Institute of Electrical and Electronics Engineers (IEEE), 2015.

Fatema Foroughnia, Somayeh Nemati, Yasser Maghsoudi, Daniele Perissin. An iterative PS-InSAR method for the analysis of large spatio-temporal baseline data stacks for land subsidence estimation. International Journal of Applied Earth Observation and Geoinformation 74, 248-258 Elsevier, 2019.

Maryam Dehghani, Mohammad Javad Valadan Zoej, Andrew Hooper, Roman F. Hanssen, Iman Entezam, Sassan Saatchi. Hybrid conventional and Persistent Scatterer SAR interferometry for land subsidence monitoring in the Tehran Basin, Iran. ISPRS Journal of Photogrammetry and Remote Sensing 79, 157-170 Elsevier, 2013.

Mahmud Haghshenas Haghighi, Mahdi Motagh. Ground surface response to continuous compaction of aquifer system in Tehran, Iran: Results from a long-term multi-sensor InSAR analysis. Remote Sensing of Environment, 221. 534-550 Elsevier, 2019.

Active layer freezing and thawing induces subsidence and heave of the ground surface. Permafrost thawing, active layer thickening and ice melting induce long-term subsidence trends. On slopes, the additional effect of gravity leads to gradual creep or abrupt slides/falls of rock/soil masses. Documenting ground movement is therefore important to assess infrastructure stability and slope hazards in/around Arctic settlements. The movement patterns also indirectly document the ground thermal dynamics, valuable for the environmental monitoring of polar regions.

Since 2015, the Sentinel-1 satellites have provided unprecedented capability for large-scale monitoring of ground movement using InSAR technology. In mainland Norway, the InSAR Norway Ground Motion Service (GMS) provides openly available displacement time series over the whole country. In Svalbard, recent research has shown the value of InSAR to map the distribution, magnitude and timing of subsidence/heave patterns and document the kinematics of permafrost landforms on mountain slopes. However, an InSAR-based GMS for the archipelago has not yet been implemented.

The UNIS PermaMeteoCommunity project develops a response system for permafrost hazards in Longyearbyen. Real-time observations of meteorology and permafrost variables combined with modelling will provide a preparedness tool for the Longyearbyen community. To complement in-situ measurements, InSAR data is processed to map ground movement in/around Longyearbyen. Based on 2015–2022 Sentinel-1 images, Small Baseline Subset (SBAS) time series are generated for each snow-free season. The results are visualized in an interactive WebGIS based on the NORCE Geo Viz technology used for InSAR Norway. It allows for identifying moving areas, plotting time series and comparing the displacements with other datasets. With a further integration into the operational response system, the results will contribute to better understand the relations between environmental variables and hazardous processes.

In parallel, the newly funded Fram Centre PermaRICH project “Advanced Mapping and Monitoring for Assessing Permafrost Thawing Risks for Modern Infrastructure and Cultural Heritage in Svalbard” focuses on taking advantage of both InSAR-based and in-situ geodetic measurements to parametrize models of foundation stability and assess the structural performance of selected objects both in Longyearbyen and Ny-Ålesund. By integrating geohazard mapping, movement monitoring and geotechnical modelling, the objective is to estimate the risk of infrastructure destabilization and suggest adaptation measures to key stakeholders.

The InSAR components of the PermaMeteoCommunity and the Fram Center PermaRICH projects follow the common objective to go further with the operational use of InSAR technology for assessing the impact of thawing permafrost in built Arctic environments. The results from both projects will also contribute to the development of pilot products for the implementation of an InSAR Svalbard GMS. The long-term objective is to deliver InSAR ground movement maps and time series over the entire archipelago in an open access web platform dedicated to the Svalbard community.

Repeat-pass interferometry is an efficient technique for measuring surface displacements during volcanic unrest. However, in tropical environment, tropospheric delays largely contribute to the phase changes in interferograms and its contribution can mask ground deformation signals of small amplitude that can be related to deep magma replenishment or small pressurization. These artefacts may also alter larger and more localized signals induced by shallow sources. In the past years, global weather models (ERA-Interim/ERA5, NARR, HRES ECMWF) have been currently used to systematically correct tropospheric artefacts on interferograms processed on tectonic and volcanic areas with different level of performance depending on the context. Due to their coarse spatial and temporal resolution, global weather models are efficient for correcting long wavelength signals (>10 km) persistent over few hours. Therefore, their correction of short wavelength signals commonly observed on the volcanic edifices is much less efficient. In addition, the strategy of systematic corrections has limitation as it leads to cases in which the corrected interferogram contain more noise than the initial one if the weather model is inaccurate.

A solution to improve atmospheric corrections over volcanoes is to integrate additional information from local ground stations, and especially the Zenithal Tropospheric Delays (ZTD) derived from GNSS measurements. We test the method on two active volcanoes: Piton de la Fournaise (PdF) and Merapi. Our first objective is to carry out a statistical analysis to compare the performance of global weather models on a set of Sentinel-1 interferograms processed over a 1-year period in the two test sites. Overall, ERA-5 provides better performance than GACOS; however, a reduction of the atmospheric noise is observed only for less than 50% of the total interferograms. The second objective is to propose a processing pipeline to take into account the local information from GNSS using two end-member cases: dense network (PdF) and coarse network (Merapi). With ~40 stations in PdF, tropospheric delay maps are produced routinely without any external information. In this case, the GNSS-based corrections induce a reduction of the atmospheric noise for 70-80% of the total interferograms and clearly outperform the performance of weather-based corrections. It has implication for ground deformation monitoring because atmospheric-free interferograms could be obtained only hours after SAR data is acquired as the corrections do not rely anymore on the delivery of the weather models. In case the network is not dense enough to produce tropospheric delay maps such as in Merapi, the information from GNSS helps to identify the epochs when tropospheric delay maps deduced from weather models are inaccurate. This will support a strategy in which tropospheric corrections can be applied only for selected epochs.

SAR tomography is a remote sensing technique that enables the reconstruction of the three-dimensional (3D) elevation of a scene using data acquired from multiple SAR images with different view angles. The reconstruction process involves solving an underdetermined inverse problem that requires the use of advanced algorithms and careful selection of acquisition parameters. The performance assessment of the reconstruction models required the use of simulated data to evaluate the robustness of the model with respect to different acquisition parameters. Different simulations were performed for this matter, like a simulation of one resolution cell with two scatterers at different elevations, number of acquisitions, and SNR for a continued representation like Capon reconstruction model [1] and SVD Weiner [3]. Other works mainly focused on sparse reconstruction using CS reconstruction with l1-l2 norm minimization [2][4], uses a simulation of different elevation profiles of one resolution cell with two scatterers at different elevations by modifying the measurements number and SNR level, and other simulation that covers multiple elevations using a simulated range profile line with different separation between ground and building walls. The most robust evaluation takes all previous parameters into consideration, followed by an evaluation with respect to different amplitude ratio of two scatters, the difference in phase and position using the probability of detection curve to evaluate the performance of the SL1MMER for different SNR [5], CS-GLRT in [7]. Nonetheless, these simulations don’t take the geometry of the target and other acquisition parameters into consideration. Since most of the data used for SAR tomographic reconstruction in urban areas are acquired from a high-resolution X-band SAR sensor, due to its weak penetration that helps recover the geometry of the target. In this paper, we present simulated urban scenes to assess the performance and robustness of different SAR tomographic reconstruction methods. In this simulation, we took into account the key acquisition parameters of high-resolution X-band sensors to examine the robustness of the reconstruction models with respect to different parameters such as SNR, number of acquisitions M, baseline distribution, range/azimuth resolutions, height/shape of the building, intensity/phase/geometry (slope) of each reflectance, and other parameters used for this simulation taking into account the SAR geometry distortions. We assess this simulation by presenting different simulated interferograms with different perpendicular baselines for different building shapes and elevations, followed by a reconstruction using different conventional reconstructions models such as SVD-Weiner [3], MUSIC [6] for a different range profile of the simulated scene for different SNR. Nevertheless, due to the high-resolution acquisitions of the X-band SAR sensors, the l1 l2 norm minimization, and SL1MMER [5] sparse reconstructions are more suitable to assist the simulated data. An assessment of the simulated data using these sparse reconstructions is presented followed by an evaluation of the performance of these sparse reconstructions with respect to multiples parameters.

Landslides are an important hazard worldwide in particular in mountainous environment. Monitoring the evolution of the slope motion is hence crucial to detect zones at risk and further understand and control their evolution. Monitoring landslides may be done via the installation of in-situ sensors requiring efforts to maintain the instruments in difficult field conditions. Remote sensing offers the advantage to monitor the Earth at a regular frequency by remote satellite. Among the many processing strategies to monitor landslides using satellite data, InSAR has drastically evolved in the past 30 years and became a widely used technique to monitor ground deformation. Numerous processing chains are now available and there are many examples of its interest for landslide application. However, landslides remain in most cases challenging to monitor with this technique and it is not always easy to understand pros and limitations of the different processing chains available.

In this work we propose to analyze and compare the output products of four different advanced InSAR processing chains: a) SNAPPING based on the Permanent Scatterer Interferometry (PSI) approach (Foumelis et al, 2022), b) P-SBAS based on Small-Subset Baseline Analysis (SBAS) approach (Casu et al, 2014), c) SqueeSAR based on PS and DS interferometry (Ferretti et al, 2011) and d) the product of the Copernicus European Ground Motion Service (EGMS, Level 2B). We selected three test areas with known landslides in different environnments: Villerville (France), Canton de Vaud (Switzerland) and Tavernola (Italy). The SNAPPING and P-SBAS processing chains are accessible through the Geohazard Exploitation Platform (GEP) and the results were obtained with default parameterization of these services. The SqueeSAR and the EGMS products were processed independently.

We use different metrics to estimate the similarity of the ground motion time series in space and in time as well as the coverage and the information density of each products. We also analyze the georeferencing of the results by comparing the location of measurement points with man-made structures and known reference points. Finally, we also determine the sensitivity of each technique to monitor landslides by inter-comparing the coverage of measurement points in specific landslide targets. The results of this inter-comparison shows that InSAR is a mature technique and that the different products are in general in agreement over large region although their coverage and density may differ significantly. However, significant discrepancies exist in the estimation of the velocity and displacement time series in the studied landslides and this will be discussed.

The Observational Products for End-Users from Remote Sensing Analysis (OPERA) project at the Jet Propulsion Laboratory (JPL) will provide a near-global land-surface Radiometric Terrain Corrected product derived from Copernicus Sentinel-1 (RTC-S1) synthetic aperture radar (SAR) data [1]. Each OPERA RTC-S1 product will provide terrain-corrected burst-based Sentinel-1 (S1) backscatter projected over a constant Universal Transverse Mercator (UTM) grid with a geographic scope that includes all land masses excluding Antarctica and temporal sampling coincident with the availability of Sentinel-1 single-look complex (SLC) data. The OPERA RTC-S1 product is processed with the open-source OPERA RTC-S1 workflow and the InSAR Scientific Computing Environment (ISCE3) framework [2] using the same algorithms that have been developed for the upcoming NASA-ISRO Synthetic Aperture Radar (NISAR) mission [3, 4].

RTC-S1 images provide frequent all-weather day-and-night observations that can be used in numerous applications, including detection of deforestation and wildfires, agriculture, glaciology, dynamic surface water extent estimation, and many others. The RTC-S1 product will also be used within the OPERA project to map the surface water providing the Dynamic Surface Water Extent (DSWx) from Sentinel-1 DSWx-S1 product with the same near-global scope of the RTC-S1 product.

The RTC-S1 imagery will be provided as multiple single-band cloud-optimized GeoTIFFs (COGs) with metadata packaged into a single Hierarchical Data Format 5 (HDF5) file following Climate and Forecast (CF) Conventions 1.8. Due to the S1 mission narrow orbital tube [1], secondary layers, including maps of local incidence angle, layover/shadow mask, number of looks, look vector, and radiometric terrain correction area normalization factor are considered static for the project. These static layers will also be provided as COGs for each burst ID and separately from the RTC-S1.

The RTC-S1 workflow uses the algorithms developed for generating the NISAR Geocoded Polarimetric Covariance (GCOV) product. The algorithm is based on a new area-based projection algorithm and consists of two main steps [4]: 1. radiometric terrain correction [4-8] and 2. geocoding with adaptive multilooking. The new area-based radiometric terrain correction delivers high-quality terrain normalization with a significantly shorter run time (up to 26.3 times faster) compared to other state-of-the-art algorithms [4]. The shorter run time enables the correction of radar images at full SLC resolution resulting in RTC-S1 products with better terrain correction and finer details that can be processed at a large scale [4]. Instead of using traditional multilooking with a constant-size window followed by geocoding with an interpolation algorithm (e.g, sinc interpolation), the new geocoding algorithm performs the averaging of radar samples that intersect the output geographical grid with a window that varies with the topography and observation geometry. This process is carried out at full SLC resolution and does not require interpolation, providing geocoded imagery with finer resolution and free of interpolation errors such as overfitting caused by high-contrast targets [4].

In addition to describing the RTC-S1 product and algorithm, we will present the OPERA RTC-S1 algorithm verification and product validation plan. For algorithm verification, we compare the normalization factor applied to the RTC-S1 product with those obtained from other algorithms. We also compare RTC backscatter from ascending and descending satellite track and assess the flatness of RTC-S1 backscatter with respect to the local topography. For RTC-S1 product validation, we assess absolute and relative geolocation errors, evaluate the linear regression of the RTC-S1 backscatter against the local incidence angle in forested areas, and compare the radar backscatter over foreslope and backslope areas.

The OPERA RTC-S1 product will be publicly distributed through the Alaska Satellite Facility (ASF) Distributed Active Archive Center (DAAC) free of charge, with a release date scheduled for September 2023 with forward stream production.

REFERENCES

[1] Torres, R.; Snoeij, P.; Geudtner, D.; Bibby, D.; Davidson, M.; Attema, E.; Potin, P.; Rommen, B.; Floury, N.; Brown, M.; et al. GMES 559 Sentinel-1 mission. Remote Sensing of Environment 2012, 120, 9– 24. The Sentinel Missions - New Opportunities for Science, https://doi.org/https://doi.org/10.1016/j.rse.2011.05.028.

[2] P. A. Rosen et al., "The InSAR Scientific Computing Environment 3.0: A Flexible Framework for NISAR Operational and User-Led Science Processing," IGARSS 2018 - 2018 IEEE International Geoscience and Remote Sensing Symposium, Valencia, Spain, 2018, pp. 4897-4900, doi: 10.1109/IGARSS.2018.8517504.

[3] P. A. Rosen et al., "The NASA-ISRO SAR mission - An international space partnership for science and societal benefit," 2015 IEEE Radar Conference (RadarCon), Arlington, VA, USA, 2015, pp. 1610-1613, doi: 10.1109/RADAR.2015.7131255.

[4] G. H. X. Shiroma, P. Agram, H. Fattahi, M. Lavalle, R. Burns and S. Buckley, "An Efficient Area-Based Algorithm for SAR Radiometric Terrain Correction and Map Projection," IGARSS 2020 - 2020 IEEE International Geoscience and Remote Sensing Symposium, Waikoloa, HI, USA, 2020, pp. 1897-1900, doi: 10.1109/IGARSS39084.2020.9323141.

[5] W. Peake, "Interaction of electromagnetic waves with some natural surfaces," in IRE Transactions on Antennas and Propagation, vol. 7, no. 5, pp. 324-329, December 1959, doi: 10.1109/TAP.1959.1144736.

[6] Ulander, L. M. H. “Radiometric slope correction of synthetic-aperture radar images,” IEEE Trans. Geosci. Remote Sens., vol. 34, no. 5, pp. 1115–1122, Sep. 1996.

[7] Ulaby F.T., Moore R.K., Fung A.K., “Microwave Remote Sensing: Active and Passive Vol III: From Theory to Applications”, Artech House, 1986

[8] D. Small, "Flattening Gamma: Radiometric Terrain Correction for SAR Imagery," in IEEE Transactions on Geoscience and Remote Sensing, vol. 49, no. 8, pp. 3081-3093, Aug. 2011, doi: 10.1109/TGRS.2011.2120616.

Low-land permafrost areas with ice- and water-rich active layers (the seasonally thawed layer on top of permafrost) are subject to intense vertical surface deformation processes due to phase changes between ice and liquid water at seasonal to multi-year time scales. Annually, downward movement of the land surface (subsidence) associated with seasonal thaw in summer is compensated by upward movement associated with winter freezing. The amplitude of the seasonal change in elevation can reach decimetres every year. If seasonal thaw in summer dominates in the long term over upward movement associated with frost heave in winter, an effective long-term, multi-annual subsidence of the surface is observed. The precise elevation of the Earth’s surface over these multi-annual time scales can thus be a direct measure of permafrost change.

Satellite differential SAR interferometry (DInSAR) has been successfully applied in the past to measure surface deformation over low-land permafrost and to derive remotely-sensed seasonal changes in active-layer thickness. Seasonal as well as year-to-year developments in the freeze-thaw cycle and subsequent subsidence have been identified using SAR data from various satellite missions. The DInSAR phase is routinely used to estimate surface displacement, but it is also influenced by changes in soil moisture, vegetation and snow cover. An increase in soil moisture has been found to correspond to an interferometric phase that is associated with a lowering of the surface, where the magnitude of the apparent deformation is expected to increase with the wavelength because the penetration depth gets larger. Biomass growth introduces an additional phase shift, with an apparent motion away from the satellite, and vegetation height changes of a few tens of centimetres can lead to phase disturbances of several tens of degrees and a decrease in coherence, in particular at higher frequencies. An increase in the Snow Water Equivalent (SWE) of dry snow increases the range delay, with an apparent motion away from the satellite, and only small changes in SWE may introduce significant interferometric phase delays and a rapid loss of coherence. Wet snow causes an even faster loss of coherence, and thus the interferometric phase coherence over these typically moist, vegetated and snow-covered areas is also a critical factor for successful estimation of summer surface subsidence. Maintaining interferometric coherence favours lower frequencies that assure longer temporal baselines.

We analyzed a series of satellite SAR data acquired between June and September 2018 at L-band from ALOS-2 PALSAR-2, C-band from Sentinel-1, and X-band from TerraSAR-X over the central part of the Lena River Delta. The Lena River Delta is located at the Laptev Sea coast in Northeast Siberia. With an area of about 30,000 km2 it is the largest delta in the Arctic and amongst the largest in the world. The delta comprises more than 1500 islands of various sizes, which are separated by small and large river channels. It is situated in the zone of continuous permafrost and belongs to the Arctic tundra ecozone, characterized by typical tundra vegetation, covered by sedges, grasses, dwarf shrubs and a well-developed moss layer. Typical active layer thicknesses range from 25 to 50 cm and underlying permafrost soils and sediments often are very ice-rich. Landforms typically indicating melt of abundant excess ice, such as thermokarst lakes and basins, gullies and thaw slumps, are widespread in the delta. The climate features long, extremely cold winters and short, cool summers, with mean annual temperatures of −10 °C, mean February temperatures of −30 °C and mean July temperatures of 9 °C over the last decade. Snow usually starts to accumulate in September, begins to melt in May and is then typically gone in less than a month. Snow depth can significantly vary depending on topography and wind action but mostly does not exceed a few decimetres.

In our contribution, we first discuss the effect of phase coherence for the interferometric processing of SAR data in series with nominal repeat cycles of 42 days (ALOS-2 PALSAR-2), 12 days (Sentinel-1) and 11 days (TerraSAR-X). We then present and compare summer subsidence maps derived from the different sensors. Bearing in mind that the sensitivity of the phase to deformation diminishes with decreasing radar frequency - for example, a fringe corresponds to a deformation of about 12 cm at L-band, 3 cm at C-Band and 2 cm at X-band - we nonetheless found a high spatial agreement of the summer surface subsidence maps derived at the three different frequencies, suggesting surface motion as the predominant effect over changes in soil moisture, vegetation and snow cover conditions. A comparison with in-situ data indicates a pronounced downward movement of several centimetres between June and September 2018 in both InSAR and local in-situ measurements but does not reveal a good spatial correspondence. However, such a commparison is challenging as the displacements measured in-situ can vary on a sub-meter scale within a range of several centimeters depending on the microtopography, wetness, and vegetation cover.

The potential of time series of Sentinel-1 (S1) Synthetic Aperture Radar (SAR) data for monitoring crops and their phenological stages has long been recognized. Here, we aim to analyze and interpret time series of S1 data for sunflower phenology monitoring. We observed that sunflower backscattering response differs for the ascending and descending orbits for the VV polarization and VH/VV polarization ratio due to the directional behavior of the flower head.

This study proposes a method that employs Sentinel-1 Synthetic Aperture Radar (SAR) data and a machine learning model based on metrics, generalized across both space and time. We calibrated our model in Ukraine for the year 2022 using VH, VV polarization and VH/VV polarization ratio and generalized it (both spatially and temporally) to selected sites located across five countries: Ukraine for year 2018, 2019 and 2020, Hungary, France, Russia and USA for the year 2018. We observed that for the calibrated model, classification results obtained from the descending orbit (Overall Accuracy (OA) = 98%, F1-score (F1) = 97%) outperformed those obtained from the ascending orbit: (OA= 91%, F1 = 90%) due to the directional behavior of the sunflower crop.

The generalized model for sunflower crop mapping performed with an OA > 85% for all sites, with F1 being highest (>90%) for the Ukraine and Russia sites and lowest (77%) for the USA site. Furthermore, we compared the sunflower areas obtained by classification to reference area using sampling-based approach. The correlation between the remote sensed based estimates using sampling-based approach and reference sunflower area was 0.96 whereas it was 0.92 for pixel-based approach. Also, the sampling-based approach reduced RMSE of the crop area estimates from 30 thsd to 5 thsd hectares. The classification results, predicted without field label data, indicate that our proposed space-time generalized classifier, can overcome the strong reliance on training data and address issues of cloud cover in optical imagery to map sunflowers, particularly in data-sparse Eastern Europe.

The use of SAR images has increased very quickly these last years as a large number of applications become available in many fields. The availability of free radar data such as Sentinel-1 (S1) stimulates the utilization of these techniques in different domains as agriculture, civil engineering, natural disasters monitoring and many others. In this frame, interferometry is one of the key techniques that can be useful. With a revisit time of 6 to 12 days on most parts of the Earth’s landmass, Sentinel-1 time series can be produced on long periods to follow the evolution of ground surface. Interferometry techniques have been developed, for about three decades now, and, with the acceleration of data processing, it becomes easier and faster to process interferograms on large scales and long periods.

Unfortunately, computing time series of interferograms is not easy for non-radar specialists. Several software such as SNAP or online services like ASF can be used but it appeared to us useful to develop a free software to produce automatically , in an as simple as possible way, series of interferograms on regions of interest. . Indeed, it can be necessary, for all kind of applications, to process rapidly sets of interferograms on specific regions, on a given time period, to check if interferometric data is valuable for desired applications.

For 30 years now, the Radar Processing Department of CNES (French Space Agency) Technical Directorate has been one of the forerunners in the field of radar interferometry and has developed performant and validated processing chains. Based on in-house interferometric processing tools (Diapason and Orfeo Tool Box), the software named INFERNO (INterFERometry Novel) was developed for CNES Earth Observation laboratory (EOLab) by Thales Services. The objective of EOLab is to promote new applications based on satellite imagery towards any fields concerning societal issues.

INFERNO processes Sentinel-1 IW products to generate time series of interferometric coherences and interferograms on given time and location ranges. This open-source software is developed in python and based on Orfeo Tool Box (OTB) library.

Inferno was designed to remain as simple and easy to use as possible. The necessary inputs are straightforward: a first range of dates and a Region Of Interest are defined by the user and, then, Sentinel1 radar IW images available data are scrapped out of the Scihub or PEPS S1 images catalog and suggested to the user. Through a Graphical User Interface, the user chooses among different scenarios to generate interferogram time series. Additionnal output results can also be selected such as SAR images in radar geometry, orthorectified images and interferograms, calibrated, speckle filtered outputs, phase unwrapping using snaphu, quality parameters. After choosing all computations parameters, INFERNO will automatically download and process the requested data.

The interferograms output files are provided in TIFF format, with three different channels: amplitude, phase and coherency for the selected scenarios. Each user is then free to choose his favorite visualization software, as Qgis or Arcgis, always in order to keep Inferno light and easy to use.

Based on users feedback, the current version above will be enhanced in the next future. INFERNO is available under an Apache V2.0 free to be used license, and can be downloaded on github.com/CNES/inferno. For easy installation, the software is proposed as a Docker for Linux and Windows platforms.

More than 8% of the world’s population lives within 100km of a volcano with at least one significant eruption [1]. This makes volcano monitoring and eruption forecasting an important process. Satellites periodically acquire imagery that can be used to observe the behaviour of volcanoes, but the large amount of data being captured makes it impractical for humans to manually inspect every interferogram. The existing automated frameworks of deformation detection using InSAR are modelled with supervised learning which relies heavily on labelled datasets. This means the deformation with unknown characteristics by the models could be missed, thereby requiring human inspection. To deal with this problem, here we apply unsupervised machine learning techniques to InSAR interferograms to identify anomalous behaviour in the deformation patterns of volcanoes. We investigate PaDiM [2], a model that uses a pre-trained CNN (Convolutional Neural Network) feature extractor to obtain embeddings from images which are then used to generate multivariate gaussian distribution. We also experiment with GANomaly [3], a GAN (Generative Adversarial Network) where the Generator consists of an encoder-decoder-encoder ensemble. Finally, we improve the performance of GANomaly by replacing the encoder-decoder part with a U-net. We compare those anomaly detection models on three volcanoes with recent eruptions: Taal, Agung and Fagradalsfjall, captured by the Sentinel-1 satellite. We combine synthetic interferograms with real data to generalise our training samples. For each volcano, we train the models on interferograms obtained from a period before the deformation began. Using the Area Under the ROC curve as a metric, we compare the model's performance on interferograms obtained during and after periods of deformation. We observe that unsupervised methods work well on volcanoes with big deformation signals, such as Taal, but may perform less well on volcanoes where the deformation is slow and spread over a long time. Other factors that influence performance are the amount of atmospheric noise present in the interferograms and the coherence.

1. Carneiro Freire, S., Florczyk, A., Pesaresi, M. and Sliuzas, R., An Improved Global Analysis of Population Distribution in Proximity to Active Volcanoes, 1975-2015, ISPRS INTERNATIONAL JOURNAL OF GEO-INFORMATION, ISSN 2220-9964, 8 (8), 2019, p. 341, JRC116796.

1. Defard, Thomas, et al. "Padim: a patch distribution modeling framework for anomaly detection and localization." Pattern Recognition. ICPR International Workshops and Challenges: Virtual Event, January 10–15, 2021, Proceedings, Part IV. Cham: Springer International Publishing, 2021.

1. Akcay, Samet, Amir Atapour-Abarghouei, and Toby P. Breckon. "Ganomaly: Semi-supervised anomaly detection via adversarial training." Computer Vision–ACCV 2018: 14th Asian Conference on Computer Vision, Perth, Australia, December 2–6, 2018, Revised Selected Papers, Part III 14. Springer International Publishing, 2019.

Landslides are serious geologic hazards common to many countries around the world. Landslides can result in fatalities and the destruction of infrastructure, buildings, roads, and electrical equipment. Especially rapid-moving landslides that occur suddenly and travel at high speeds for miles can pose a serious threat to life and property. Landslide inventories are essential to understand the evolution of landscapes and to ascertain landslide susceptibility and hazard, which are helpful for any further hazard and risk analysis. Although several previous researchers have mapped landslides, the present archive of historical landslide inventories lacks information on the date, type, trigger, magnitude and distribution of landslides. Precise detection of landslide occurrence time is a big challenge for landslide research. Optical and Synthetic Aperture Radar (SAR) images with multi-spectral and textural features, multi-temporal revisit rates, and large area coverage provide opportunities for history landslide detection and mapping. Landslide-prone regions are frequently obscured by cloud cover, limiting the utility of optical imagery. The capacity of SAR sensors to penetrate clouds allows the use of SAR satellite data to provide a more precise temporal characterization of the occurrence of landslides on a regional scale. The archived Copernicus Sentinel-1 satellite, which has a 6-day revisit period and covers the majority of the world's land, allows for more precise identification of landslide failure timings. The time series of interferometric coherence extracted from SAR data have strong responses to sudden landslide failures in vegetated regions, which is expressed by a sudden increase or decrease in the values of coherence. Therefore, the abrupt change in coherence time series in response to the occurrence of failure can be identified and considered as the time of failure. The abrupt change and abnormalities in the time series could be efficiently detected using machine learning and deep learning. This study aims to determine the time of failure occurrence by automatically detecting sudden changes in the coherence time series. We propose a deep neural network-based time series anomaly detection strategy to detect the time of failure occurrence using SAR coherence time series. Experiments are performed using Shaziba and Shuicheng landslides in China, Takht landslide in Iran, Jalgyz-Jangak and Kugart landslides in Kyrgyzstan, Hitardalur landslide in Iceland, and Brumadinho landslide in Brazil, and compared the performance of our proposed strategy for failure detection time with widely used unsupervised algorithms including K-means, Isolate Forest, ARIMA, STL, Autoencoders, and Breakout detection.

The Central Andes subduction zone has been the theater of numerous large megathrust earthquakes since the beginning of the 21th century, starting with the 2001 Mw8.4 Arequipa, 2007 Mw8.0 Pisco and 2014 Mw8.1 Iquique earthquakes. A deeper understanding of interseismic coupling distribution between tectonic plates and seismic cycle in this area is therefore a key issue in the frame of seismic hazard assessment. In this purpose, we rely on geodetic data acquired by the dense GNSS networks that have been deployed, and on the Sentinel-1 InSAR acquisitions processed in the frame of the FLATSIM Andes project (PI: Mohamed Chlieh), and we analyse them in the frame of a PhD project cofunded by CNES (scientific referent: Felix Perosanz) and the ERC DEPPtrigger project (PI: Anne Socquet).

In a first analysis, relying on about 50 permanent GNSS time series and about 30 survey GNSS measurements acquired in Central-South Peru between 2007 and 2022, and using a trajectory model that mimics the different phases of the cycle, we extract a coherent interseismic GNSS field at the scale of the Central Andes from Lima to Arica (12°S - 18.5°S). GNSS-derived interseismic models on a 3D slab geometry indicate that the level of locking is relatively high and concentrated between 20 and 40 km depth. Locking distributions indicate a high spatial variability of the coupling along the trench axis, with the presence of many locked patches that spatially correlate with the seismic segmentation of that subduction. Our study confirms the presence of a creeping segment where the Nazca Ridge enters in subduction; we also observe a more tenuous apparent decrease of coupling related to the Nazca Fracture Zone (NFZ). However, since the Nazca Ridge appears to behave as a strong barrier, the NFZ is relatively weak and less efficient to arrest seismic rupture propagation.

The FLATSIM Andes InSAR data, covering the 2015-2021 period, will allow to better constrain the depth of the transition between brittle and ductile rheology, as well as the amount and extension of intracontinental deformation. Moreover, it would help to estimate the extent of visco-elastic relaxation following megathrust events, like the 2001 Mw8.4 Arequipa earthquake. The increased resolution would also be a key point to overcome the lack of resolution we encounter in some areas with GNSS data. We may also include a denser monitoring of creeping areas to assess if the aseismic creep is released continuously or through bursts of slow slip, in order to better constrain the frictional behavior of those barriers, and the actual value of alpha

Post-processing of these data and their joint inversion, by principal component analysis (PCAIM) and independent component analysis (ICAIM) will allow us to finely model interseismic coupling distribution along the subduction interface. From this model, we will carry out a moment budget analysis, in order to determine the maximum magnitude upcoming earthquake and its recurrence time. Finally, a significant part of the work will be dedicated to finite element modelling of the subduction zone, in order to determine rheological laws better suited to the various geological structures. This will enable taking into account complex visco-elasto-plastic behavior associated to megathrust events, as well as linking short-term and long-term crustal deformation. Finally, it would also be a step in the direction of a general interseismic coupling model at the scale of central Andes, extending south of the Arica bend where a change in the rotation direction was suggested by Arriagada et al. (2008) and Métois et al. (2016).

The Cyprus Continuously Operating Natural Hazards Monitoring and Prevention System, abbreviated CyCLOPS, is a national strategic research infrastructure unit, with main objective the systematic study of geohazards in Cyprus and the broader EMMENA region. The project was coordinated by Cyprus University of Technology in collaboration with the German Aerospace Center (DLR), and holds the support of the critical national stakeholders, such as the Geological Survey Department and the Department of Lands and Surveys. CyCLOPS is comprised of two main components; (a) a multi-parametric network of sensors (MPN) established throughout the government-controlled areas of Cyprus and (b) an Operation Centre (OC) [1]. The MPN is comprised by a permanent and a mobile segment, which is deployed at areas of interest. The permanent segment includes six permanent sites, each of which contains a Tier-1 GNSS reference station co-located with two calibration-grade triangular trihedral corner reflectors of 1.5m inner length to account for both the ascending and descending tracks of SAR satellite missions, such as ESA’s Sentinel-1. Furthermore, the GNSS equipment is co-located with precise weather stations and tiltmeters. The mounting considerations for the permanent segment are aligned with the most stringent specifications, as outlined by UNAVCO, IGS and EPN. Therefore, besides its zero-order geodetic nature, the unit aims to become a calibration and validation (Cal/Val) infrastructure for current and future SAR satellites constellations. The mobile segment is comprised by the same grade of GNSS equipment, hosted on a specifically designed mobile configuration, which enables flexibility in the deployment of the stations, even at harsh environments, to monitor dynamic phenomena, such as landslides. Furthermore, the mobile segment includes electronic corner reflectors (ECRs), which are, again, co-located with the GNSS sensors, weather stations and tiltmeters. CyCLOPS achieved full operational capacity in June 2021. Since then, it continuously monitors the geodynamic regime of the southeastern Mediterranean area along with several active landslides occurring at the western part of the island. Consequently, the objective of this research is to deliver a brief presentation of the infrastructure, the first experience after 1.5 years of system operation, and outline results from the analysis of SAR products using our Corner Reflectors. The latter can be carried out, for instance, by means of the SAR Calibration Tool (SCT), developed by Aresys Srl, to estimate accurate geometric and radiometric calibration for Sentinel-1 products over Cyprus.

Radiometric calibration will be assessed by means of a Point-Target-Analysis (PTA) on the SLCs to estimate parameters such as peak signal power, clutter power and RCS following the procedures outlined in [2]. The now almost 2 year long dataset will be analysed in full in order to verify the temporal stability of the network and to identify, for instance, drops in accuracy due to collection of precipitation in the CRs.

The geometric or geolocation accuracy will be assessed, taking into account the effects of propagation delay of the SAR signal through the troposphere and ionosphere, and geodynamical effects which influence the previously determined, e.g. through surveying, CR position such as the coordinate reference frame and solid earth tides [3,4].

References:

[1] Danezis, C. et al. (2022). CyCLOPS: A National Integrated GNSS/InSAR Strategic Research Infrastructure for Monitoring Geohazards and Forming the Next Generation Datum of the Republic of Cyprus. In: International Association of Geodesy Symposia. Springer, Berlin, Heidelberg. https://doi.org/10.1007/1345_2022_161

[2] Adrian Schubert et al., “Corner Reflector Deployment for SAR Geometric Calibration and Performance Assessment,” Ref: UZH-FRM4SAR-TN-100, Issue 1.03, 2018-08-22, UZH-WP100-CALVAL-SETUP_v103.pdf.

[3] Balss et al., “Survey Protocol for Geometric SAR Sensor Analysis,” Ref: DLR-FRM4SAR-TN-200, Issue 1.4 2018-04-26, FRM4SAR_TN200_Site_Survey_Protocol_Definition_V1_4.pdf.

[4] C. Gisinger et al., "In-Depth Verification of Sentinel-1 and TerraSAR-X Geolocation Accuracy Using the Australian Corner Reflector Array," in IEEE Transactions on Geoscience and Remote Sensing, vol. 59, no. 2, pp. 1154-1181, Feb. 2021, doi: 10.1109/TGRS.2019.2961248.

In this study, we measure the extent and magnitude of land subsidence signals over the Carrizo-Wilcox aquifer in Texas and the southern portion of the Central Valley in California. Extended droughts in both regions have strained groundwater resources used for oil and gas production, agriculture, or by surrounding communities and has led to the increasing need for efficient groundwater management. Because hydraulic head changes associated with confined aquifer pumping and recharge can lead to centimeter-level deformation, we can use spaceborne Interferometric Synthetic Aperture Radar (InSAR) surface deformation observations to better identify widespread subsidence. InSAR has been used to study drought-prone aquifers and groundwater levels. Here the subsidence signals are associated with withdrawal of fluids from the subsurface, either from oil and gas production or confined aquifer pumping.

We processed 110 C-band Sentinel-1 SAR images from 2017-2021 over a ~100 x 200 km region near San Antonio, TX and 122 C-band Sentinel-1 SAR images over the southern Central Valley, CA. These InSAR datasets suffer from severe decorrelation artifacts due to the presence of dense vegetation. When severe decorrelation is present, phase unwrapping cannot be performed reliably given that the spatially coherent signal is corrupted. Large unwrapping errors then impact final time series solutions.

Here we use Persistent Scatterers (PS) techniques to mitigate decorrelation artifacts. We employ the cosine phase similarity algorithm to choose high-quality, PS pixels that suffer from minimal decorrelation noise. In areas with very low PS density, we interpolate phase measurements between the final set of PS pixels to restore the InSAR phase continuity in space. We select the PS-interpolated interferograms with minimal phase unwrapping errors and compute the cumulative line of sight (LOS) deformation over our study regions based on a linear deformation model.

The Texas cumulative LOS deformation map derived from the repaired interferograms shows a region over 100 km long of up to 10 cm of LOS subsidence overlaying the Eagle Ford shale, the location of ongoing, extensive hydraulic fracturing. In the Central Valley, preliminary results show a subsidence region up to 90 cm LOS. For both datasets, the InSAR measurements match the GPS data at available stations with sub-centimeter error.

Future work includes analysis with in-situ well data to further explore the deformation due to pumping and groundwater withdrawal and subsequent aquifer compaction. Subsidence mapping over the large- scale, complex aquifers will help transform our understanding of groundwater resources and their sustainable management.

The COMET LiCSAR system automatically processes Sentinel-1 data to derive InSAR products for active tectonic and volcanic regions globally [1]. LiCSAR data are used to assess tectonic velocities and strain, earthquake rupture zones, volcanic deformation, and have applications in mass movement and cryospheric research. InSAR Products are disseminated through an online portal (https://comet.nerc.ac.uk/comet-lics-portal/), which enables location-based search and download of the data, and visualisation of quick-look images. Here, we present recent developments to the web portal, the results of user feedback, and outline directions of future development.

The LiCSAR portal is the gateway to over one million open access Sentinel-1 InSAR products, processed using the LiCSAR system [1]. Products include coherence, wrapped and unwrapped interferograms, multi-looked intensity images, Generic Atmospheric Correction Online Service for InSAR (GACOS) files [2], and metadata. Data coverage is shown by frames on an interactive Leaflet map. In the last year, we developed a new query-based web tool to simplify the search for data using interactive sliders. Users can apply multiple constraints to find frames meeting the input criteria, for example frames with a given time series length, or that have recent data processed and include GACOS corrections. Additionally, users can draw an area of interest to find corresponding frames and their processing status.

Other COMET data portals use LiCSAR data in a variety of applications and are in active development. The COMET Volcano Deformation Database (https://comet.nerc.ac.uk/comet-volcano-portal/)[3,4] uses LiCSAR data to analyse volcanic deformation using LiCSBAS time series processing [5]. Users can plot, analyse, and export displacement time series for volcanoes globally. Currently, a limited subset of volcanoes with good data coverage are publicly visible; however, the full database is viewable upon registration to the portal. Machine learning aids in the identification of deformation signals [6], which will help observatories monitor and respond to volcanic unrest. Another example is the COMET Subsidence Portal (https://comet-subsidencedb.org/)[7], which uses LiCSAR data to quantify subsiding basins in Iran. Finally, the Earthquake InSAR Data Provider (EIDP) (https://comet.nerc.ac.uk/comet-lics-portal-earthquake-event/) automatically processes Sentinel-1 data in the LiCSAR system for earthquakes that meet a set of criteria and are likely to produce surface deformation [1]. The EIDP catalogue contains over 500 events, which have individual event pages displaying the processed interferograms on an interactive map. These pages also catalogue the coseismic and postseismic data processed for each event. Interferograms are automatically tweeted by @COMET_database and are provided in various data formats, including KMZs for overlaying in Google Earth. Future developments will include cross-correlation derived displacements from Sentinel-2 imagery for larger earthquakes that rupture the surface.

The LiCSAR portal is accessed over 1,300 times each month and usage is increasing through time [8]. The LiCSAR system and online dissemination tools develop in response to feedback, which can be provided using a feedback survey on the LiCSAR Portal home page. Feedback suggests that academics form the largest user base, followed by geological/geophysical surveys and public sector workers. The mostly commonly used products include unwrapped and wrapped interferograms and coherence data. Additionally, the most desired future products identified by users were displacement time series. Effectively communicating uncertainties is also an area of future development, given the often complex interpretability of InSAR products [8].

References:

1. Lazecký, M.; Spaans, K.; González, P.J.; Maghsoudi, Y.; Morishita, Y.; Albino, F.; Elliott, J.; Greenall, N.; Hatton, E.L.; Hooper, A., et al. LiCSAR: An Automatic InSAR Tool for Measuring and Monitoring Tectonic and Volcanic Activity. Remote. Sens. 2020, 12, 2430, doi:https://doi.org/10.3390/rs12152430.

2. Yu, C.; Li, Z.; Penna, N.T.; Crippa, P. Generic Atmospheric Correction Model for Interferometric Synthetic Aperture Radar Observations. Journal of Geophysical Research: Solid Earth 2018, 123, 9202-9222, doi:https://doi.org/10.1029/2017JB015305.

3. Ebmeier, S.K.; Andrews, B.J.; Araya, M.C.; Arnold, D.W.D.; Biggs, J.; Cooper, C.; Cottrell, E.; Furtney, M.; Hickey, J.; Jay, J., et al. Synthesis of global satellite observations of magmatic and volcanic deformation: implications for volcano monitoring & the lateral extent of magmatic domains. Journal of Applied Volcanology 2018, 7, 2, doi:10.1186/s13617-018-0071-3.

4. Rigby, R.; Burns, H.; Watson, C.S.; Lazecky, M.; Ebmeier, S.; Morishita, Y.; Wright, T. COMET_VolcDB: COMET Volcanic and Magmatic Deformation Portal (2021 beta release) (1.1-beta). Zenodo. https://doi.org/10.5281/zenodo.4545877. 2021, http://doi.org/10.5281/zenodo.3876265, doi:http://doi.org/10.5281/zenodo.3876265.

5. Morishita, Y.; Lazecky, M.; Wright, T.J.; Weiss, J.R.; Elliott, J.R.; Hooper, A. LiCSBAS: An Open-Source InSAR Time Series Analysis Package Integrated with the LiCSAR Automated Sentinel-1 InSAR Processor. Remote Sensing 2020, 12, 424.

6. Anantrasirichai, N.; Biggs, J.; Albino, F.; Hill, P.; Bull, D. Application of Machine Learning to Classification of Volcanic Deformation in Routinely Generated InSAR Data. Journal of Geophysical Research: Solid Earth 2018, 123, 6592-6606, doi:10.1029/2018jb015911.

7. Payne, J.; Watson, A.; Thomas, M.; Crowley, K.; Maghsoudi, Y.; Lazecky, M.; Rigby, R.; Ebmeier, S.; Elliott, J. Nation-wide characterisation of actively subsiding basins in Iran using 7 years of Sentinel-1 InSAR time series analysis, Living Planet Symposium, 2022. 2022.

8. Watson, C.S.; Elliott, J.R.; Ebmeier, S.K.; Biggs, J.; Albino, F.; Brown, S.K.; Burns, H.; Hooper, A.; Lazecky, M.; Maghsoudi, Y., et al. Strategies for improving the communication of satellite-derived InSAR ground displacements. Geosci. Commun. Discuss. 2022, 2022, 1-39, doi:10.5194/gc-2022-15.

Interferometric Synthetic Aperture Radar (InSAR) is now theoretically able to map and track in time, subcentimetric ground deformation. However, time series of phase change, often directly interpreted as deformation, are contaminated by noise and biases originating from the radar wave interaction with the ground and the atmosphere.

While spatial averaging of complex interferograms (multilooking) is often required for spatial unwrapping in SBAS-like strategies, it also induces a phase error in each interferogram. Redundant interferometric networks average out this error providing that it is Gaussian centred on zero. When this condition is not satisfied, errors due to multilooking can introduce cumulative biases in deformation estimates of several centimetres per year through conventional time series analysis. Building interferograms with long temporal baselines is known to attenuate this bias, but this is rarely feasible under temperate vegetated climates. We review existing mitigation strategies and describe the problem analytically, before suggesting corrections based on empirical laws.

As multilooking error cannot be measured directly, we analyse the distribution of closure phase, a quantity reflecting the sum of multilooking errors for three interferograms. We study the interconnected statistical effects on closure phase of coherence (and its various definitions), land cover types, seasonal variations and multilooking window size under various climates. Processed Sentinel 1 images are in Normandie (France), Ontario (Canada), Balochistan (Pakistan) and Gauteng (South Africa). We find that closure phase distribution in time may be highly non-Gaussian, especially for small baseline interferograms and larger averaging window size. Specifically, it tends to be positive, right-skewed with outliers. As expected, vegetated and agricultural lands are most affected by systematic errors which display a seasonality probably related to drops in coherence due to biomass growth.

Global catalogues of volcano deformation have previously been compiled from literature on specific volcano deformation episodes and report a variety of spatio-temporal parameters. However, due to methodological differences across the literature, the data can suffer from incompleteness or relatively large uncertainties. Sentinel-1’s acquisition policy presents an opportunity to overcome these limitations and create a new, more systematic and comparable global catalogue of volcano deformation.

Here, we explore methods of systematically extracting spatial deformation characteristics from Sentinel-1 interferograms. We initially focus on extracting source parameters for deformation (location, depth, volume change etc…) systematically using GBIS, a MATLAB-based software package for Bayesian non-linear inversion of deformation data from unwrapped interferograms. We test a variety of pre-processing options, particularly for downsampling, to be able to apply GBIS in an objective and systematic manner, rather than optimising the model on a case-by-case basis. Additionally, we calculate the Akaike Information Criterion (AIC) from the root-mean-square error (RMSE) of the residuals for multiple models on each interferogram (Mogi, Okada Dyke, etc…) to determine the best fitting model in each case.

Our approaches were first validated on synthetic interferograms with generated turbulent and stratified noise, before being applied to real data in the form of a subset of Sentiel-1 interferograms from the East African Rift (EAR). The EAR dataset covers 64 Holocene volcanoes and contains 18 deformation signals detected at 14 volcanoes. We chose to use this dataset as it contains signals at a variety of spatial scales, from <1 km² to 2,600 km², representing variable processes including pre- and co- eruptive intrusions, post-eruptive deflation, lava flow subsidence and reservoir inflation.

Future aims include integrating these approaches with existing techniques to extract further spatial and temporal characteristics of deformation episodes from Sentinel-1 interferograms. From this, we aim generate a new global volcano deformation catalogue and investigate patterns between deformation using clustering techniques. Applying clustering techniques will allow for the classification of similarly deforming volcanoes, which could be used to identify ‘analogue volcanoes’.

Using ‘analogue volcanoes’, identifying and comparing a volcanic system that shows similar characteristics to the one being investigated, is an established approach used by monitoring authorities and volcano observatories during periods of unrest. This is particularly helpful where the volcano being investigated is poorly monitored or lacks historical data (of deformation for example), and a better-monitored analogue exists. These approaches aim to improve interpretation of the likely causes and forecasting of the evolution of new unrest signals, including deformation.

Looking Into the Continents from Space with Synthetic Aperture Radar (LiCSAR) is a system built for large-scale interferometric (InSAR) processing of data from Sentinel-1 satellite system, developed within the Centre for Observation and Modelling of Earthquakes, Volcanoes and Tectonics (COMET). Utilising public data sources, and data and computing facilities at the Centre for Environmental Data Analysis (CEDA) UK, LiCSAR automatically produces geocoded wrapped and unwrapped interferograms in combinations suitable for time series processing using Small Baselines (SB)-based InSAR techniques, such as the LiCSBAS open-source tool, for large regions globally. This contribution will report on up-to-date technical solutions implemented in LiCSAR, and present selected processing results demonstrating capabilities and applications of the system for studying tectonic and volcanic deformations.

LiCSAR system is established as a set of open-source tools (primarily bash scripts and custom python3 libraries), while the core SAR/InSAR processing elements are running GAMMA software. Data management combines functionality of CEDA facilities and specific LiCSInfo database running on a MariaDB server. The processing is prioritised following an earthquake or during volcanic crises through an Earthquake InSAR Data Provider (EIDP) subsystem where data are processed partially on a High Performance Computing facility, permitting rapid generation of a co-seismic InSAR products in the first hours following a new post-seismic Sentinel-1 acquisition becoming available.

The main LiCSAR products are generated from standard Sentinel-1 Interferometric Wide Swath (IWS) data in frame units where a standard frame is a merge of 13 IWS burst units per each IWS swath, covering approx. 250x250 km. The frame InSAR products and additional generated data (backscatter intensity images, pixel offset tracking outputs, tropospheric corrections by GACOS service etc.), are distributed in a compressed GeoTIFF format at 0.001° resolution in the WGS-84 coordinate system, through the COMET LiCSAR Portal, European Plate Observing System (EPOS) and the CEDA Archive. The final products are open and freely accessible. As of March 2023, over 1,105,000 interferometric pairs have been generated by processing over 266,000 epochs from Sentinel-1 acquisitions for 2,015 frames, prioritising areas of the Alpine-Himalayan tectonic belt, the East African Rift, and global volcanoes. The dataset is increasing by approx. 4,000 epochs per month.

Reliable source parameters for earthquakes provide vital
support to the progressive development of the Comprehensive
Nuclear-Test-Ban Treaty (CTBT) verification regime, e.g., for
location calibration, and validation of Earth structure models.
Recently regional seismic networks have been used to produce
full moment tensor solutions for small seismic events, alongside
more traditional global long-period surface-wave dependant
inversions. In this work we investigate for a number of
case studies, the differences in location and depth between a
seismically derived solution and an InSAR derived solution in
areas of sparse coverage by global seismic networks. Due to
InSAR being a global satellite born method it offers a none-
network dependant solution to constrain the location and depth,
in areas where the network density required for a robust seismic
moment tensor inversion is not available.
However, InSAR derived interferograms consist of phase
change contributions from a myriad of contributing factors.
Each of these phase shifts is highly geographically dependent.
As such, an investigation into the effects each of the different
sources of phase contribution have on the detectablity thresholds
and the quality of source parameter inversion compared to
seismically derived parameters in different local conditions
was conducted. Special focus was given to atmospheric and
ionospheric conditions in the chosen regions, using GACOS
atmospheric modeling to handle this correction (Yu et al., 2018).
This work focuses on sentinel-1 C-band data, as this gives
sufficient global coverage for to investigate a variety of different
environments with a quick repeat. However, this
means that the detection threshold is correlation limited due to
local factors such as vegetation and rapid surface changes, the
impact of this was also investigated. For the inversion of InSAR
data, Geodetic Baysian inversion software was used which
provides a robust Bayesian approach to the inversion problem,
with the ability to add bespoke source models (Bagnardi &
Hooper, 2017). This was compared against the results for both
GROND and pyrocko BEAT seismic inversion tools (Heimann
et al., 2018; Vasyura-Bathke et al., 2019). The outputs are
compared both in terms of absolute values and the uncertainties
associated with the depth and location.
REFERENCES
Bagnardi, M. & Hooper, A. J., 2017. Gbis (geodetic bayesian
inversion software): Rapid inversion of insar and gnss data
to estimate surface deformation source parameters and
uncertainties, in AGU Fall Meeting Abstracts, vol. 2017,
pp. G23A–0881.
Heimann, S., Isken, M., K ̈uhn, D., Sudhaus, H., Steinberg, A.,
Daout, S., Cesca, S., Bathke, H., & Dahm, T., 2018. Grond:
A probabilistic earthquake source inversion framework.
Vasyura-Bathke, H., Dettmer, J., Steinberg, A., Heimann, S.,
Isken, M. P., Zielke, O., Mai, P. M., Sudhaus, H., & J ́onsson,
S., 2019. Beat: Bayesian earthquake analysis tool.
Yu, C., Li, Z., Penna, N., & Crippa, P., 2018. Generic
atmospheric correction online service for insar (GACOS), in
EGU General Assembly Conference Abstracts, p. 11007. UK Ministry of Defence © Crown Owned Copyright 2023/AWE

InSAR time-series analysis is an important method for the monitoring of natural and anthropogenic hazards such as earthquakes, volcanoes, landslides, and subsidence due to groundwater extraction or mining as it can provide information crucial for hazard mitigation and preparedness. Accurate estimation of a pixel's coherence, which is used as a proxy for the noise level, is particularly important in InSAR time series to select pixels with low noise characteristics for the extraction of accurate ground deformation measurements. The coherence of a pixel is estimated by calculating the spatial correlation between nearby pixels within some estimation window. When this estimation window contains different types of scatterers, however, the estimation can be biased to give an incorrect value. Instead, sibling-based time-series methods such as RapidSAR (Spaans & Hooper 2016) offer superior coherence estimation because only pixels with similar scattering characteristics (Statistically Homogenous Pixels or siblings) in the window are used in the estimation. These sibling methods require a time series of interferograms to select the ensembles and as a result, siblings may become unsuitable for coherence estimation post-selection when scattering characteristics have changed in new acquisitions.

Two main scenarios exist for when siblings may become invalid: (1) the scattering characteristics of some of the siblings of a pixel change (for example, they lose coherence), (2) the scattering characteristics of the pixel of interest itself changes. The first scenario might cause an apparent decrease in coherence even though the central pixel’s coherence is unchanged, leading to the exclusion of the pixel for part of the time series. The second scenario might mean that the coherence estimation of that pixel remains essentially the same, even though the pixel’s coherence could have decreased significantly and whose phase should no longer be interpreted. To ensure the coherence estimation is accurate, we must be certain that each set of siblings is valid for each interferogram.

Cases where the coherence might decrease because of seasonal variation or farming practices may recover, allowing the siblings to still be valid later in the time series, but for areas undergoing anthropogenic changes such as construction, the siblings are unlikely to recover. Our work focuses on determining when the siblings become unsuitable and avoiding having to re-estimate siblings for each new acquisition. We explore the use of bootstrap sampling and jackknife resampling to statistically infer the variance of the sibling ensemble and to determine the suitability of sibling ensembles. We then use random forest classifiers to predict whether the sibling ensemble of each pixel is valid. We compare and validate our models by analysing an area containing rural and urban terrain and demonstrate that our methods can be used to identify pixels which have poor sibling selections. Our methods may be particularly useful for real-time high-resolution change detection.

Proper usage of distributed scatterer (DS) can improve both the density and quality of Synthetic aperture radar interferometry (InSAR) measurements. A critical step in DS interferometry is to restore consistent phase series from SAR interferogram stacks. Most state-of-art algorithms, e.g. phase triangulation (PTA) approach, adopt an approximate likelihood function to calculate the likelihood by replacing the true coherence matrix with its estimation, i.e. the sample coherence matrix. However, this approximation has drawback that the coherence estimates are greatly biased when the coherence is low.

In this study, we give a mathematical formula to truly represent the likelihood value of consistent phase series. Unlike the one used in many phase linking methods; it does not use the sample coherence matrix for approximation in calculating the likelihood value. A DS interferometry framework using the new likelihood function for consistent phase series estimation and DS points selection is given correspondingly. We then evaluate the performance of the proposed DS interferometry approach by comparing it with the state-of-the-art approaches.

The simulation study reveals the proposed phase estimation method (TMLE) outperforms existing phase linking methods with significantly less RMSE, especially for the low coherent scenario, i.e. short-term and periodic decorrelation model. Meanwhile, the can better distinguish solutions from different coherence models than the widely used posterior coherence, showing good performance to serve as a quality measure for phase linking.

The real-world case study shows a similar finding as compared with the simulation experiments. The difference between TMLE and PTA is distributed in a wide posterior coherence range, while more obvious for low coherent pixels. The TMLE gives less noisy estimated interferogram than the conventional PTA. The results from parametric bootstrapping shows that TMLE has less RMSE than PTA in different type of scatterers. The map also show better performance than posterior coherence in distinguishing different scatterers. In particular, the water scatterer can be more easily distinguished from the soil scatterer by than the posterior coherence. The final deformation map derived from the proposed DS interferometry framework has significantly better DS density and coverage than the conventional approaches.

The contributions of this study are as follows:

1) We gave a precise function to evaluate the likelihood of consistent phase series.

2) We designed a multiple-starting-points strategy to optimize consistent phase series under the new likelihood function.

3) We designed several regularization ways on sample coherence matrix to generate a series of phase linking solutions as starting points.

4) We examined the advantage of the proposed likelihood function in selecting high quality scatterers.

UK peatlands are of great environmental importance, they are a major carbon store locking-in approximately 3.2 billion tonnes of carbon and cover 12% of UK land area (CEH, 2021).

This research is part of the Natural Environment Research Council (NERC) funded project “Towards a UK fire danger rating system: Understanding fuels, fire behaviour and impacts” (https://ukfdrs.com/). Work package 1 focuses on the use of Earth Observation techniques to assess (a) the spatial distribution of vegetation fuel-loads across the UK and (b) to develop a dynamic fuel map based on seasonal change and land cover management in the South Pennines, England. This research focuses on (b) the dynamic fuel map.

The South Pennines covers the Peak District National Park (PDNP) which is the oldest national park in the UK and extends further north to Marsden Moor. The Marsden Moor Estate owned by the National Trust in West Yorkshire, is a Site of Special Scientific Interest (SSSI), a Special Area of Conservation (SAC) and Special Protection Area (JNCC, 2021) (https://sac.jncc.gov.uk/site/UK0030280). The South Pennines blanket bog habitat is home to rare upland species such as the mountain hare and red listed Birds of Conservation Concern 4 (BoCC4) such as the skylark, curlew and lapwing (British Trust for Ornithology, 2021).

Wildfire disturbance in UK peatlands is of growing concern, for example since 2019, the National Trust reported a total of £700,000 worth of damage caused by wildfires on the Marsden Moor Estate (National Trust, 2021). Over the past three years there have been large wildfire events at Marsden Moor (26 February 2019, 22 April 2019, 23 March 2020 and 25 April 2021) with the biggest fire in April 2019 with a reported 700 hectares of peatland damaged impacting this fragile landscape (National Trust, 2021). Regular wildfire activity extends throughout the South Pennines region as recorded by Incident Recording System (IRS) data provided to the UKFDRS Project by the Home Office from 2009 - 2022.

This paper presents a SAR intensity and InSAR coherence multitemporal approach to monitor wildfire occurrence and to assess the impact of these events at the landscape scale for the South Pennines area. Fuel properties of peatland vegetation in the South Pennines vary spatially due to variation in land management activites/wildfire occurrence and also seasonally due to phenological change. We examined the dynamics of land cover types from 2017 – 2022 using a Sentinel-1A and -1B intensity time series for both VV and VH polarisations. Stuctural changes of the vegetation types is analysed using InSAR coherence. This work extends previous SAR intensity and InSAR coherence analysis reported by Millin-Chalabi (2015) using ERS-2, Envisat ASAR and ALOS PALSAR sensors for burned areas from 2003 - 2008 in the PDNP.

The latest 10m land cover map from the Centre for Ecology and Hydrology is used to implement a stratified sampling technique for extracting SAR intensity and InSAR coherence values for key land cover types e.g. bog, heather, heather grassland and acid grassland. Other environmental variable will be taken into consideration when sampling e.g. precipitaton. topography and burn severity. Areas unburnt will also be sampled to act as a control and mode of comparison to burned area perimeters supplied by Moors for the Future Partnership and the European Forest Fire Information System (EFFIS).

The outcomes of this work will be combined in the future with dynamic fuel map analysis using optical sensors led by Labenski (2023) to provide a detailed understanding of the wildfire regime and potential wildfire risk for the South Pennines region.

A-DInSAR (Advanced Differential Synthetic Aperture Radar Interferometry) is one of the most powerful and widespread remote sensing techniques for monitoring Earth’s surface. It allows detecting displacements over large areas that could be related to deformative phenomena, such as landslides, subsidence, volcanic activities, etc., with millimetric accuracy. Nowadays, there are many SAR missions producing images functional to these analyses, such as Sentinel-1, COSMO-SkyMed, SAOCOM, etc., and there are also many other SAR images in databases of SAR expired missions. Consequently, it is possible to analyse A-DInSAR dataset over ground surface obtained from different satellites with different wavelengths in post-processing. The most widely used SAR bands in displacement monitoring are the X, C and L band. It is known that a satellite’s wavelength influences the ability of detecting information from the land and, therefore, the performance of A-DInSAR techniques. Considering this, A-DInSAR results could be extremely different in the same area depending on the satellite. To investigate efficiently the land surface, the development of new tools that can simultaneously exploit results obtained from different wavelengths is necessary.

To fill this gap, NHAZCA S.r.l. developed Data Fusion, in the frame of “MUSAR” project together with “CERI” – Centro di Ricerca Previsione e Prevenzione dei Rischi Geologici research center and funded by ASI (Agenzia Spaziale Italiana). Data Fusion consists of an algorithm that allows to combine the displacement along the line of sight (LOS) of A-DInSAR data from satellites with different frequencies. The results are synthetical measurement points representing displacement along the East-West and Up-Down directions. These points are called Ground Deformation Markers (GD-Markers). The Data Fusion algorithm requires orbital parameters for each orbit, such as heading and incidence angle. Some parameters are wisely selected by an operator for the generation of GD-Markers, their extension and interpolation with respect to the original A-DInSAR results. In addition to displacement, another information provided by the algorithm is the calculated error and the number of original measurements points used for the estimation, both in total and for each satellite.

The first applications of this new tools, using as input data the A-DInSAR results obtained from COSMO-SkyMed (X-band), Sentinel-1 (C-band) and SAOCOM (L-band) data, demonstrate the efficacy in detecting and investigating deformation phenomena due to geological and hydrogeological factors, such as landslides and subsidence. Positively, Data Fusion enables the user to cover a larger area than any input dataset, to recognize with better accuracy deformation processes and their spatial distribution, to obtain dependable measurements of displacements that are compatible also with other evidence, and the displacement value is in agreement with the original A-DInSAR data. In addition, it manages to remove some measurement outliers. This result is due to the use of A-DInSAR data characterized by different wavelength, that are complementary or redundant, for the creation of GD-Markers. Indeed, they are a reliable estimation of displacement, also in places without scatterers but close to other measurement points. To make Data Fusion functional to any operator, the tool has been included in PS-Toolbox, a NHAZCA software that contains different post-processing tools.

1 Introduction

Subglacial hydrology and its effect on ice flow is an important area of study for both the Greenland and Antarctic ice sheets. Events such as subglacial lake drainages are associated with local subsidence and uplift of the ice surface,and in some cases horizontal flow changes, and can be studied with in-situ measurements or by remote sensing.

The extensive temporal and spatial coverage of the Sentinel-1 SAR constellation has led to the development of operational, routinely generated, ice velocity products covering large geographic regions (e.g., the Greenland Ice Sheet) with a temporal resolution on the order of 12 days ([1]-[3]). These products provide horizontal velocities based on offset-tracking and/or InSAR, and are mosaics based on acquisitions from multiple tracks. The high spatial resolution (50 m) and low noise level (<1 m/yr) achievable with InSAR[4] allows the study of localized and subtle dynamic variations in ice motion, but since InSAR only measures displacements in the radar line-of-sight (LoS) direction, horizontal velocities are derived by combining InSAR measurements from crossing tracks (ascending/descending) - which are not routinely acquired - or by combining them with much noisier azimuth shifts from offset-tracking. In deriving the horizontal velocity, the vertical ice motion is usually assumed to be along the local topography (surface-parallel flow assumption), and any subsidence/uplift is usually assumed to be negligible. In recent studies([5]-[7]), however, the Sentinel-1 InSAR line-of-sight measurements have been used to infer centimeter-scale vertical displacements, and study the dynamic evolution of these, linking them to transient subglacial hydrological events. This is possible by adopting some assumptions on the horizontal component ice flow, or by careful interpretation of the data. In regions where crossing tracks are available, and by assuming the horizontal flow direction to be constant, the horizontal and vertical flow can be separated, allowing for the mapping of uplift/subsidence events propagating over extended areas [6]. In [5] and [7], the very localized nature of the observed propagating LoS displacement anomalies is interpreted as being consistent with uplift and subsidence caused by propagating subglacial water.

In this study we present a prototype LoS-displacement timeseries product that is intended to be provided to interested users through the ESA Climate Change Initiative (ESA CCI) Greenland and Antarctic Ice Sheet projects and can be used to identify and further study subglacial hydrology events such as those described in [5]-[7]. The product contains time-series of geocoded LoS displacements derived from single Sentinel-1 InSAR pairs (6- or 12-day baseline), along with supporting information necessary for users without InSAR expertise to interpret the data, e.g., projecting them to vertical displacement. Since crossing tracks are not required, the spatial and temporal coverage is expected to be good, except in the presence of surface melt or precipitation, where InSAR measurements are not possible due to lack of coherence. The product is generated with the IPP (Interferometric Post Processing) processor developed at the Technical University of Denmark, and which was also used to generate the InSAR measurements in [6] and [7]. The final paper and presentation will show examples of the product from both Greenland and Antarctica.

2 References

[1] T. Nagler, et.al., 2015, “The Sentinel-1 Mission: New Opportunities for Ice Sheet Observations”, Remote Sensing, https://doi.org/10.3390/rs70709371

[2] A. Solgaard, et.al., “Greenland ice velocity maps from the PROMICE project”, Earth System Science Data, 2021, https://doi.org/10.5194/essd-13-3491-2021

[3] Joughin, I. 2021. “MEaSUREs Greenland 6 and 12 Day Ice Sheet Velocity Mosaics from SAR, Version 1”, NASA NSIDC, https://doi.org/10.5067/6JKYGMOZQFYJ

[4] J.K. Andersen, et.al., “Improved Ice Velocity Measurements with Sentinel-1 TOPS Interferometry”, Remote Sensing, 2020, https://doi.org/10.3390/rs12122014

[5] N. Neckel, et.al., “Evidence of Cascading Subglacial Water Flow at Jutulstraumen Glacier (Antarctica) Derived From Sentinel‐1 and ICESat‐2 Measurements”, Geophysical Research Letters, 2021, https://doi.org/10.1029/2021GL094472

[6] N. Maier, et.al., ”Wintertime supraglacial lake drainage cascade triggers large-scale ice flow response in Greenland”, Geophysical Research Letters, 2023, https://doi.org/10.1029/2022GL102251

[7] J.K. Andersen, et.al., ”Episodic Subglacial Drainage Outbursts Below the Northeast Greenland Ice Stream”, submitted to Geophysical Research Letters, 2023, preprint available at https://doi.org/10.22541/essoar.167839992.21036667/v1

The increased availability of frequent Synthetic Aperture Radar (SAR) data over the last decade has revealed the wide range in volcano deformation patterns that was not observable before. Signals at individual volcanoes are complex as they contain contributions from multiple deformation processes (e.g., volcanic, tectonic, hydrothermal, structural, anthropogenic etc.) and noise sources (e.g., atmospheric, orbital, soil moisture etc.). Independent component analysis (ICA) has been shown as a useful tool to identify and separate deformation patterns at volcanic systems. Here, we present the application of ICA to distinguish between magmatic and hydrothermal deformation signals at silicic caldera systems.

We apply ICA to line-of-sight (LOS) displacement maps constructed using Sentinel-1 interferograms processed through the automated COMET-LiCSAR system and LiCSBAS, an open-source InSAR timeseries analysis package. We use Corbetti Caldera, located in the southern central Main Ethiopian Rift, as our initial case study. It has been showing steady deformation since mid-2009, with an uplift rate of approximately 4.8 cm yr^-1 between 2015 to 2022. From initial ICA results, we can separate this dominant uplift signal from a continuous lower magnitude fault bound deformation pattern as well as a clear seasonal trend. This second deformation signal matches the hydrothermal signal that was observed prior to the onset of continuous uplift in 2009.

We aim to extend our analysis to other caldera systems (e.g., Campi Flegrei, Italy; Laguna del Maule, Chile; Tullu Moje, Ethiopia) with known deformation signals and hydrothermal systems, to examine the sensitivity of the ICA and its ability to separate volcanic processes. Understanding the individual contributions to volcanic deformation patterns is critical to understand the architecture of the magmatic system.

The lateral shear margins play an important role in ice-stream dynamics by controlling the motion. The study of the forces partitioned within the ice stream is significant to understand the ice stream stability. In this study, we used the Interferometry technique to identify these lateral shear zones of the Greenland Ice Stream with ERS-1 and Sentinel-1A/1B of C-band Synthetic Aperture Radar (SAR) dataset. SAR Interferometry (InSAR) is useful in many applications of cryosphere like DEM (Digital Elevation Model) generation, Mass changes from DEM differencing, Ice velocity retrieval, and grounding line identification or changes. Additionally, the InSAR coherence is also useful to identify the glacier features. In this study, we used coherence to identify the lateral marginal shear zones. The ERS-1 SAR interferometry pair is selected in 1991, October of 3-days temporal baseline. The Sentinel-1A/1B SAR interferometry pair is selected in 2020, October of 6-days temporal baseline. The same C-band and the season datasets (October) are selected to avoid the penetration and seasonal effects in the results. The three decadal marginal shear zone changes are observed through these two pairs. The ice streams are selected over the region of Northeast Greenland region. The hydrologic weakening of the shear zones due to the meltwater-induced basal sliding can increase the flow of marginal shear zones. Hence, coherence is useful to identify the lateral shear zones. However, these marginal shear zones of different regions are identified in earlier studies. In this study, we notably observed the shear zones for most of the ice streams of width approximately 1 km. Interestingly, these marginal shear zones are not observed for one of the ice streams during 1991. However, in 2021, we observed the shear zones for the same ice stream during the recent year (2020) of a width more than 1 km. The study finds the development of the shear zone for the ice stream from 1991 to 2020. The shear strain rates of the marginal zones are generally high. The development of shear zones is related to the shear strain rates, hydrology system, and surface accumulation rates. Additionally, the study of the development of shear zones helps to understand the evolutionary changes of the ice streams.

The development of Synthetic Aperture Radar Interferometry (InSAR) technology and the open-sourcing of Sentinel-1 data make it easier for wide-field landslide investigation. For the C-band SAR system, InSAR measurements are severely affected by tropospheric atmospheric disturbance and unwrapping errors in alpine valley regions. Notably, the topography-dependent stratified delay with spatial heterogeneity over wide areas cannot be accurately corrected by conventional empirical phase-elevation models or external data-based methods.
In this study, a nonparametric estimation method (NEM) is proposed to isolate the stratified tropospheric delay and phase unwrapping errors from the InSAR-derived time series based on independent component analysis (ICA). ICA is used to decompose the InSAR-derived time series into a set of sources with different spatial-temporal characteristics. By isolating the sources with locally topography-dependent characteristics and those having a step jump beyond 2π in the time series, the corrected time series can be reconstructed. Distinct advantages of NEM are that no prior information of deformation or error estimation models is required, and it’s computationally efficient.
We simulate a set of InSAR-derived time series to verify the validity of NEM, which contains linear deformation, stratified delay, unwrapping error, atmospheric turbulence, and random noise. The quantitative assessment indicates that NEM has higher precision in regions with a lower level of atmospheric turbulence, and the accuracy will not be affected by the magnitude variation of deformation velocities.
We then perform NEM on a real dataset over the reservoir region of Lianghekou hydropower station, where InSAR observations might be heavily disturbed by the stratified tropospheric delay and phase unwrapping errors due to the complex meteorological conditions and steep terrain. 63 scenes of descending orbit Sentinel-1 data over the reservoir region, acquired between June 2018 and November 2020, are processed by StaMPS-SBAS. We compare NEM with other typical methods, including the external data-based method (ERA5 and GACOS), the spatial-temporal filtering method, linear model (LM). The NEM-based result holds the smallest standard deviation (STD) of deformation velocity maps and average time series in stable regions, i.e. 5.3 mm/y and 1.6 mm/y. By investigating the results' time series, we find that NEM has the best behavior on seasonal fluctuations and step jumps correction.
The test results using both simulated and real datasets have shown that NEM can accurately correct the stratified tropospheric delay and phase unwrapping errors for wide field landslides investigation. Moreover, NEM could also be applied to other fields with different criteria of spatial-temporal characteristics, such as the classification of deformation features, decomposition of deformation patterns, etc.

The political borders of Iran encompass one of the most tectonically active regions in the world. Part of the larger Alpine-Himalayan orogenic belt, convergence between the Arabian and Eurasian plates is driving active deformation and seismicity throughout the Zagros Mountains, the Alborz, the Kopeh Dag, and the Makran subduction zone. Accurate geodetic estimates of ground-surface velocities and strain rates are critical to our understanding of both the localised seismic hazard, and the distribution and mechanics of deformation throughout the country. Previous geodetic estimates from regional GNSS observations are limited by sparse station coverage, while InSAR-derived velocity fields have focused on subregions over major crustal structures due to the computational cost of processing the data.

Here, we present ground-surface velocities and strain rates for a 2 million km² area encompassing Iran, derived from the joint inversion of InSAR-derived ground-surface velocities and GNSS data. This is made possible by the COMET-LICSAR processing system, which we use to generate 85,000 interferograms from seven years of Sentinel-1 acquisitions. We correct for tropospheric noise using the GACOS system, which combines ECMWF weather models and the 90 m SRTM digital elevation model to mitigate both the stratified and turbulent signals of tropospheric delay. We estimate average velocities using LiCSBAS, an open-source software package for performing small-baseline time-series analysis. We correct for rigid plate motions, tie the InSAR velocities into a Eurasia-fixed reference frame, and perform a decomposition to estimate East and Vertical velocities at a 500 m resolution.

Our InSAR-GNSS velocity field reveals a complex mosaic of signals, from large-scale crustal deformation to localised subsidence. We model rates of interseismic strain accumulation and locking depths along four active strike-slip faults: The Main Kopet Dag Fault, the Sharoud Fault Zone, the Doruneh Fault, and the North Tabriz Fault. We investigate groundwater subsidence (publicly accessible on the COMET Subsidence Portal), co- and post-seismic deformation, active salt diaprism, and potential sediment motion. From our InSAR-GNSS velocity fields, we derive high-resolution strain rate estimates on a country- and local scale, using both Velmap and filtering methods to suppress noise. We discuss the challenges in generating a InSAR velocity field at this scale, and the difficulties of mapping diffuse strain rates in areas with abundant non-tectonic and anthropogenic signals.

The 1000-km-long Haiyuan fault system on the northeastern edge of the Tibetan Plateau contributes to accommodating the deformation in response to the India/Asia collision. In spite of its importance, the kinematics of the fault including the geometry and along-strike slip rate have not been completely defined. In this study, we use synthetic aperture radar data acquired between 2014 and 2021 by Sentinel-1 satellites to investigate the present-day strain accumulation on the Haiyuan fault system. We produce a high-resolution velocity map for the ∼300,000 km² Haiyuan region using the Small BAseline Subset method. Our new velocity fields reveal deformation patterns dominated by the eastward motion of Tibet relative to Alaxan and localised strain accumulation along the Haiyuan, Gulang and Xiangshan-Tianjingshan faults. The western ∼300 km-long section of the Haiyuan fault, which was previously unmapped, seems to follow Tuolaishan and terminate at Halahu. We compute the along-strike slip rate using a Bayesian Markov Chain Monte Carlo inversion approach, and find that the overall strike-slip rate along the Haiyuan fault system gradually increases from the western end (1.8±0.3 mm/yr close to Halahu) to the east (6.4±0.5 mm/yr before entering Liupanshan), and further east, it decreases from 6.4±0.5 mm/yr to 1.3±0.7 mm/yr. The Haiyuan fault absorbs most of the left-lateral strike-slip motion with a rate of ∼4.2±0.4 mm/yr, and the Gulang and Xiangshan-Tianjingshan faults take up a fraction of 2.2±0.6 mm/yr. We re-map the previously identified shallow creeping zone on the Laohushan segment for a length of 45 km, slightly larger than the previous estimate of 35 km. The average shallow creep rate, 3 mm/yr between 2014–2021, is consistent with the rate before 2007 (2–3 mm/yr), implying that the shallow creep is a steady behaviour.

Landslide disaster is one of the most serious disasters to people's life and property safety and public infrastructure due to its high frequency and wide influence, especially the landslide and collapse disasters widely existed in complex mountainous areas, with strong concealment and great harmfulness, and difficult to monitor and research. The traditional measuring tools, such as GPS and leveling, whose spatial density of the observation network is low. The coverage of unmanned aerial vehicle remote sensing (UAVRS), light detection and ranging (LiDAR) and ground based-synthetic aperture radar (GB-SAR) are greatly limited to investigate and monitor large-scale ground deformation. The optical remote sensing is greatly affected by weather conditions, and cannot observe the small deformation signal. In contrast, synthetic aperture radar interferometry (InSAR) has been widely used in surface deformation monitoring due to its characteristics of high monitoring precision, high spatial resolution, high temporal repetition observation, wide coverage and small impact of climate conditions. High precision deformation monitoring is very important for the study of landslide disaster, but there are still many limitations in landslide monitoring in complex mountainous areas.

First of all, various in-situ monitoring devices still infeasible to continuously evaluate the long-term displacements of the whole mining area due to its limited spatial coverage. In our research, the Multi-temporal InSAR technology is adopted to monitor the line of sight (LOS) displacement of Fushun West Opencast Coal Mine (FWOCM) and its surrounding areas in Northeast China. Comparison with ground measurements and cross correlation analysis via cross wavelet transform with monthly precipitation data are also conducted.

Secondly, one-dimensional line-of-sight (LOS) deformation monitoring ability of D-InSAR method limited a single satellite platform data to reflect the three-dimensional deformation characteristics of landslide surface. In this paper, a surface-parallel flow model is proposed to reconstruct the landslide surface three-dimensional deformation field with two observation results from different geometric images based on the geological data and DEM slope information. Experiments were carried out on Jiaju landslide in Sichuan Province, and the effectiveness of the method and model was verified by GPS observation data.

Thirdly, most landslide investigations focus on pre-disaster deformation signal extraction or co-disaster landslide-affected area estimation but ignore the stability analysis of landslides in post-disaster stage. In this study, the evolution life cycle of the Sunkoshi landslide during different periods (pre-, co- and post-disaster stages) is characterized using various InSAR techniques with multi-source SAR data. The deformation pattern and possible driving factors in the pre-disaster stage are explored, the sliding area is determined and the collapse volume is estimated, and the post-disaster stability of the landslide is evaluated.

Finally, the Distributed Scatterers SAR Interferometry (DS-InSAR) time series analysis method adopt batch processing mode. When new observation data acquired, the entire archived data is reprocessed, completely ignoring the existing results, and difficult to realize the real-time updating of data processing. In this paper, a Recursive Sequential Estimator with Flexible Batches (RSEFB) is proposed to block the large dataset flexibly without requirements on the number of images in each subset. This method updates and processes the newly acquired SAR data in near real-time, and obtains long-time sequence results without reprocessing the entire data archived.

Synthetic aperture radar interferometry (InSAR) is one of the most common techniques for the retrieval of ground topography. It is used to generate digital elevation models (DEMs) by exploiting the phase difference between two Synthetic Aperture Radar (SAR) images, which are acquired with a small spatial separation. In particular, bistatic or single-pass InSAR data is very convenient for generating high-quality DEMs, since the two acquisitions are simultaneous and therefore unaffected by temporal changes. The TanDEM-X mission, which consist of two X-Band SAR satellites flying in a helix formation, has been very successful in generating a global DEM at 12m posting, that is widely used for a variety of scientific applications. One of the main direct applications of newly acquired InSAR DEMs is monitoring topographic changes, by performing DEM differencing.

Single-scene DEMs from the TanDEM-X mission may contain residual offsets and tilts in the order of a few meters, caused by residual phase and baseline errors. Therefore, the mutual calibration of two DEMs is a critical aspect for monitoring changes. Generally, the calibration of a single InSAR-derived DEM is performed utilizing reference measurements, which mainly consist of selected tie-points with known height and location derived from GPS measurements, ICESat footprints or other LiDAR data. This procedure is very time-consuming and expensive since it is often performed manually. References have to be timely consistent and therefore depend on the availability of such external measurement in the considered area. This endangers the success of monitoring topographic changes on most regions that are difficult to access.

To deal with these restrictions, we propose a novel technique, that elaborates on the selection of natural tie-points based on the assessment of persistent scatterer candidates from Sentinel-1 time-series. Thanks to the continuous global coverage of Sentinel-1, with a maximal global revisit time of 12 days, it is possible to overcome the lack of reference calibration tie-points. The hypothesis is that the selected points from Sentinel-1 are natural targets which under certain conditions, such as an appropriate signal-to-noise ratio and interferometric coherence to assure a high quality of the selected tie-points, can be used for mutual calibration of two TanDEM-X DEMs and for the derivation of accurate DEM changes. This approach is conceived to be independent of the possible availability of reference measurements from GPS or LiDAR and to be fully automatic without any manual intervention.

On March 19, 2021, Mount Fagradalsfjall erupted for the first time after approximately 781 years in a dormant state. The observations of the Fagradalsfjall volcano were conducted during 2021 which the eruption period lasted for 6 months until 18 September 2021. 90 synthetic aperture radar (SAR) images acquired from the Sentinel-1 satellite from January 2021 to December 2021 to generate time series between 6 days. The time-series measurement was conducted using the combination of Persistent Scatterer (PS) points and the Distributed Scattered (DS) points to retrieve the high density of measurement points in the study area. The PS points were selected using an amplitude dispersion index of 0.4 and the further PS processing was similar to the StaMPS processing. Meanwhile, the DS points were selected by Generalized Likelihood Ratio (GLR) test to identify Statistically homogeneous pixels (SHP) to the SAR data. In addition, adaptive spatial coherence and temporal coherence were estimated to increase the pixel density to determine the DS point candidates. The combination of PS and DS measurement in this study was exploiting the Improved Combined Scatterers Interferometry with Optimized Point Scatterers (ICOPS) algorithm. The ICOPS method used the machine learning algorithm and optimized hot spot analysis (OHSA) after the PS and DS points were combined. The machine learning that was used in this study was a Convolutional Neural network (CNN) to find the optimal measurement points with high reliability of displacement pattern based on their coefficient of correlation between each measurement point. The OHSA method will further identify hot spot points statistically based on the Getis-Ord Gi* statistics calculation. The result from the OHSA which clustered the data based on the z-score (standard deviation) and p-value (independent probability) will be used to determine the significance of the measurement points with their neighbors spatially. The validation was conducted by comparing the ICOPS result with the time-series process with the measurement of GPS in Reykjavik city. The result showed a good correlation in the deformation patterns. The deformation around the Fagradalsfjall volcano was suggested due to the activity of the magma reservoir beneath the earth’s surface that was formed by dike intrusion. Further analysis can be conducted by applying multi-track analysis to find the 3D deformation pattern due to the eruption.

Peat soils are known to sequester vast quantities of carbon with 644 Gigatonnes (Gt), or 20-30 % of global soil carbon, stored in peat, despite covering only 3-5 % of the land area. In Europe, peat soils cover about 530,000 km² (5 %) and hold around 42 Gt of carbon. For example, Irish peat bogs have also sequestered 2.0-2.3 Gt of carbon over the past 10,000 years, with 40 % of Irish peatland carbon stored in raised bogs covering about 3 % of the land. Their small areal extent means that raised peatland represents a rare ecosystem subject to intensive conservation efforts over the past. In parallel, links proposed recently between tropical peatland Greenhouse Gas (GHG) emissions and peat-surface displacements, as estimated remotely by Interferometry of Synthetic Aperture Radar (InSAR), could provide a basis for estimation of peatland GHG emissions on a global scale via low-cost remote sensing techniques. In addition, recent studies propose that maps and time series of apparent peatland surface motions derived from satellite-based SAR/InSAR are a proxy for ecohydrological peat parameters (i.e., groundwater level and soil moisture). However, links between SAR and InSAR estimates and peat ecohydrological parameters remain uncertain for temperate bogs located in Ireland and Britain, especially raised bogs, and until recently, there has been a lack of ground validation of these apparent surface motions at raised peatlands. In our study, we analyse SAR and InSAR products (intensity maps, interferograms, coherence maps and temporal evolutions of displacements) from Sentinel-1 C-Band data for three well-studied Irish and Welsh raised bogs: Ballynafagh bog (Co. Kildare, IE), Cors Fochno (Wales, UK) and Cors Caron (Wales, UK). From various in-situ measurements (peat surface movement, groundwater levels, soil moisture, weather conditions, etc.), we analysed the linkages between SAR/InSAR estimates and ecohydrological parameters. For Ballynafagh bog, which was affected by wildfire in June 2019, the InSAR-derived VV-polarisation coherence and displacements are not affected by vegetation changes caused by the wildfire. In contrast, the VV-polarisation SAR intensity shows an increase which can be linked to vegetation removal. This bog apparently is subsiding at the centre and rising at other parts (-9 mm.yr^-1 to +5 mm.yr^-1) during the 2017-mid-2021 period. These apparent long-term evolutions are affected by annual oscillations of displacements in correlation to the variations of water-table levels (i.e., dry/wet periods) and to the meteorological conditions (rainfall and temperature). In-situ data show that the InSAR coherence is directly related to the soil moisture within the peat resulting in an oscillation of InSAR coherence according to the temporal baselines of interferograms. In parallel, InSAR processing of ascending and descending acquisitions spanning May 2015 to September 2021 indicates that the peat surface of Cors Fochno is also subsiding at the centre and rising at the edges (-5 mm.yr^-1 to +5 mm.yr^-1) while the peat surface of Cors Caron is mostly subsiding (max. -8 mm.yr^-1). Both bogs are also affected by annual oscillations. The time-series of InSAR-derived apparent surface motions show a high similarity with the peat surface displacements measured in-situ using a novel camera-based method. The InSAR data capture the amplitude and wavelength of peat surfaces oscillations well; with Pearson’s values of 0.6 and 0.7 for Cors Fochno and Cors Caron respectively, and 72 % of the deviations are lower than 5 mm (92 % < 10 mm). The true surface motion is slightly underestimated by InSAR during drought periods (summer). Our results can be interpreted as evidence that the satellite-derived C-band radar waves penetrate through the 10-20 cm thick mossy vegetation layer and into the upper few cm of the underlying peat. InSAR displacements could be modified by soil moisture (associated to potential phase ambiguity), resulting in biased InSAR-derived displacements during the dry periods. Overall, our results confirm that InSAR can enable accurate monitoring of the surface motions of temperate raised peatlands.

In early 1990s, a European consortium led by French and Greek universities and geophysical observatories initiated an institution of long-term observation in the western Gulf of Corinth, Greece, named the Corinth Rift Laboratory (CRL, http://crlab.eu). Its principal aim is to better understand the physics of the earthquakes, their impact and the connection to other related phenomena such as tsunamis or landslides.

The Corinth Rift, is one of the narrowest and fastest extending continental regions worldwide. Its western termination was selected as the study area with the criterion of its high seismicity and strain rate. The cities of Patras and Aigio, as well as other towns were destroyed several times since the antiquity by earthquakes and, in some cases, by earthquake-induced tsunamis. The historical earthquake catalogue of the area reports five to ten events of magnitude larger than 6 per century. Episodic seismic sequencies are often. Over the past two decades, a dense array of permanent sensors was established in the CRL, gathering 80+ instruments, the majority of them being acquired in real time.

The CRL is nowadays one of the Near Fault Observatory (NFO) of the European Plate Observing System (EPOS, https://www.epos-eu.org/tcs/near-fault-observatories) and the only one with international governance.

With the development of synthetic aperture radar interferometry (InSAR) and high-resolution optical imagery space missions, remote sensing occupies an increasingly important place in the observatory. Space observations, especially those from InSAR, contain unique, dense and global information that cannot be obtained through field observations. Although low Earth orbit satellites cannot provide continuous real-time observations, the time lag can be sufficiently short for the space products to be useful for monitoring needs. The increased geophysical continuous activity and density of in-situ instruments such as GNSS and strainmeters, renders this natural laboratory site as a platform for validation/calibration/correction of InSAR and MT-InSAR products as well for benchmarking of routine ones. The community may be benefit from exploiting the available Virtual/Transnational Access (VA/TNA) services provided through EPOS/ERIC and Horizon projects like Geo-Inquire.

For the observation of the CRL observatory, the European Space Agency’s Geohazards Exploitation Platform (GEP) gathers, in a well-organized manner, products routinely made by different services, with a double benefit for the observatory: (1) computational resources and algorithms hosted and maintained by the service provider and (2) capability to elaborate solutions with different services for greater confidence and robustness.

An additional advantage is the didactic and user friendly design of the GEP that permits to disseminate it to schools.

From the science point of view, a current weakness of the GEP is the lack of visibility on the implemented algorithms, especially for the services not based on open-source packages. This issue is taken into account by the United Nations Global Geospatial Information Management (UN-GGIM) in its recommendations on coordinated geospatial information management.

Our current efforts intend to strengthen the contribution of GEP to the CRL observatory and to turn the space component stronger in this NFO. The utilization of GEP advanced InSAR services, such as P-SBAS and SNAPPING, for monitoring terrain motion over the Gulf of Corinth, as constraint of regional GNSS measurements, will be demonstrated.

Since 2016, a yearly summer school, the CRL-School, is organized in the framework of the NFO CRL for the postgraduate students and secondary education teachers. Since 2016 a School is being organized at the end of September, every, in the framework and in the research objectives of the NFO CRL for the postgraduate students and secondary educations school teachers. This experiential summer, is tailored to teach in this natural laboratory, and in the field the major components and theoretical background of the observations performed in the NFO. Space observations occupy an important role in the school, with the presence of experts from space agencies and the GEP consortium. The participants have the opportunity to analyze the space data directly in the field, in front of the in-situ instruments as well as in front of geological and other objects of interest. The CRL-School is particularly relevant to the activities of ESA’s European Space Education Resource Office (ESERO) network of currently twenty offices in the ESA member states, focusing on strengthening Science, Technology, Engineering, and Mathematics (STEM) and Space Education in primary and secondary education.

Badlands are typical landforms on clayey, bare and sparsely vegetated slopes, characterized by high rates of erosion due to water washout [1]. Erosion reduces the soil capacity to support life, leading to progressive or abrupt decrease of the total plant biomass, a simplification of the vegetation structures and a modification of the plant spatial distribution. Badland runoff can trigger flash floods and landslide movements that are difficult to predict, with potentially devastating consequences. Furthermore, high loads of sediment, salts or agricultural chemicals transferred from runoff into streams and downstream water bodies can have important ecological impacts and cause problems for human health.

The study of erosion rates and processes generally involves in situ measurements or, regarding satellite remote sensing, indices derived from satellite optical imagery. Such approaches, however, present significant uncertainties, especially if there is a need to investigate large areas, over long periods of time. More recently, coherence measured on interferometric synthetic aperture radar (InSAR) has been proposed as a tool to observe badland soil erosion phenomena with high spatial and temporal resolution [2]. In this framework, we present here some experiments on long time series of C-band Sentinel-1 SAR images, with the aim to investigate badland erosion processes through integration of geomorphic digital elevation analysis, rainfall, and satellite PSInSAR data, on different test sites in the Basilicata region, in southern Italy. In this area, two main morphologies of badlands can be distinguished: Calanchi and Biancane [3, 1]. Calanchi have a ‘knife-edge’ geometry, characterized by a network of rills, separated by ridges [4]. These asymmetrical forms are generally found at high elevations, maintaining the slopes at a steep constant angle. Biancane are dome-shaped forms, and have been interpreted by some authors as the end-product of calanchi erosion [e.g. 3]: for this reason, they are found at lower elevations and dominate at the base of the slopes. There are also forms that have intermediate morphological and physico-chemical characteristics, called hummocky [5] or mammellonari. Processes that characterize such slopes, including erosion, result also in landslide phenomena. The climate of the study area can be classified as Mediterranean, with a mean annual rainfall that varies between 530 and 750 mm. Since the beginning of the 21st century, the rainfall trend shows a general increase in both total and daily precipitation [6].

For our study, time series of Sentinel-1 SAR images acquired in the interferometric wide swath (IW) mode were collected and processed over the area, in both ascending and descending geometries. The time series are composed of more than 300 images each (acquisition window of 5 years), with a temporal resolution of 12 days in the first year, reaching 6 days from 2016 up to December 2021, thanks to the availability of the Sentinel-1B sensor (on December 23, 2021 Sentinel-1B experienced an anomaly, leaving it unable to deliver radar data). For each geometry of interest, precise, sub-pixel coregistration was performed through the ESA SNAP software tool. Interferograms were then formed between pairs of images with short temporal baselines, focusing in particular on combinations spanning up to 18 days. Stacks of coherence images spanning fixed temporal baselines were processed separately and time series composed of the “cascaded” coherences were analyzed, in correlation with corresponding time series of cumulated daily rainfall levels, collected from rain gauge stations located close to the test sites. In addition, each coherence time series was also fitted with a periodic function.

Average coherence on badland areas appears higher than on other nearby areas, either naturally vegetated (shrubs or Mediterranean scrub) or cultivated. Episodes of partial coherence loss on gullies appear temporally correlated with time series of precipitation cumulated over the time intervals between each InSAR pair. The climatic conditions at our test site make it challenging to analyze individual rainfall events and investigate their impact on spatial coherence [e.g. 7]. However, our statistical analysis indicates that cumulated rainfall between SAR acquisition separated by short intervals (6 to 18 days) has a significant correlation with abrupt decreases in short-term InSAR coherence levels. The same time series of InSAR coherences on cascaded short-baseline image pairs exhibit a different behavior on other areas with crops or spontaneous vegetation: here, the correlation with rainfall is lower, and a seasonal trend is instead statistically significant (with p-values lower than 0.1 over large extensions).

Our results strongly suggest that we can observe badland soil erosion phenomena with high spatial and temporal resolution. A critical aspect is the potential for large-scale applications. Despite the relatively small size of our test area, badlands, or bare soils subject to surface erosion in general, are widespread in many parts of the world. With the wide and increasing availability of long time series of SAR data at the global level, this opens up new avenues for investigating important processes such as soil erosion on a large scale.

References

[1] Gallicchio, S., Colacicco, R., Capolongo, D., Girone, A., Maiorano, P., Marino, M., Ciaranfi N. (2023). Geological features of the Special Nature Reserve of Montalbano Jonico Badlands (Basilicata, Southern Italy). Journal of Maps, accepted, https://doi.org/10.1080/17445647.2023.2179435.

[2] Refice, A, Partipilo, L., Bovenga, F., Lovergine, F.P., Nutricato, R., Nitti, D.O., Capolongo, D. (2022). Remotely sensed detection of badland erosion using multitemporal InSAR. IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium, Kuala Lumpur, Malaysia, 2022, pp. 5989-5992. doi: 10.1109/IGARSS46834.2022.9883555.

[3] Alexander, D.E. (1982). Difference between “calanchi” and “biancane” badlands in Italy. R. Bryan, A. Yair (Eds.), Badland Geomorphology and Piping, Geo Books, Norwich, UK (1982), pp. 71-88

[4] Piccarreta, M., Capolongo, D., Boenzi, F., Bentivenga, M. (2006). Implications of decadal changes in precipitation and land use policy to soil erosion in Basilicata, Italy. Catena, vol. 65, Issue 2, pp. 138-151. https://doi.org/10.1016/j.catena.2005.11.005.

[5] Del Prete, M., Bentivegna, M., Amato, M., Basso, F., Tacconi, P. (1997) - Badland erosion processes and their interactions with vegetation: a case study from Pisticci, Basilicata, Southern Italy. Geografia Fisica e Dinamica Quaternaria, 20(1), 147-155.

[6] Piccarreta, M., Pasini, A., Capolongo, D., Lazzari, M. (2013). Changes in daily precipitation extremes in the Mediterranean from 1951 to 2010: The Basilicata region, southern Italy. Int. J. Climatol. 2013, 33, 3229–3248. https://doi.org/10.1002/joc.3670.

[7] Cabré, A., Remy, D., Aguilar, G., Carretier, S., Riquelme, R. (2020). Mapping rainstorm erosion associated with an individual storm from InSAR coherence loss validated by field evidence for the Atacama Desert. Earth Surf. Process. Landforms, vol. 45, pp. 2091–2106. https://doi.org/10.1002/esp.4868.

Since the 21st century, the urbanization of the human living environment has accelerated, and many various bridge facilities have emerged. With the increase in operation time and daily load, some bridges showed different degrees of settlement, deformation, cracks, and bulges, which seriously affected the safety of bridges in daily use. Therefore, using a reliable technique for bridge periodic deformation monitoring is of great research importance to prevent bridge collapses that cause public casualties and property damage.

Bridge deformation monitoring by traditional contact monitoring techniques (such as GPS, level, and total station.) has the disadvantage of a long monitoring period and is susceptible to environmental influence. The InSAR technology is a non-contact monitoring means. Applying InSAR to infrastructure monitoring, such as bridges and high-rise buildings, has the advantages of round-the-clock monitoring, high accuracy, and low cost. In addition, the technology does not affect the traffic of bridges during the monitoring period, and high-resolution X-band SAR data can be applied to bridge fine deformation monitoring work with the advantages of higher monitoring point density and sensitivity.

This research takes continuous box girder and cable-stayed bridges in Shenyang, Liaoning Province, as research objects. Thirty images from March 2015 to April 2017 provided by TerraSAR-X satellite and 29 from August 2015 to June 2017 provided by COSMO-SkyMed satellite were used as data sources and processed by SBAS-InSAR technique to obtain the deformation information of the bridge in the LOS direction. The least-squares linear fitting method is applied to extract the temperature influence factors by combining the bridge's structural characteristics and material properties and constructing a bridge thermal dilation model to separate the bridge's thermal dilation and trend deformation. The bridge deformation is the result of the combined effect of periodic thermal dilation and linear trend-type deformation, so separating the thermal dilation from the trend deformation can help us better study the characteristic deformation mechanism of the bridge. Then the multi-source LOS thermal dilation is combined with the bridge structure and sensor geometry parameters, based on the natural neighborhood interpolation method, to obtain the longitudinal thermal dilation field. Based on the time and space interpolation methods and the principle of singular value decomposition, the LOS trend deformation obtained from the multi-source SAR data is geometrically aligned, interpolated, and fused to solve the bridge deformation longitudinal and vertical deformation field. In addition, the finite element model is established through the three-dimensional structure of the bridge and related structural mechanics principles.The normal stress information of the bridge is extracted by finite element modeling analysis based on the vertical fine deformation information of the bridge. The deformation and force characteristics are deeply explored to study the bridge's deformation mechanism and causes.

The research results show that the method can obtain the relationship between the deformation characteristics of bridges and their specific structures, which can also accurately extract the bridge thermal dilation and bridge 3D deformation, thus providing reliable data support for bridge health monitoring.

InSAR validation and comparison is required by the end-users to assess the quality of the results. Validation is applied via comparison of the InSAR data with ground truth data resulting from an independent source of measurements e.g. levelling. Cross comparison is used to evaluate the consistency of the products resulting from various InSAR techniques. Up to now, several projects have compared InSAR velocities and time series, e.g. PSIC4 (Crosetto et al., 2007; Raucoules et al., 2009), Terrafirma (Capes et al., 2009), Digital Environment (Sadeghi et al., 2021).

European Environment Agency provided European Ground Motion Service (EGMS) uses Senrinel-1 data to deliver consistent and reliable information regarding natural and anthropogenic ground motion over the Copernicus Participating States and across national borders, with millimeter accuracy (Crosetto 2020). With free availability of this data set, a new opportunity appears to allow comparison of a locally processed InSAR data set and assessment of the level of consistency between the locally processed InSAR data set and EGMS.

Spottitt Sp. zo.o. is developing a project co-financed by the European Union under the European Regional Development Fund. The project aims to develop a range of satellite based infrastructure monitoring solutions for owners of critical infrastructure such as power, gas and water network operators. One of the areas of particular interest to these infrastructure owners is the remote monitoring of the stability of their assets and the stability of the land in and around their assets. Land and asset motion negatively impact network integrity and reliability. Owners of power, gas and water networks spend millions on invasive monitoring of their high-risk assets and additional millions on repairs and mitigation activities across their networks. Network infrastructure owners are keen to understand whether free and open source Sentinel 1 data and InSAR techniques can be used to accurately, cost effectively and remotely monitor their entire networks for land and asset motion issues thus improving network performance and reliability.

We processed Quasi-PS InSAR analysis using SARPROZ software (Perissin 2011 and 2012) and Senrinel-1 data sets from 2020 on three rural test sites in Poland, all with 25 km of overhead power lines, and 25km of underground water pipelines. And three rural test sites in Italy, France, and UK all with 25 km of underground gas pipelines. To assess the quality of our results, we compared our estimated velocities and displacement time series with EGMS data sets which used the same Sentinel-1 images. Our methodology is based on the Digital Environment inter-comparison method (Sadeghi et al., 2021). We compared density, coverage, velocity and deformation time series after the pre-processing steps including solving any geo-coding issues between our outputs and the EGMS product. We will show and discuss the results in the full paper.

Acknowledgments:

Spottitt Sp. zo.o. is developing a project co-financed by the European Union under the European Regional Development Fund.

References:

Capes, R., Marsh, S., Bateson, L., Novali, F., & Cooksley, G. (2009). Terrafirma User Guide: A guide to the use and understanding of Persistent Scatterer Interferometry. ESA GMES Service Element, Available:https://core.ac.uk/download/pdf/385324.pdf.

Crosetto, M., Agudo, M., Raucoules, D., Bourgine, B., de Michele, M., Le Cozannet, G., Bremmer, C., Veldkamp, J., Tragheim, D., & Bateson, L. (2007b). Validation of Persistent Scatterers Interferometry over a mining test site: results of the PSIC4 project. In, Envisat Symposium,ESA (pp. 23-27)

Crosetto, M.; Solari, L.; Mróz, M.; Balasis-Levinsen, J.; Casagli, N.; Frei, M.; Oyen, A.; Moldestad, D.A.; Bateson, L.; Guerrieri, L.; Comerci, V.; Andersen, H.S. The Evolution of Wide-Area DInSAR: From Regional and National Services to the European Ground Motion Service. Remote Sens. 2020, 12, 2043. https://doi.org/10.3390/rs12122043

Perissin, D., Wang, Z., Wang, T., "The SARPROZ InSAR tool for urban subsidence/manmade structure stability monitoring in China", Proc. of ISRSE 2011, Sidney, Australia, 10-15 April 2011.

Perissin, D., and Wang, T., "Repeat-pass SAR Interferometry with Partially Coherent Targets", IEEE Transactions on Geoscience and Remote Sensing, Volume 50, Issue 1, Pages 271-

280, 2012.

Sadeghi, Z., Wright, T.J., Hooper, A.J., Jordan, C., Novellino, A., Bateson, L., Biggs, J., Benchmarking and inter-comparison of Sentinel-1 InSAR velocities and time series, Remote Sensing of Environment, Volume256,2021,112306,ISSN0034-4257,https://doi.org/10.1016/j.rse.2021.112306.

Acting as an effective carbon stock, forests are of paramount importance for the global carbon cycle. This delicate ecosystem is currently threatened and degraded by anthropogenic activities and natural hazards, such as deforestation, agricultural activities, farming, fires, floods, winds and soil erosion. Therefore, the availability of reliable, up-to-date measurements of forest resources, their evolution and the resulting impact on the carbon cycle is of great importance for environmental preservation and climate change mitigation. In this scenario, Synthetic Aperture Radar (SAR) systems, thanks to their capability to operate also in presence of clouds, represent an attractive alternative to optical sensors for remote sensing surveys over forested areas, especially over tropical forests, which are heavily affected by adverse weather conditions all year round.

In this work, we investigate the added-value of single-pass SAR interferometry (InSAR) with respect to repeat-pass InSAR and to classical SAR backscattering information, for mapping forests at large scale by using artificial intelligence. We present a study on the potential of Deep Learning (DL) for the regression of forest height from TanDEM-X bistatic single-pass data and from Sentinel-1 repeat-pass data. We propose a novel fully convolutional neural network (CNN) framework, trained in a supervised fashion using reference canopy height measurements derived from the LVIS airborne LiDAR sensor from NASA. The reference measurements were acquired during the joint NASA-ESA 2016 AfriSAR campaign over five tropical sites in Gabon, Africa. Together with the DL architecture and the training strategy, we present a series of experiments to assess the impact of different input features. In particular, regarding TanDEM-X, we concentrate on the use of:

• SAR backscatter in HH polarization,
• single-pass InSAR-related features such as the bistatic coherence and the volume decorrelation factor, which are not affected by temporal changes occurring during the acquisition of the interferometric image pair
• and geometry-related features such as the terrain elevation model and the local incidence angle.

The use of bistatic single-pass interferometry allows for exploiting the coherent information related to scattering mechanisms from a volumetric target, which is closely linked to the intrinsic characteristics and structure of vegetation. Our feature analysis shows that the TanDEM-X regression performance is primarily driven by bistatic InSAR features and that ancillary information about the acquisition geometry as well as scene topography is crucial to deliver peak performance. Differently, when considering Sentinel-1 data, due to the repeat-pass nature of the mission it is not possible to separate the volume decorrelation component from the temporal decorrelation one. In this case, the InSAR coherence becomes less informative compared to TanDEM-X and most of the information content can be extracted from the two polarization channels of the backscatter (VV and VH). Even with the limited penetration capability of X and C band radar waves into vegetation, the obtained results are extremely promising and already in line with state-of-the-art methods based on both physical-based modelling and data-driven approaches, with the remarkable advantage of requiring only one single input acquisition at inference time.

The grounding line is a subsurface geophysical feature that divides a grounded ice sheet and floating ice shelf. Knowledge of its precise location is required for estimating ice sheet mass balance, as ice discharged from the interior is typically calculated at the grounding line [1]. Grounding lines move back and forth as ice shelves bend and flex due to ocean tides. Identifying their migration patterns can provide insights into understanding ice sheet dynamics and overall ice sheet stability [2] and thereby improve the accuracy of numerical ice sheet models.

The spatial and temporal resolution of past and current satellite missions has enabled regular, continent-wide observation of Antarctica and other isolated glaciers with floating ice tongues. In particular, the high sensitivity of Interferometric SAR measurements to ground deformation has resulted in its application to grounding line location (GLL) mapping [3]. Specifically, the deformation at the grounding zone resulting from tidal flexure of the ice shelf is isolated from ice motion and topography in Differential InSAR (DInSAR) interferograms, under the assumption of steady ice velocity within the chosen temporal baseline. The tidal deformation is visible as a dense fringe belt and its landward limit is manually digitised as the GLL. Apart from being labour and time intensive, manual delineations are also inconsistent due to varying interpretations of experts in identifying the landward fringe, especially in areas with poor coherence or intricate fringe patterns.

The concept of automatic GLL delineation has recently gained attention and seen the development of several methodologies. [4] demonstrated a semi-automatic method that estimates the fringe frequency of wrapped phase in DInSAR interferograms. The grounding zone can be directly identified by computing the gradient of the estimated frequencies, thereby avoiding phase unwrapping. However, this approach requires an a priori grounding zone location. [5] developed a deep learning based automatic delineation pipeline in which the proposed DNN was trained on real and imaginary components of DInSAR phases from Sentinel-1 acquisitions.

This study further investigated the feasibility of DNNs for mapping the interferometric grounding line. The proposed DNN, based on the architecture of the Holistically-Nested Edge Detection network [6], was trained in a supervised manner, using manual delineations from the GLL product developed within ESA’s Antarctic Ice Sheet climate change initiative (AIS cci) project [7] as ground truth (Fig. 1 (a)). The GLL product contains manual delineations on 478 DInSAR interferograms computed from Sentinel-1A/B, ERS-1/2 and TerraSAR-X images acquired during 1992 - 2021.

The training feature stack consists of four interferogram-based features: real and imaginary components, interferometric phase and pseudo coherence (which is estimated by applying a boxcar filter to interferometric phase) derived from the corresponding DInSAR interferograms and five auxiliary features derived from several compiled datasets: TanDEM-X Polar DEM [8], horizontal and vertical components of ice velocity [9], tidal amplitude [10] and atmospheric pressure [11] (Fig. 1 (b)).

An automatic workflow that handles the preparation of the training feature stack, training and inference of the neural network and the post processing of network generated delineations was developed. The performance of the neural network was evaluated as the median deviation of the network generated GLLs from the manual delineations, quantified using the PoLiS metric [12]. Additionally, the importance of individual features was indirectly gauged by training several networks with different feature subsets and comparing their median deviations from the ground truth.

The DNN generated GLLs follow the landward limit of ice sheet flexure reasonably well, with the best network variant achieving a median deviation of 209 m from manual delineations.The contribution of auxiliary features was shown to be very weak, with their inclusion in the feature stack only slightly improving the delineation capability of the network. This finding is advantageous in terms of saving time, computational effort and memory in creating and storing the feature stack.

References

[1] E. Rignot and H. Thomas, “Mass balance of polar ice sheets,” Science, vol. 297, no. 5586, pp. 1502–1506, 2002. DOI: 10 . 1126 / science . 1073888. eprint: https : / / www . science . org / doi / pdf / 10 . 1126 / science . 1073888. [Online]. Available: https://www.science.org/doi/abs/10.1126/science.1073888.
[2] C. Schoof, “Ice sheet grounding line dynamics: Steady states, stability, and hysteresis,” Journal of Geophysical Research: Earth Surface, vol. 112, no. F3, 2007.
[3] E. Rignot, “Tidal motion, ice velocity and melt rate of petermann gletscher, greenland, measured from radar interferometry,” Journal of Glaciology, vol. 42, no. 142, pp. 476–485, 1996.
[4] A. Parizzi, “Potential of an Automatic Grounding Zone Characterization Using Wrapped InSAR Phase,” in IGARSS 2020 - 2020 IEEE International Geoscience and Remote Sensing Symposium, Waikoloa, HI, USA: IEEE, Sep. 2020, pp. 802–805, ISBN: 978-1-72816-374-1. DOI: 10.1109/IGARSS39084.2020.9323199.
[5] Y. Mohajerani, S. Jeong, B. Scheuchl, I. Velicogna, E. Rignot, and P. Milillo, “Automatic delineation of glacier grounding lines in differential interferometric synthetic-aperture radar data using deep learning,” Scientific reports, vol. 11, no. 1, pp. 1–10, 2021.
[6] S. Xie and Z. Tu, “Holistically-nested edge detection,” in Proceedings of the IEEE international conference on computer vision, 2015, pp. 1395–1403.
[7] A. Groh, Product user guide (pug) for the antarctic ice sheet cci project of esa’s climate change initiative, version 1.0, 2021. [Online]. Available: https://climate.esa.int/media/documents/ST-UL-ESA-AISCCI-PUG-
0001.pdf.
[8] M. Huber, Tandem-x polardem product description, prepared by german remote sensing data center (dfd) and earth observation center, 2020. [Online]. Available: https://www.dlr.de/eoc/en/desktopdefault.aspx/
tabid-11882/20871_read-66374.
[9] T. Nagler, H. Rott, M. Hetzenecker, J. Wuite, and P. Potin, “The sentinel-1 mission: New opportunities for ice sheet observations,” Remote Sensing, vol. 7, no. 7, pp. 9371–9389, 2015.
[10] L. Padman, S. Erofeeva, and H. Fricker, “Improving antarctic tide models by assimilation of icesat laser altimetry over ice shelves,” Geophysical Research Letters, vol. 35, no. 22, 2008.
[11] E. Kalnay, M. Kanamitsu, R. Kistler, W. Collins, D. Deaven, L. Gandin, M. Iredell, S. Saha, G. White, J. Woollen, et al., “The ncep/ncar 40-year reanalysis project,” Bulletin of the American meteorological Society, vol. 77, no. 3, pp. 437–472, 1996.
[12] J. Avbelj, R. M ̈uller, and R. Bamler, “A metric for polygon comparison and building extraction evaluation,” IEEE Geoscience and Remote Sensing Letters, vol. 12, no. 1, pp. 170–174, 2014.

Background:
Snow water equivalent (SWE) is an essential climate variable due to its importance for regional and global water resource. For mapping of SWE from local to global scales, remote sensing techniques are the only efficient method. Microwave techniques are a preferred choice for depth-sensitive mapping during winter conditions with little daylight or strong cloud coverage. Compared to km-scale passive microwave radiometry, SAR based methods provide the spatial resolution required to resolve variations in SWE related to local topography. Substantial efforts on SWE retrieval have focused on using radar backscatter at different frequencies and polarizations. These studies have met with mixed success because the models do not capture the dynamics of the snowpack. Alpine, but also polar snowpack, generally has a complex scattering and absorption behavior caused by spatial and temporal inhomogeneity of the snow structure due to compaction, sublimation, freeze-thaw cycles, and liquid water content [Tan 2015] [Zhu 2018] [Zhu 2021].

It is known that dry snow has relatively low attenuation at frequencies < 10 GHz and acts as a dielectric layer above the ground if the ice structures of different scales (grains, grain-clusters, ice crusts and snow layers) within the snowpack are of significantly smaller scale than the wavelength. An almost linear relation of SWE to microwave propagation delay has been proposed and demonstrated [Guneriussen 2001, Leinss 2015]. Given that there is little change in the configuration of scatterers in the time interval between radar measurements and that the snowpack remains dry, then interferometric phase measurements can potentially be used to track changes in SWE. In this approach, short period interferograms from temporally adjacent pairs of observations are calculated for the entire stack. The interferometric phases are summed at each point in time to determine the cumulative phase due to propagation through the snowpack as a function of time. If the time intervals are sufficiently short, changes in the propagation path length are expected to be less than 𝜆⁄2 meaning that the short-period interferometric phase is in the range of ±𝜋, thus avoiding the need for phase unwrapping.

For conditions where melt events are frequent, like, e.g., alpine snow, the main challenge to the interferometric approach to SWE retrieval is not only the loss of interferometric coherence by changing scattering properties, but also large changes in the index of refraction due to the addition liquid water from melting snow layers. Another source of error is due to insufficient temporal sampling of the interferometric phase signal. During transient melt conditions (frequently coinciding with strong snow fall) the phase signal can change very rapidly causing phase changes exceeding ±𝜋. Loss of interferometric coherence translates directly into possibly loosing track of the SWE related phase signal.

Even though promising solutions have been proposed to mitigate the problem of phase unwrapping [Eppler 2022] on the km-scale, and to address the phase-calibration including fusion of optical snow cover maps with radar data [Tarricone 2022], the choice of the optimal frequency (or set of frequencies) for interferometric estimation of SWE is still a topic of current research. While L-band measurements are relatively robust against coherence loss and melting [Tarricone 2022], X- and Ku-band measurements can provide very accurate information about SWE changes under optimal dry snow conditions [Leinss 2015].

Methods and Data:
In this contribution we present interferometric data acquired by the Gamma WBSCAT coherent scatterometer. WBSCAT covers the frequency range from 1-40 GHz and is capable of making coherent polarimetric measurements of radar backscatter multiple times each day. The instrument was installed in Davos-Laret, located at an altitude of 1514 meters a.s.l. in Switzerland. As part of the ESA Snowlab (2018-2019) and Snowlab-NG (2019-2020) projects, the radar measurements are part of a comprehensive data set including radiometric microwave emission to estimate the liquid water column height [Naderpour 2022], meteorological data (air and snow surface temperatures, precipitation), and snowpack characteristics, e.g., snow height, moisture content, snow water equivalent (SWE), snow density, and snow structure.

WBSCAT is based on a vector network analyzer (VNA) using internal standards to calibrate the instrument. The instrument worked reliably during these observation seasons producing time-series of radar scattering coefficient 𝜎0, interferometric phase, and coherence. WBSCAT was mounted on a 2.5-meter rail, inclined 45 degrees from horizontal, located on a static tower, 8 meters above the ground surface (Figure 1). Radar tomographic profiles were calculated from measurements acquired over the rail aperture and show scattering layers in the snowpack [Frey 2023].

Data were acquired every 8-hours beginning in late November and continuing until late April in three overlapping frequency bands 1-6, 3-18, and 16-40 GHz and at three different incidence angles (25, 35, 45 degrees) over a 90-degree azimuth sector, sampled every 3-4 degrees. Each frequency band was bandpass filtered for a set of frequencies using a Kaiser window, followed by oversampling and FFT to obtain the range-compressed radar echo profiles. The complex-valued range echoes from sequential acquisition pairs are used to form 8-hour interferograms and coherence maps for each sub-band. Data samples at slant ranges near the center of the antenna elevation pattern are used to estimate the coherence and interferometric phase of each acquisition pair. Time series of integrated phase differences were calculated by summing interferometric phase differences under the condition that the coherence was above a specified threshold.

Results:

Integrated 8-hour phase differences and coherence are compared with the in-situ measurements of snow height, snow surface temperature, and snow-water equivalent (Figure 3-5), and liquid water column height (Figure 6). The time-series of correlation coefficients are shown for frequency sub-bands centered at 2, 3, and 5 GHz in Figures 7 (a-c). The integrated phase differences are shown for these frequencies in Figures 7 (d-f). These data were collected with an incidence angle of 45 degrees.

Periods of low correlation coincide with temporal increases in the column liquid water content (Figure 7a-c vs. Figure 5), e.g., between 2019-12-16 and 2019-12-23 and during five events in February 2020. During these periods of low coherence, the 8-hours interferometric phase (not shown) contains large variation that were filtered out by the coherence threshold of 0.7. After 2020-03-09 the snowpack compacts (Figure 3) due to melting conditions with runoff after 2020-04-01 (Figure 6). In the radiometry-derived liquid water column (Figure 5) daily freeze- thaw cycles are observed.

The integrated phase differences at 2, 3 and 5 GHz, shown in Figure 7 (d-f), show a good correlation with the temporal evolution of SWE (Figure 6) as already shown for dry snow (Leinss 2015). Surprisingly, the magnitude of the integrated phase is not proportional to the radar frequency as it would be expected for a frequency- independent propagation delay. The frequency-dependence of the permittivity of liquid water (Buchner 1999), together with lost phase cycles due to coherence loss, might explain this observation. Another surprising observation is that during the five melt cycles in February 2020, the 2 GHz integrated phase differences (Figure 7d) shows a significant negative trend despite increasing SWE. A reason could be that during snow melt coherence is lost, while during the subsequent refreeze period the propagation delay continuously decreases (cf. Figure 5). Note also that large snowfall events are often characterized by periods when the temperature is near freezing with low correlation. The large amount of snow during such events can result in large phase jumps with magnitude greater than 𝜋, resulting in lost phase and underestimation of the integrated phase delay.

Discussion:
The Davos-Laret site is characterized by periods of freezing and thawing of the snowpack practically during the entire season resulting in a varying snow moisture content. This liquid water content counters to the assumption that scattering comes primarily from the ground rather than the snowpack. During the periods when the snowpack remained frozen, as indicated from the high interferometric correlation, low temperatures, and low column water content, the integrated phase closely tracks the SWE. Selection of the frequencies better suited for SWE estimation is determined by the trade-off of requirements that on one hand decorrelation is minimized and the phase variation ambiguity can be resolved if the magnitude of the phase change exceeds 𝜋, and on the other hand, that there is sufficient sensitivity to changes in SWE. One of the observations from this data set is that the integrated phase is insensitive to changes in the height of the snowpack but responds to the amount of snowfall. Furthermore, the short-term interferometric phase changes exceed 𝜋 even at low frequencies (< 3 GHz) implying that spatial and/or temporal phase unwrapping are required to resolve the phase ambiguities in the integrated phase.
Acknowledgements:
This work was performed at Gamma Remote Sensing in collaboration with the WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland as part of the ESA-funded project: “Scientific Campaign Data Analysis Study for an Alpine Snow Regime SCANSAS (ESA SCANSAS), C ontract No. 4000131140/20/NL/FF/ab. Contract No. 4000131140/20/NL/FF/ab. ESA SnowLab campaign and data processing: ESA/ESTEC Contract No. 4000117123/16/NL/FF/MG the and ESA Wide-Band Scatterometer development: ESA/ESTEC Contract No. 4000117123/16/NL/FF/mg.

References:
R. Buchner, J. Barthel, and J. Stauber, “The dielectric relaxation of water between 0°C and 35°C,” Chem. Phys. Lett., vol. 306, no. 1-2, pp. 57–63, 1999, doi: http://dx.doi.org/10.1016/S0009-2614(99)00455-8
O. Frey, A. Wiesmann, C. Werner, R. Caduff, H. Löwe, M. Jaggi, SAR Tomographic Profiling of Seasonal Alpine Snow at L/S/C-Band, X/Ku-Band, and Ka-Band Throughout Entire Snow Seasons Retrieved During the ESA SnowLab Campaigns 2016-2020, FRINGE ESA meeting, Leeds, England, 11-15 Sept. 2020

T. Guneriussen, K. A. Høgda, H. Johnsen, and I. Lauknes, “InSAR for estimation of changes in snow water equivalent of dry snow,” IEEE Trans. Geosci. Remote Sens., vol. 39, no. 10, pp. 2101–2108, 2001, doi: http://dx.doi.org/10.1109/36.957273.

S. Leinss, A. Wiesmann, J. Lemmetyinen, and I. Hajnsek, “Snow water equivalent of dry snow measured by differential interferometry,” IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 8, no. 8, pp. 3773–3790, 2015-06-17, doi: 10.1109/JSTARS.2015.2432031.

J. Eppler, B. Rabus, and P. Morse, “Snow water equivalent change mapping from slope-correlated synthetic aperture radar interferometry (InSAR) phase variations”, The Cryosphere, 16, 1497–1521, https://doi.org/10.5194/tc-16-1497-2022, 2022.

R. Naderpour, M. Schwank, D. Houtz and C. Mätzler, "L-Band Radiometry of Alpine Seasonal Snow Cover: 4 Years at the Davos-Laret Remote Sensing Field Laboratory," in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 15, pp. 8199-8220, 2022, doi: 10.1109/JSTARS.2022.3195614.

S. Tan, W. Chang, L. Tsang, J. Lemmetyinen and M. Proksch, "Modeling Both Active and Passive Microwave Remote Sensing of Snow Using Dense Media Radiative Transfer (DMRT) Theory With Multiple Scattering and Backscattering Enhancement," in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 8, no. 9, pp. 4418-4430, Sept. 2015, doi: 10.1109/JSTARS.2015.2469290.

J. Tarricone, R. Webb, H. P. Marshall, A. W. Nolin, and F. J. Meyer: Estimating snow accumulation and ablation with L-band InSAR, The Cryosphere Discuss. [preprint], https://doi.org/10.5194/tc-2022-224, in review, 2022.

J. Zhu, S. Tan, J. King, C. Derksen, J. Lemmetyinen, and L. Tsang. "Forward and Inverse Radar Modeling of Terrestrial Snow Using SnowSAR Data," in IEEE Transactions on Geoscience and Remote Sensing, vol. 56, no. 12, pp. 7122-7132, Dec. 2018, https://doi.org/10.1109/TGRS.2018.2848642.

J. Zhu, S. Tan, L. Tsang, D. Kang, and E. Kim. “Snow water equivalent retrieval using active and passive microwave observations,” Water Resources Research, 57, e2020WR02756, 2021, https://doi. org/10.1029/2020WR027563

This work presents a novel method for assimilation of meteorological data, SAR-derived Surface Soil moisture (SSM) and interferometric SAR (InSAR)-derived Ground Surface Motion (GSM) for monitoring peatlands condition and hydrology in the Forth Valley, Scotland. We use SAR imagery from the Sentinel-1 SAR satellite and meteorological observations obtained from short-latency ground weather stations. We present preliminary findings that qualitatively analyse the relationship between SSM, GSM, PET, Net rainfall and peatland Water Table Depth (WTD) measurements captured by a network of ground water loggers. We also present the GSM in peatland sites in good condition and compare it with GSM in sites where it is known that bare peat exists. We show that degraded peatland show no signs of hydrology-driven seasonality and presents a negative trend in surface motion indicating subsidence.

In recent decades, with the increase of population, the land reclamation is often occurring in both mountainous regions and coastal areas to extend the land for urban construction and airport construction in many countries. In China, for example, Lanzhou city is one of the typical cities with many civil engineering projects for mountain excavation and city construction (MECC) on the Loess Plateau since 1997, which has changed the landscape significantly and resulted in the surface deformation in both vertical and horizontal directions. To monitor the multi-dimensional surface deformation reliably, the height changes cannot be omitted, as it changes frequently from meters to over 50 meters. Therefore, there exist four questions, that is, firstly, whether do SAR images keep coherent before and after land reclamation? Secondly, can height change time series be estimated with multi-temporal InSAR technique? Thirdly, what is the surface deformation time series during the land reclamation over several years? And lastly, can we get the multi-dimensional surface deformation by fusing ascending and descending SAR images? Therefore, we propose an improved time series InSAR technical flowchart with the emphasis on the following key steps.

Firstly, we determine the subsets of interferometric pairs for a generic pixel according to the landfill time, which can be detected according to jump of the cumulative deformation phase.

Secondly, the height changes are estimated as the DEM errors in each subsets individually with the Least Squares (LS) method, where long spatial baseline, short time baseline and high coherence interferograms are involved. Then DEM errors are corrected in all interferograms in each subsets, respectively.

Thirdly, the surface deformation time series in line-of-sight is estimated for interferograms with short spatial and short temporal baselines with Least Squares (LS) or Singular Value Decomposition (SVD) method.

Lastly, the two dimensional surface deformation time series in vertical and east-west directions are estimated by fusing ascending and descending LOS deformation results.

Three tracks Sentinel-1 SAR images from October 09, 2014 to May 17, 2022 are tested over Chengguan District, Lanzhou City, China, which is one of the typical MECC region. In total 513 SAR images are involved. Firstly, height changes are successfully obtained ranging from -80 meter to 70 meter, where correlation coefficient of height estimation is achieved over 0.89 between two results from independent SAR tracks. Secondly, the cumulative vertical deformation and east-west deformation time series is retrieved by using one ascending and two descending tracks SAR data. The maximum cumulative vertical deformation exceeds -600 mm from November 2014 to May 2022. And the maximum cumulative east-west deformation exceeds -300 mm from November 2014 to May 2022. We can conclude that the main reason for the two dimensional deformation is the soil compaction in vertical and opposite horizontal directions.

Permafrost is a common characteristic of Arctic landscapes, where it refers to ground that remains at or below 0 °C for a duration of at least two consecutive years. Permafrost underlies approximately 15 % of the landmass in the Northern Hemisphere and is becoming more susceptible to rapid thawing as the climate continues to warm (Obu et al. 2019). When ice-rich permafrost thaws it can alter the surface characteristics of a landscape which is commonly referred to as thermokarst. Retrogressive Thaw Slumps (RTS) are emerging as one of the most dynamic types of thermokarst, varying strongly in shape and thawing behavior. The prevalence and distribution of rapid thaw on a pan-Arctic scale are not well understood and so is its potential contribution in the Arctic carbon-climate feedback (Kokelj et al. 2009). High-resolution Digital Elevation Models (DEMs) are a valuable tool for monitoring surface characteristics of thermokarst features and track changes over time, which in turn improves our understanding of large-scale landscape changes and their implications for hydrology, biochemistry, permafrost stability, and hazard risk management (Jorgensson and Grosse 2016). To derive these DEMs, a range of techniques are employed, including ground-based and aerial LiDAR (e.g., Patton et al. 2021), optical stereo-imagery from airborne (e.g., Lim et al. 2020) and satellite platforms (e.g., Günther et al. 2015). The high-resolution ArcticDEM has been used to supplement optical satellite data in monitoring highly dynamic thermokarst features such as RTS towards the pan-Arctic scale (Yang et al. 2023). However, these methods are subject to spatial coverage and availability constraints, or data quality issues and data gaps due to limitations such as cloud cover, seasonal snow, vegetation, and illumination conditions for passive optical sensors.

Another high-resolution DEM covering the Arctic landscape has been available with the start of the TanDEM-X satellite in 2010, forming together with the TerraSAR-X satellite the TanDEM-X constellation, a bistatic single-pass radar system. The temporal, spatial and vertical resolution of the TanDEM-X mission (10-12 m spatial resolution and approx. 2 m vertical accuracy over Arctic regions) merits investigation for a comprehensive monitoring of rapid permafrost thaw and directly retrieve information about volumetric change rates and thus carbon mobilization. This approach has already been successfully applied to single-pass InSAR-based time-series DEM analysis to detect and quantify volumetric change rates and potential carbon mobilization of RTSs in several test sites in the Arctic permafrost region (Bernhard et al. 2020, Bernhard et al. 2022a, Bernhard et al. 2022b). Uncertainties that still remain include the potential error in the volumetric change rate estimation due to viewing geometry of the SAR sensor, the assumption of complete penetration of the dry winter snowpack of the radar waves, as well as systematic differences between wave polarizations with respect to penetration of snow and vegetation.

In this paper we present the learnings from a time-series TanDEM-X case study in the Mackenzie River Delta that addresses the pending uncertainties when applying TanDEM-X derived DEMs to RTS monitoring. Our study involves a general analysis of the produced DEM accuracy for Arctic permafrost regions, as well as targeted investigations at known RTS locations. The accuracies of the generated DEMs are compared with the high-resolution DEM from a LiDAR campaign (Anders et al. 2018) and the ArcticDEM products to improve the understanding of the underlying accuracies. Potential discrepancies in height accuracies due to viewing geometry of the SAR sensor are assessed through the comparison of DEMs generated from TanDEM-X observations with different orbit directions. Furthermore, the impact of snow and vegetation cover on the penetration of the radar waves to the ground and resulting height discrepancies is investigated. For this investigation we choose the upland region to the east of the Mackenzie River Delta which is located in the western Canadian Arctic and is characterized by a subarctic climate. The region is dominated by tundra vegetation and contains large amounts of ground ice. Studies found a high concentration of relatively small RTSs with head wall heights of 2-10 meters (Kokelj et al. 2013). In addition to the global TanDEM-X bistatic single-pol observations (availability in Arctic permafrost landscapes: 2010/11/12 and 2016/17), additional observations with a variety of observation properties are available for the study region: Bistatic dual-polarization observations are available in 2018/19, as well as high temporal resolution time-series (11-day repeat pass) during multiple periods between 2011 and 2022. The data from the TanDEM-X Science Phase in 2015 offers high baselines yielding vertical accuracies on sub-meter level. All observations with height of ambiguities greater than 80 meters are removed ensuring acceptable vertical accuracy needed for RTS detection. DEMs are generated with standard InSAR techniques from the pairs of TanDEM-X images and are differenced on multiple timescales. RTS locations and shapes provided by Bernhard et al., 2022a are used to analyze DEM accuracy for RTS feature characterization.

References

Anders, Katharina; Antonova, Sofia; Boike, Julia; Gehrmann, Martin; Hartmann, Jörg; Helm, Veit; Höfle, Bernhard; Marsh, Philip; Marx, Sabrina; Sachs, Torsten (2018): Airborne Laser Scanning (ALS) Point Clouds of Trail Valley Creek, NWT, Canada (2016). PANGAEA, https://doi.org/10.1594/PANGAEA.894884, Supplement to: Antonova, Sofia; Thiel, Christian; Höfle, Bernhard; Anders, Katharina; Helm, Veit; Zwieback, Simon; Marx, Sabrina; Boike, Julia (2019): Estimating tree height from TanDEM-X data at the northwestern Canadian treeline. Remote Sensing of Environment, 231, 111251, https://doi.org/10.1016/j.rse.2019.111251

Bernhard, P., Zwieback, S., Leinss, S., & Hajnsek, I. (2020). Mapping Retrogressive Thaw Slumps Using Single-Pass TanDEM-X Observations. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 13, 3263–3280. https://doi.org/10.1109/JSTARS.2020.3000648

Bernhard, P., Zwieback, S., Bergner, N., & Hajnsek, I. (2022a). Assessing volumetric change distributions and scaling relations of retrogressive thaw slumps across the Arctic. The Cryosphere, 16(1), 1–15. https://doi.org/10.5194/tc-16-1-2022

Bernhard, P., Zwieback, S., & Hajnsek, I. (2022b). Accelerated Mobilization of Organic Carbon from Retrogressive Thaw Slumps on the Northern Taymyr Peninsula. https://doi.org/10.5194/tc-2022-36

Günther, F., Overduin, P. P., Yakshina, I. A., Opel, T., Baranskaya, A. V., & Grigoriev, M. N. (2015). Observing Muostakh disappear: Permafrost thaw subsidence and erosion of a ground-ice-rich island in response to arctic summer warming and sea ice reduction. The Cryosphere, 9(1), 151–178. https://doi.org/10.5194/tc-9-151-2015

Jorgenson, M. T., & Grosse, G. (2016). Remote Sensing of Landscape Change in Permafrost Regions. Permafrost and Periglacial Processes, 27(4), 324–338. https://doi.org/10.1002/ppp.1914

Kokelj, S. V., Lantz, T. C., Kanigan, J., Smith, S. L., & Coutts, R. (2009). Origin and polycyclic behaviour of tundra thaw slumps, Mackenzie Delta region, Northwest Territories, Canada. Permafrost and Periglacial Processes, 20(2), 173–184. https://doi.org/10.1002/ppp.642

Kokelj, S. V., & Jorgenson, M. T. (2013). Advances in Thermokarst Research. Permafrost and Periglacial Processes, 24(2), 108–119. https://doi.org/10.1002/ppp.1779

Lim, M., Whalen, D., J. Mann, P., Fraser, P., Berry, H. B., Irish, C., Cockney, K., & Woodward, J. (2020). Effective Monitoring of Permafrost Coast Erosion: Wide-scale Storm Impacts on Outer Islands in the Mackenzie Delta Area. Frontiers in Earth Science, 8, 561322. https://doi.org/10.3389/feart.2020.561322

Obu, J., Westermann, S., Bartsch, A., Berdnikov, N., Christiansen, H. H., Dashtseren, A., Delaloye, R., Elberling, B., Etzelmüller, B., Kholodov, A., Khomutov, A., Kääb, A., Leibman, M. O., Lewkowicz, A. G., Panda, S. K., Romanovsky, V., Way, R. G., Westergaard-Nielsen, A., Wu, T., … Zou, D. (2019). Northern Hemisphere permafrost map based on TTOP modeling for 2000–2016 at 1 km2 scale. Earth-Science Reviews, 193, 299–316. https://doi.org/10.1016/j.earscirev.2019.04.023

Patton, A. I., Rathburn, S. L., Capps, D. M., McGrath, D., & Brown, R. A. (2021). Ongoing Landslide Deformation in Thawing Permafrost. Geophysical Research Letters, 48(16). https://doi.org/10.1029/2021GL092959

Yang, Y., Rogers, B. M., Fiske, G., Watts, J., Potter, S., Windholz, T., Mullen, A., Nitze, I., & Natali, S. M. (2023). Mapping retrogressive thaw slumps using deep neural networks. Remote Sensing of Environment, 288, 113495. https://doi.org/10.1016/j.rse.2023.113495

The Interferometric Synthetic Aperture Radar (InSAR) technique allows the precise monitoring of ground displacements over wide areas based on radar data. Several satellites carrying synthetic radar antennas are orbiting around the Earth at the time. In particular, the Sentinel ESA mission provides open data from two SAR satellites operating at the global scale with a return time of six days. This allows the scientific community to dispose of consistent and daily updated dataset for a wide range of applications.

In this context, out of commercial missions, it is fundamental for the community to dispose of open-source and free software packages for SAR data processing. One of them is the widely used Stanford Method for Persistent Scatterers (StaMPS) for InSAR processing, which provides time series of range variations over a cluster of points starting from both amplitude and phase raw observations. These points are the so-called Persistent Scatterers (PS), namely pixels in a series of interferograms characterized by amplitude stability and signals not obscured by the phase noise. For each PS, StaMPS basically provides the mean velocities of displacement in the range direction over the inspected observing period. Besides, the software gives ancillary parameters such as the Phase Coherence, the RMS of the estimated velocities, the topography, the wrapped and unwrapped phase. StaMPS outputs are also the time series of unwrapped phase observations, expressed in terms of displacements, and the time series of corrections related to satellite ephemerids, atmosphere, orbits, master and slaves. To perform a smart and detailed analysis of these InSAR output time series, the TimeSAPS software package has been developed.

TimeSAPS works starting from StaMPS outputs and for each PS it allows to perform analysis characterizing the time series in terms of linear trends, periodical signals and the related phase and amplitude, frequency power spectrum and residuals with respect to both linear and periodical models. In detail, linear trends and periodical signals are estimated at once using the Gauss-Markov model with a least squares approach. As for the characteristic frequencies of the periodical signals, these can be defined by the users or estimated through a Lomb-Scargle periodogram. In both cases, the composition of up to five sine-wave signals can be computed to represent deformation models characterized by highly irregular shapes. In other words, TimeSAPS provides users with a tool capable of analyzing the StaMPS outputs behind the linear characterization of the PS displacements.

Further strengths of the software packages are its implementation in the Matlab environment, the same used for StaMPS and its capability of producing output in the shapefile format, directly importable in whatever GIS environment. Furthermore, the analysis can be basically applied to any kind of InSAR output, independently by the used SAR processing software, just by converting them into the StaMPS file format.

As the largest water transfer in the world, China’s South-to-North Water Diversion Project (SNWDP) consists of the East Route Project (ERP), the Middle Roue Project (MRP), and the pending West Route Project (WRP). The MRP, constructed beginning in 2002 and started operation in 2014, transfers water from the Yangtze River to arid northern China. It is near the south-north direction with a total length of 1432 km and is composed of underground box culverts, buildings (dams, aqueducts, bridges, inverted siphons, and ventholes), and open concrete-lined canals. Together with the local poor geological conditions such as swelling soil, mining, or groundwater overexploitation areas along the route, this man-made canal is vulnerable to geological disaster.

The Sentinel-1 data with a wide swath makes it practical to obtain large-scale ground deformation along the MRP, and the integration of multi-sensor InSAR measurements contributes to investigations into the long-term displacement evolution of specific canal sections. In this study, multi-scale deformation monitoring for the whole MRP by Sentinel-1 data was conducted, and the potential unstable canal sections were identified, most of which are caused by regional deformation. For example, the buried box culverts of Tianjin Branch Route (TBR) passes through the severe subsidence funnels in North China Plain induced by overexploitation of groundwater, and a few short canal sections in Henan Province are deformed due to surrounding coal mining areas or swelling soil areas. Only a few canal sections are deformed due to construction health, such as the canals in Jiaozuo City and Ye County. The large buildings, such as the Danjiangkou Reservoir and Shehe Aqueduct are stable.

The high-fill canals and deep-cut canals are prone to deform due to construction health. Take the Jiaozuo high-fill canal as an example, Sentinel-1 data covering the MRP operation period were processed to analyze the deformation evolutions and behaviors. The positive correlation between the canal settlements and embankment heights together with long-term consolidation curves reveals that the deformation is caused by post-construction consolidation of filling materials. Moreover, the different parts of embankment exhibit distinct deformation behaviors responding to the extreme rainstorm in July, 2021, the intrinsic relations of which with canal structure and soil wetting need to be further determined. For the deep-cut canal in swelling soil area, the uplift deformation related to the unloading rebound occurs. In addition, the distributed scatterers (DS) InSAR method was used to process the high-resolution TerraSAR-X data, revealing the deformation characteristics of embankment crests and back slopes in more detail.

By contrast, the stability of half-cut and half-fill canals is affected by surrounding deformation. For the case of the Changge canal, its deformation evolution derived from multiple satellites, including ENVISAT ASAR, ALOS-1 PALSAR-1, Sentinel-1, and TerraSAR-X, covering the pre- and post-operation, reveals that its instability is related to surrounding coal mining activities. The 2D deformation and distortions along the canal obtained from multi-track InSAR results illustrate that this canal section is subject to both horizontal and vertical distortion in a short distance. Furthermore, for fine monitoring, the ascending and descending TerraSAR-X results were interpreted on a structure level with consideration of SAR applicability. The distribution of InSAR measurement points on different canal structures and the sensitivity of LOS deformation to monitor a specific deformation vector were discussed by calculating the InSAR visibility and sensitivity.

In conclusion, the MRP is overall stable except for some short canal sections and the TBR. The deformation related to canal itself mainly occurs on high-fill canals or deep-cut canals. Satellite InSAR can obtain long-term and large-scale deformation evolution of other artificial water transfer projects with high efficiency and low cost. The deformation behaviors of different canal types as well as the structure level interpretation apply to canals with similar structure, beneficial for cause diagnosis and maintenance work.

Land subsidence is a major geohazard that causes significant damage to infrastructure and poses a threat to people. In Iran, land subsidence has been reported in several regions, primarily due to the over-extraction of groundwater for irrigation. The use of open data and remote sensing technologies can provide valuable insights into the extent and impact of land subsidence on both the population and infrastructure.
In this study, we used open data from multiple sources to estimate the risk of subsidence to the population and infrastructure across Iran. We used the entire archive of Sentinel-1 and performed a Small Baseline analysis with interferograms multilooked to 100 meters spatial resolution. The unwrapped phase time series were corrected for elevation-correlated and broad-scale phase changes using a patch-wise approach with patch sizes of 25 km. Then, seismic signals were identified and removed from the time series by considering significant events in the USGS earthquake catalog. The final average velocity was masked by slope based on the Copernicus DEM. Subsidence candidates were identified based on average deformation rates, converted to vertical, greater than 1.5 cm/year. Finally, subsidence zones were determined by calculating connected components larger than 5 km2.
We used the angular distortion estimated from land subsidence rates and the population density from Worldpop data to assess land subsidence risk to the population. First, we adjusted the 1-km-resolution Worldpop data for the actual population based on national census statistics available at the district level. Next, we upscaled the population density to 100x100 meters using the built-up areas from the Copernicus Land Cover map. The angular distortion and the population density were then combined in a 3x3 risk matrix to estimate land subsidence risk to the population. Different categories of hazard and vulnerability were defined based on the Jenk natural breaks of angular distortion and population density. Our results demonstrate that one-fifth of Iran's population lives in areas directly affected by land subsidence. While 7 million of them reside in low subsidence risk areas, 8 million are in medium-risk areas, and 1.5 million are in high-risk areas.
We also combined the angular distortion with linear infrastructure data, including roads, railways, and power lines, from the OpenStreetMap (OSM) database. We used this information to estimate the risk of subsidence on infrastructure across the country. The results suggest that 31 of 51 rail lines across the country are posed with subsidence risks, with 0.5% of railways at high subsidence risk. Furthermore, 0.5% of roads and 1% of power lines are at high risk of subsidence.
The use of open data was critical to the success of our study. By leveraging openly available data from multiple sources, we were able to develop a comprehensive subsidence map for Iran. This map provides valuable information to policymakers and planners who can use it to develop strategies for mitigating the impact of subsidence on infrastructure and the population. Therefore, we made our results available as raster maps in Web Map Service (WSM) and vector data in MapBox Vector (MVT) formats. These data can be loaded into free GIS software, allowing researchers and policymakers to combine the data with other information.

The determination of ground deformation can be realised by applying various measurement methods such as levelling, laser scanning, gravimetry, satellite navigation systems, synthetic aperture radar (SAR), and many others. However, providing sufficient spatio-temporal resolution of 3-D deformation with high accuracy can be very challenging using only one method. Therefore, the application of multiple complementary methods allows the establishment of an overall system for the determination of three-dimensional displacement values and movement rates. In this study, we focus on exploiting strengths and reducing weaknesses of Global Navigation Satellite Systems (GNSS) and Differential Interferometry SAR (DInSAR) techniques by providing a new methodology of integration involving Kalman filter algorithms for non-linear ground displacements.

An unquestionable advantage of GNSS technology is the possibility of continuous monitoring of deformations in three-dimensional space. Moreover, the evolution of GNSS estimation methods allows for obtaining a highly precise position determination with a relatively slight latency (ranging from a few seconds to less than one hour). The limitation of satellite navigation technology is the spatial range of the measurements. Ground deformations can only be observed at the point where the GNSS antenna is located. Additionally, acquisition, installation, and maintenance of equipment may also involve high costs. At least several dozen GNSS receivers are needed to acquire a ground system for monitoring horizontal and vertical movements across an area of interest. Moreover, some technical issues related to, e.g., power loss may introduce significant interruptions in the time series of observations.

In contrast to the GNSS technique, the InSAR methods enable the detection of large-scale subsidence areas with the possibility to use free products and software (eg, Sentinel-1 and SNAP). Large-scale InSAR investigations provide a better overview of local landform changes. The radar imagery coverage ranges from 5 to 250 km with ground resolution from 0.5 to 20 m. Unfortunately, InSAR methods also have some limitations related to data acquisition technology related to a few days latency in acquiring new products in only one LOS (line-of-sight) direction. Due to the nearly north-south trajectory of the SAR satellites, the system has limited sensitivity to ground movements in this direction. Furthermore, the InSAR time series of displacements can be affected by outlier values related to the limitations of the technique, e.g., decorrelation in vegetated areas, local atmospheric effects, or other phase unwrapping problems.

The main goal of this research is to determine a persuasive integration between the data acquired by the DInSAR and GNSS methods regarding the capabilities and limitations of these two techniques. The paper presents an original methodology for the integration of two different techniques, optimal for strong non-linear motions, conducted for an area affected by underground mining works. The process of fusion is based on the Kalman filter approach, which is able to ingest the time series of GNSS topocentric coordinates with significant gaps and noisy time series of DInSAR ascending and descending LOS velocities subject to troposphere artefacts or improper SAR phase unwrapping errors.

Geotechnical slope stability monitoring is a critical aspect of managing the safety and integrity of constructed and natural slopes. Slopes can be affected by various factors such as rainfall, seismic activity, soil erosion, and human activities, which can result in landslides, slope failures, and infrastructure damage. It is, therefore, essential to monitor slope stability to ensure the safety of infrastructure and for protecting the environment.

Slope monitoring can be done using both in-situ measurements and remote sensing observations. In-situ measurements involve placing instruments directly on, and within, the slope, to collect detailed and accurate data, but may be limited to a specific location or small area. Remote sensing observations, on the other hand, involve using technologies such as LiDAR, satellite imaging, and aerial photography to remotely gather data on slope conditions. In recent years, Interferometric Synthetic Aperture Radar (InSAR) has emerged as a powerful remote sensing tool for monitoring slope deformation patterns. InSAR can provide measurements over large areas, making it possible to monitor multiple slopes simultaneously. Also, it can deliver continuous monitoring without the need for physical instrumentation, reducing the cost and labour required for monitoring.

However, the use of InSAR techniques can be limited in vegetated slopes, where the number of coherent scatterers is reduced or non-existent. In vegetated areas, several factors, including vegetation type, density, and moisture content, as well as the radar wavelength can cause decorrelation and loss of coherence between radar images used in interferometric synthetic aperture radar (InSAR) analysis. This can make it difficult to identify coherent scatterers, reducing the accuracy and precision of the deformation measurements.

In this work, we present two novel approaches to improve the results of StaMPS-SBAS InSAR technique in monitoring vegetated slopes. The first approach is based on optimization of Single Look Complex (SLC) images using a metaheuristic optimization algorithm. In some cases, certain SLC images can lead to a decrease in the number of detected coherence pixels in Interferometric SAR (InSAR) analysis. This can happen due to several factors, including low signal-to-noise ratio, high atmospheric disturbances, and strong decorrelation, caused by vegetation or other factors. To address this issue, an optimization approach is employed to identify the optimal SLC images from a full dataset to increase the number of coherent pixels. To evaluate the effectiveness of the optimization approach, we apply it to a dataset of Sentinel-1 SLC images acquired over the Hollin Hill landslide observatory site in Yorkshire, United Kingdom. We then perform StaMPS-SBAS analysis on the optimized SLC images and compare the results with full dataset. The results show that the optimized SLC images lead to increase the number of reliable coherent pixels, resulting in better estimates of ground deformation.

In the second approach, we present a pixel selection strategy for StaMPS-SBAS processing, which is based on machine learning. Firstly, a set of scatterer candidates are detected via Amplitude Difference Dispersion Index (ADDI) and processed using StaMPS-SBAS and their Temporal Coherence (TC) is estimated. An Artificial Neural Network (ANN) is then trained to predict the TC value of the candidates. Afterward, the trained model is used to predict the TC value of all pixels. Finally, all pixels are categorized as coherent or incoherent based on their TC value. The pixels that are categorized as coherent are then identified as new PS candidates and processed by StaMPS. We apply this strategy to a dataset of Sentinel-1 images acquired over the Hollin Hill landslide and compare its results to the StaMPS pixel selection strategy. Our findings indicate that this approach successfully improves the results of the StaMPS-SBAS technique.

Over the past few years, supervised classification using Deep Neural Networks (DNN) has been used to learn and detect geohazard related InSAR fringes. Most of these networks have been trained using synthetic datasets that do not always represent the true nature of reality. Due to the low occurrence rate of geohazards, there are insufficient datasets or methods to generate training datasets for training DNNs. Data augmentation methods are available to increase the size of the training set but they apply generic transformations using augmentation techniques to pre-extracted training tiles. This may undesirably affect the positioning of features of interest (FOI) in the tiles. Therefore, we identified a need to develop a method that focuses on an FOI, e.g. sinkholes, in the original data space, extracts a subset over the FOI, applies the desired augmentations to the dataset, and finally, downsamples the subset to the tile size. This gives additional flexibility in terms of extracting subsets at various scales.

To address this need, we developed a training data extraction and augmentation method called eXtract using Bounding Box (XBBox). This method takes the extents of an inner (B1) and an outer (B2) bounding box, the size of the tiles and the translation stride parameter as inputs. It calculates all possible combinations of subsets while ensuring a ‘lock’ on B1 which contains the feature of interest and stays within the bounds defined by B2. These subsets are created using augmentations of translation, reflection and, as a novelty of this method, in scale space using SAR multilooking. The method gives the extracted and augmented training tiles as output.

We implemented this method over a sinkhole site in Wink, Texas, USA, where a 500 m wide sinkhole emerged in 2015. It was captured by high resolution TerraSAR-X spotlight SAR datasets of a 0.23 m × 0.94 m resolution. Due to the sinkhole size and the fine spatial resolution of the sensor, sinkhole-related fringes were clearly visible from the InSAR images. Using just two sets of sinkholes-related concentric fringe loops and twelve InSAR epochs, we were able to extract 164,792 training tiles. These were used to train a UNet model for the semantic segmentation of sinkholes. Our method showed excellent convergence with training and validation accuracy of 99.74% and 98.29% respectively.

Future applicability of this method could be diverse. In addition to InSAR fringes, this method could be used to extract training data from amplitude datasets, where features of interest needs to be included in the training tiles.

One of the fundamental assumptions of multi-temporal InSAR is that scatterers remain coherent over the entire analyzed time period. As time series lengthen, there is an increased likelihood of surface changes, and scatterers may only be coherent for part of the time series. We refer to these as Temporary Coherent Scatterers (TCS) (Hu et al., 2019). If we assume presence of Continuously Coherent Scatterers (CCS) in areas that have undergone a surface change during the time series period, the ensemble coherence of these will be low, consequently leading to gaps in the estimates at those locations.

Incorporating TCS time series analysis approach provides an alternative to estimate time series from scatterers that are only coherent for part of the period. The TCS InSAR approach uses the statistical analysis of amplitude time series to detect periods of the presence of the same point scatterer (PS) or distributed scatterer (DS) over consecutive SAR images and does not require any contextual information as input (Hu, et al, 2019). Resultant partitioned time series are consequently unwrapped separately with respect to higher-order continuously coherent reference PS network. The result is an increased number of observation points for displacement monitoring.

The TCS InSAR approach was applied to a project between SkyGeo and the Office of Groundwater Impact Assessment (OGIA), Queensland. OGIA are responsible for the cumulative assessment of groundwater impacts from Coal Seam Gas (GSG) development. A component of this assessment requires OGIA to evaluate the potential for subsidence resulting from resource development and predict how subsidence trends will evolve..

To quantify historical subsidence in the region, SkyGeo processed Sentinel-1 data between 2015 and 2022, using a Persistent Scatterer Interferometry (PSI) approach. Between 2015 and 2022, over 100 new well pads were constructed and began extraction. Using a traditional PSI approach, few or no scatterers were obtained at the new well pads. After applying the TCS InSAR method, we obtain a subset time series as each well pad, once construction is completed.

The results in Queensland demonstrate that TCS can significantly increase the number of observations. 90% of wells constructed during the time period of InSAR processing have PS in the new TCS results, providing additional insights into subsidence trends. Also this results in improved decomposition of the complex, compound subsidence signal over wide areas; ultimately better supporting the mapping of effects of reservoir depletion and prevention of undesirable effects on groundwater.

References

Hu, F., Wu, J., Chang, L., & Hanssen, R. F. (2019). Incorporating temporary coherent scatterers in multi-temporal InSAR using adaptive temporal subsets. IEEE transactions on geoscience and remote sensing, 57(10), 7658-7670.

The intertidal flats characterized by high- and low-tides are transitional buffer zones between land and sea space. They have gently inclined terrains with a very low slope that develop along the coastlines and are exposed occasionally depending on the tide level. They play important roles in providing ecological habitats for various flora and fauna species, protecting coastal residents from storms and floods, and generating huge economic value as tourism. These intertidal flats are easily threatened by frequent erosion and sedimentation processes with anthropogenic impacts like reclamation or embankment construction and natural causes such as climate change or storms. To protect and rehabilitate invaluable intertidal flats, periodic morphological monitoring using remotely sensed images is essential. There are several techniques for extracting the topographic height of the intertidal flats; 1) in-situ terrestrial surveys, 2) airborne or drone LiDAR surveys, 3) waterline extraction with multi-temporal images, and 4) interferometric synthetic aperture radar (InSAR) techniques. In this study, we focus on the construction of a highly accurate digital elevation model (DEM) using space-based synthetic aperture radar observations on the dynamic intertidal flat environment. The InSAR technique using the phase difference between two consecutive SAR images can provide very detailed surface displacement and topographic elevation information. The construction of DEM over intertidal flats using repeat-pass InSAR is somewhat challenging because the intertidal flats are not always exposed due to flow conditions by the tide effects. In addition, the small or moderate geometric baseline in the general InSAR observations mission cannot provide enough height of ambiguity (HoA) to extract the height sensitivity of the low slope regions. The HoA is defined as the height difference corresponding 2 cycle of interferometric phase. It is closely related to phase-to-height sensitivity which is inversely proportional to the perpendicular baseline. To overcome these two obstacles of 1) temporal decorrelation and 2) low HoA, we adopted the bistatic SAR observations with large perpendicular baselines acquired during the TanDEM-X scientific phase. The study area is the German Wadden sea, inscribed as a UNESCO World Heritage Site. We collected two co-registered single-look slant range complex (CoSSC) data with large perpendicular baseline (~1.57 km and ~1.99 km) to compare the height of sensitivity in the intertidal zone. The HoA have been calculated as 8.79 m and 4.37 m, which are much lower than that of the conventional TanDEM-X interferometric pair (30-45 m) and a preferable condition for a low slope area. We calculated differential interferograms to reduce phase aliasing even in a low mountainous topography owing to a large perpendicular baseline with 1-arc SRTM DEM. The validation using ICESat-2 altimeter data with high vertical accuracy of ~10 cm has been conducted and compared with the TanDEM-X global DEM (~90 m spatial resolution) and the SRTM 1-arc DEM (~30 m spatial resolution). Constructed TanDEM-X DEMs (R² > 0.95) and reference DEMs (R² > 0.85) showed great correlations with ICESat-2 altimeter elevation over the inland region. The reference DEMs show very little correlation with altimeter data in the intertidal zone, while constructed TanDEM-X DEMs showed good agreements (R² > 0.7). Note that the DEM with a smaller HoA (~4.37 m) represents much better agreements (~0.92 R²) than the larger HoA (~0.79 R²). It implies that HoA might significantly contribute to the vertical accuracy at the low slope intertidal topography. Our findings suggest that instantaneous InSAR measurement with almost-zero temporal and large perpendicular baselines can successfully construct topographic height on the intertidal flat. Periodic observations with specific flight modes such as the TanDEM-X science phase could be beneficial for monitoring the intertidal zone that is difficult to access.

PS or PS/DS InSAR processing is challenging in areas affected by decorrelation for a part of a year. Due to the fact that causes of decorrelation, such as vegetation and snow cover, are variable in space and time, invalidated images may be different for each PS/DS in the area of interest, leading to spatially variable results, which must be interpreted carefully.

The case of PSInSAR and external information about snow cover is straightforward with regard to image masking, but brings interpretation problems: if a site is sliding down during the summer, what is happening in winter under the snow? Does it move at all, or does it move faster, skipping several ambiguities?

For distributed scatterers in vegetated areas, the problem becomes even more complex. Distributed scatterers may be found based on the amplitude distribution [1] in time and space. Small temporal baseline interferograms are calculated, and phases and coherence are evaluated for each DS, averaged over the DS pixels; for other algorithms, (adaptive) spatial filtering is performed. Coherent interferograms are selected for each DS (or pixel) based on coherence thresholding, or all interferograms are processed (possibly weighted).

However, it is important to stress out that coherence of pure-noise interferograms is non-zero, in the interval of 0.2-0.3, depending on the number of pixels averaged. Our algorithm uses simulated statistics to estimate the coherence threshold to filter out DSs corresponding to pure noise.

In the case of seasonally incoherent DSs, the time series is split into several disconnected segments, making monitoring of more seasons in one time series impossible. The small baseline method [2] sets the displacement velocities between the segments to the lowest possible value, minimizing the optimization criteria. We have developed an approach that interconnects the segments by an approximation of the displacement velocities before and after the excluded interval. Still, none of these approaches may correspond to the real displacement trends in cases of their seasonal variability, e.g due to soil swelling, seasonal variability of soil moisture or cyclic soil freezing and thawing.

The interpretation of time series emerging from spatially filtered interferograms must consider the non-zero (triangular) closures. Before the estimation of displacement rate from image phases, the image phases have to be calculated from interferogram phases, in order to get non-biased results [3]. As the non-zero phase closures are caused (at least partially) by soil moisture variability [4], soil moisture changes contribute to the finally estimated time series of a (filtered) point. This is different from possible soil swelling due to moisture changes (such swelling would not influence phase closures, only displacement noise).

And finally, the interpretation of time series emerging from a method where some interferograms are incoherent or invalidated, must be even more careful: the ambiguity problems mentioned above apply, and the soil moisture influence is even enlarged by the fact that some of the images could not be corrected for soil moisture due to invalidated interferograms. In addition, there are problems of displacement velocity approximation in the invalidated seasons: the approximation was done based on some criteria which do not need to be realistic in the monitored area.

References:

[1] Ferretti, Alessandro, et al. "A new algorithm for processing interferometric data-stacks: SqueeSAR." IEEE transactions on geoscience and remote sensing 49.9 (2011): 3460-3470.

[2] Berardino, Paolo, et al. "A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms." IEEE Transactions on geoscience and remote sensing 40.11 (2002): 2375-2383.

[3] Manunta, Michele, et al. "The parallel SBAS approach for Sentinel-1 interferometric wide swath deformation time-series generation: Algorithm description and products quality assessment." IEEE Transactions on Geoscience and Remote Sensing 57.9 (2019): 6259-6281.

[4] De Zan, Francesco, et al. "A SAR interferometric model for soil moisture." IEEE Transactions on Geoscience and Remote Sensing 52.1 (2013): 418-425.

Floating ice shelves fringe 74% of Antarctica's coastline, providing a direct link between the ice sheet and the surrounding oceans. A better understanding of Antarctic ice shelves and the physical processes affecting them has been the main objective of ESA’s Polar+ Ice Shelves project. The study’s main objective has been the advance in the use satellite observations and modelling to a better understanding of Antarctic ice shelves and the physical processes affecting them. A suite of geophysical products based on Earth Observation datasets from the last decade and modelling has been defined and produced over selected target ice shelves in Antarctica. One of these products, the ice shelf area change, is an important indicator of ice shelf stability in a warming climate, being affected by grounding line retreat as a possible consequence of ice thinning and calving events including ice shelf disintegration or collapse.

An ice shelf is bounded at its seaward margin by the calving front while its inland border to the grounded ice of the Antarctic continent is given by the grounding line. Our calving front location is derived from Cryosat-2 swath elevation, while the grounding line is detected as the upper limit of ice shelf tidal flexure from Sentinel-1 and, prior to 2015, ERS-1/2 interferometric data. Time series of individual grounding lines from Sentinel-1 SAR triplets acquired at various dates within the ocean tide cycle have been processed and averaged over one entire year in order to obtain a gapless yearly grounding line. Eventually, time series of complete ice shelf delineations are obtained from the combination of these two products. It is possible to track absolute and relative area change of an ice shelf and additionally to partition the change into the individual contributions induced by the calving front and grounding-line migration. The annual ice shelf perimeters of the Amery Ice Shelf from 2011 to 2020 is visualized in the attached Figure 1. More similar examples over major ice shelves will be shown at the workshop.

In-orbit test of Lu Tan-1 (LT-1) started at the beginning of 2022 when the first satellite named as LT-1 A was launched at January 26. The second satellite LT-1 B was launched at February 27. The two satellites are especially designed for the interferometric applications, i.e., digital elevation model (DEM) generation and deformation monitoring. The helix bistatic formation (HBF) was established at June and the rainy and cloudy areas covering Easter Sichuan, Western Guizhou, Southern Yunnan, Southern Tibet were the main target regions where the optical satellite failed to collect the ground surface information. In December, LT-1 were converted into the pursuit monostatic formation (PMF) which would be lasted till the end of the satellite constellation life cycle. We would spend months to collect the data over the areas of interests, 30 images were expected to be provided and the deformation accuracy would be assessed using differential interferometric synthetic aperture radar (SAR, InSAR, DInSAR), stacking and mutli-temporal InSAR (MTInSAR) technologies.

Interferometric performance of LT-1 is determined by the eight decoherent components expressed using eight parts, i.e., baseline decoherence, temporal decoherence, signal-to-noise ratio decoherence, volume decoherence, ambiguity decoherence, quantization decoherence, Doppler decoherence, as well as processing decoherence. Most of the decoherence values are similar for both HBF and PMF due to the identical satellite mechanical elements. For example, typical values of the signal-to noise ratio, ambiguity, Doppler and processing decoherence values are better than 0.91, 0.96, 0.98 and 0.96, respectively. But the decoherence parts related to the satellite formation, i.e., baseline, temporal and volume coherence are different. Because in the HBF, the interferometric phase is half of that in the PMF if the other conditions are exactly the same. Temporal decoherence of HBF is considered to be 1 because the signal is accepted by the antenna at the same time. That of the PMF is related to the temporal lags. However, LT-1 maintains good coherence in the city areas even the temporal baseline is longer than half a year. We are about to assess the temporal decoherence in the operational stage after in orbit test.

Determinative coherence of LT-1 is related to the baseline. Critical baseline of HBF is two times more than that of PMF. The stripmap 1 mode is preferred because of the 3 m high resolution. The critical baselines are always longer than 55,285 m if the incidence angle is 35° and the slope angle is 0° for HBF. Under the same circumstance, the critical baseline is 27,642 m for PMF. The other important factor that should be considered is the range resolution. If we use the stripmap 2 mode for deformation monitoring, the critical baseline is one quarter of that of stripmap 1. Therefore, we suggest using stripmap 1 mode to keep high coherence values. We do not assess the HBF interferometric capacity in this paper because the digital elevation model (DEM) have already being processed successfully. Given that the main task in the following 8 years is deformation monitoring, baseline decoherence of PMF is more important.

The recursive orbit control radius (ROCR) is the key factor in PMF to keep coherent for the deformation monitoring task. ROCR is controlled by the space telemetry tracking and command system every week considering the drift of the satellites compared to the predetermined orbit. ROCR of LT-1 is 350 m, the corresponding baseline in the interferometric geometry is less than 700 m. The orbits are controlled even for HBF mode, meaning that the data collected for DEM generation can also be used for deformation monitoring. The interferometric coherence is greater than 0.97. 301 interferograms during the in-orbit test are arbitrarily collected and the perpendicular baseline which is useful to determine the baseline decoherence is provided. The minimum perpendicular baseline is 3.8 m and the maximum is 522.4 m, 90% of the interferograms are less than 396.4 m, if the parallel baseline follows the same distribution, 90% of the the interferometric baseline would be smaller than 555.0 m. However, we paid no attention to the parallel baseline which was of no affects to the deformation monitoring if the proper processing chain was adopted.

The volume decoherence, which is related to the vegetation height, is also determined by the ROCR. The looking angle difference introduces the propagation paths diversity. Volume coherence is a function of height of ambiguity (HoA) as well as the vegetation height. 90% HoA would be greater than 86.2 m in the PMF given that the looking angle ranges from 20 – 46 degrees in the interferometric mode for stripmap 1, the corresponding decoherence would be greater than 0.97 even in the regions where the vegetation height is around 36 m.

The quantization decoherence is assessed using the real data. In this paper, we selected a region covering Qinghai Province. We assessed the quantization ratio of 10:6, 10:4, 10:3 and 10:2, the commonly used one is 10:6. The quantization parameters are injected to the satellite instructions. Then the images with different quantization ratio values are collected and provided from the ground segment to our application system. The coherence values decreased from 0.94 to 0.87, 0.81 and 0.61. If we assessed the phase dispersion using Cramer-Rao bound, the corresponding phase standard deviation would increase from 14.7 to 23.0, 29.3, and 52.6 degrees, leading to the deformation dispersion to 0.52, 0.75, 0.96, and 1.72 cm. Although this was not very universal, the obvious degradation was observed if big quantization ratio was applied. Therefore, se suggest use 10:6 operationally in the first year after satellite is delivery successfully to us.

The interferometric coherence of LT-1 is of good performance to provide InSAR DEM observations and deformation observations. ORCR, which is the basic parameter for interferometric applications, is controlled to be less than 350 m, thus ensuring the basic interferometric coherence. In the following years, we will use the LT-1 data for DInSAR, stacking and MTInSAR technologies to generate deformation field product, deformation velocity product and multi-temporal deformation product, respectively. The products are expected to be useful in the 3,940,000 km² highly and moderately susceptible geohazard areas deformation monitoring in China.

The economy and society in Egypt are highly dependent on the Nile river water. The Grand Ethiopian Renaissance Dam (GERD) construction is expected to reduce Nile water volume inflow in Egypt by 12% to 25%. This will contribute to the current water shortage in Egypt, increasing freshwater demands, groundwater discharge rates, and land subsidence risk. At the same time, this risk is also increased by the steep population growth in recent years in Egypt, which has led to the urbanization of new and larger areas and the relocation of the Nile water to these new sites, such as the Toshka lakes. Therefore, there is an emergent need for a surface deformation monitoring scheme, especially over the Nile Valley, where a dense population and metropoles cities exist. Given the rapid and dynamic changes across the Nile valley, it is crucial to understand the factors contributing to surface deformation to establish a mitigation strategy depending on the analysis of the relationship between surface deformation rates and surface deformation-related factors. In the last three decades, the Interferometric Synthetic Aperture Radar (InSAR) technique has been proven as a well-established technology to monitor land surface deformation with millimeter precision over large areas. Especially with the launch of Sentinel-1a&b SAR satellites in 2014 and 2015, we can obtain SAR data for free, which has global coverage and a short repeat cycle of 6 or 12 days, and develop surface deformation monitoring system at local and regional scales, and with high spatio-temporal resolution.

In this research, we present the preliminary results of a prototype system that uses Sentinel-1 SAR data characterized by VV polarization, with ascending orbital direction and acquired over the years from 2017 to 2021, and open-source GMTSAR tools to monitor the surface deformation rates from InSAR and associate them with possible causative factors. Particularly, we applied a Small Baseline Subset (SBAS) time series InSAR approach to monitoring surface deformation over a large area of the Nile Valley, starting from Aswan to Toshka, Egypt, as a case study. The study area covers 54107.2 km2. Then, the deformation obtained with the present methodology were analyzed against the data available of a different factor of influence of surface deformation (e.g., rainfall, water body change, total terrestrial water storage, land use-landcover, temperature, etc.) to understand their relations and their impact. By linking the surface deformation to the causative factor, we aim to understand the system dynamics better. This can be utilized by the decision-makers so that they can take into account the surface deformation risk due to the change of the Nile water and quantity during the regional planning, especially over the Aswan-Toshka area.

SAR is different from other sensors in that it can acquire complex images that contain not only amplitude information but also phase information. The phase information of SAR images is extremely sensitive to changes, so it can be well applied to the measurement of sub-wavelength changes. The method adopting phase information to detect potential changes in the scene is called coherent change detection (CCD). However, the relationship between the coherence of typical objects and SAR frequency has not been fully studied. As a result, the application of CCD in various fields has not yet been fully explored. The scattering mechanism of the target under SAR radiation is very complicated; different types of targets have different scattering types under the radiation of different SAR frequencies. Therefore, it is more than significant to choose an appropriate frequency to observe the changed area. Choosing an appropriate frequency to observe the changed area is conducive to reliably detecting the changes of interest in the scene. On the contrary, using an inappropriate frequency for observation will result in a high false-alarm rate, a poor detection rate and unreliable detection results.

This paper focuses on the relationship between the coherence of typical objects and SAR frequency. A large number of experiments have been carried out and effective experimental data have been obtained with the DVD-InSAR system developed by the Aerospace Information Institute, Chinese Academy of Sciences, which can observe the same scene at six frequencies simultaneously. Combining all six or more frequencies into one airborne SAR system is unprecedented. This study will make it possible for researchers to compare the radar backscatter characteristics and study coherence characteristics across frequencies simultaneously. The relationship between the coherence of different typical objects and SAR frequency is analyzed in detail in this paper.

The DVD-InSAR system has multiple working modes, including strip-map, spotlight, cross-track and along-track interferometry modes. The P, L, S, C, X and Ka bands SAR subsystems share a set of positioning and orientation systems (POS) and have the same timing source. These six-band SAR systems can work at the same time and acquire SAR images of the same scene simultaneously.

The temporal decorrelation of the targets characterizes their mechanical and dielectric stability. In order to analyze the relationship between the temporal decorrelation and SAR frequency of the selected study area, we chose the repeat-pass interferometry observation mode of the DVD-InSAR system to obtain an experimental dataset. Multiple flights were conducted in the selected study area with the DVD-InSAR system. In order to fully analyze the coherence characteristics, sufficient samples of different typical objects were first selected from the coherence map of the study area. The typical objects mainly included building, vegetation, bare land, road, railway and water regions.

In this paper, analysis of multi-frequency interferometric coherence characteristics of typical objects for coherent change detection is presented. We discuss the method for multi-frequency interferometric processing, and presents the experimental results and analysis of the work.

This research was supported by the National Natural Science Foundation of China (No. 62231024).

Campi Flegrei is a volcanic caldera located in Southern Italy, west of the city of Naples, well known by the scientific community because of the very high volcanic risk associated. It is indeed a highly urbanized area undergoing periodic phases of unrest, causing inflation or deflation with ground deformation rates up to several mm/month and other related effects such as shallow depth seismic swarms, soil temperature variations and degassing in the center of caldera, mainly in the Solfatara-Pisciarelli volcanic district. The ground displacement, known as the Campi Flegrei bradyseism, has been also mapped by archaeological records. It is directly connected to the volcanic activity and can be exploited to retrieve information about the source geometry and its depth, thus providing important indications for hazard assessment and risk mitigation purposes.

This work provides the mean ground deformation rates and ground displacement time series of the Campi Flegrei caldera (Italy) retrieved by satellite remote sensing data analysis from 1992 to 2021.

Synthetic Aperture Radar (SAR) images acquired by ERS 1-2 (1992-2002), ENVISAT (2003-2011) and COSMO-SkyMed (2011-2021) are processed by multi-temporal SAR Interferometry (InSAR) approach using the same technique, parameters, and reference system, to obtain for the first time a homogeneous and time-continuous dataset.

The multi-temporal InSAR approach allowed us to obtain a very huge number of point targets with good coherence, and thus to detect ground deformations of the caldera with dense spatial coverage. Since 1992, with the launch of the first space mission equipped with a SAR sensor operating for many years, InSAR data have been largely applied in the study of Campi Flegrei, with particular focus on the intense inflation phase started in 2011 and still ongoing, with about 100 cm to date in the maximum deformation area, located in the town of Pozzuoli along the coastline.

As a last step of our analysis, we carried out a validation of the InSAR products by comparison with the measurements provided by precise levelling technique and cGNSS stations. These ground-based techniques provide precise information about the Campi Flegrei surface deformations, but only in a limited number of measuring points. From the levelling technique, the altitude of the benchmarks along levelling lines, constraining the vertical displacement in the time interval between two measurement campaigns, has been retrieved. In addition, the cGNSS technique provides measurements with high temporal sampling of deformation along the 3D displacement component, i.e. North-South (N-S), East-West (E-W) and Vertical (UP).

To conclude, our outcomes from InSAR data processing offer an overview on the temporal behaviour of ground deformations at Campi Flegrei along an unprecedented time window of about 30 years. The datasets are open access and compliant with FAIR principles, so they can be exploited by the scientific community for supporting and improving the knowledge of the dynamics of the caldera.

The TanDEM-X mission acquires data with two satellites flying in bistatic formation for Digital Elevation Model (DEM) generation since more than ten years. The collected data from the years 2010 to 2015 was used for the generation of the first global TanDEM-X DEM, which includes multiple acquisitions for the whole Earth. Since then enough data for a second global DEM, the TanDEM-X DEM 2020 [1], was acquired with at least one or even multiple acquisitions depending on the area. This dataset was acquired between 2017 and 2022. Since then additional acquisitions are conducted. Altogether, the TanDEM-X DEM acquisitions which yield a unique multitemporal data set.

The data acquired for the TanDEM-X DEM 2020 is processed to so-called CRaw DEM scenes by the Integrated TanDEM-X Processor (ITP) [2,3]. Additional to the generation of the second global DEMs, these CRaw DEM scenes are used for the generation of TanDEM-X DEM Change Maps [4]. These represent the differences between mosaics of the CRaw-DEM scenes and an edited version of the first global TanDEM X DEM. These DEM Change Maps already show a broad variety of applications for change indications in different areas and land covers all over the Earth. The possible applications contain mining areas, deforestation, glaciers and many more.

To go even further, not only the CRaw DEM scenes, but all TanDEM-X DEM data can be exploited for the generation of stacks of DEM changes. In contrast to the DEM Change Maps, which give the difference of one discrete point in time to a time span, the stacks provide change information between multiple specific points in time. This also allows the calculation of change velocities.

These multitemporal DEM change stacks can give information about the temporal DEM height development over a timespan up to 13 years. The number of usable acquisitions varies for different areas. Over Iceland this number goes up to almost ten acquisitions over the glaciers. The Patagonian Ice field is also covered by partially more than five acquisitions. Long-time monitoring of glacier regions and their changes is crucial, especially in the context of climate change research.

The DEM Change Maps and Stacks of DEM Change Maps show a dramatic ice loss in Iceland and Patagonia over the last decade. However, different acquisition dates and especially acquisition seasons show the need for an additional quantitative study with a more precise choice of data and indicate a need for taking the different penetration depths in different seasons into account.

Even though the current version of the TanDEM-X DEM Change Maps stacks does not claim to give an exact measurement of DEM changes i.e. ice loss, it gives a great starting point for these global measurements in the future and already a qualitatively measurement over large areas.

References

[1] B. Wessel et al., "The new TanDEM-X DEM 2020: generation and specifications," EUSAR 2022; 14th European Conference on Synthetic Aperture Radar, Leipzig, Germany, 2022, pp. 1-5.

[2] T. Fritz, C. Rossi, N. Yague-Martinez, F. Rodriguez-Gonzalez, M. Lachaise, and H. Breit, “Interferometric processing of TanDEM-X data,” in 2011 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2011, pp. 2428–2431.

[3] M. Lachaise and T. Fritz, “Update of the Interferometric Processing Algorithms for the TanDEM-X high resolution DEMs,”in EUSAR 2016: 11th European Conference on Synthetic Aperture Radar. VDE, 2016, pp. 1–4.

[4] M. Lachaise, C. Gonzalez, P. Rizzoli, B. Schweisshelm, and M. Zink, “’The new TanDEM-D DEM Change Maps Product’,” in ´2022 IEEE International Geoscience and Remote Sensing Symposium IGARSS. IEEE, 2022

Coastal Erosion from Space is a project funded by ESA and its primary objective is to determine the feasibility of using a range of satellite images (both optical and SAR) to monitor coastal changes, as well as to collect Coastal State Indicators (CSI) to describe coastal dynamics and evolution. The objective of this project is to develop a global service for monitoring coastal erosion, assessing environmental risks, and assessing the potential impacts of climate change on coastlines.

As a result of this activity, ISARDSAT has developed a processing chain for generating coastal change products using Sentinel 1 data, even if it can be applied to other SAR missions. As a result of Sentinel 1, which operates regardless of the weather conditions and sunlight, we are able to monitor coastal evolution using hundreds of freely accessible satellite data under the Copernicus programme that provides extremely high spatial (10mx 10m) and temporal (6 days revisit time) resolution. In contrast with optical images, which are unusable for this type of application when the Area of Interest (AOI) is even partially obscured by clouds, SAR technologies provide a significant advantage.

There are four main processes in the methodology:

Firstly, a georeferenced image is generated for each available S1 data, separately for ascending and descending tracks. This process, also known as pre-processing, is composed of several sub-steps that have been developed in the SNAP toolbox provided by ESA.

The second process (which consists of Enhancement, Segmentation, Healing, and Vectorization) produces a vector line, called a waterline (WL), which represents the boundary between land and water. An input configuration file can specify a set of parameters for configuring these sub-steps. This process aims to improve the quality of the output. It is achieved by reducing as much as possible the erroneous features that may appear in the initial estimation of the waterline.

Following this, two parameters are computed for each WL as part of the process known as "Quality control":

• The distance x_i between each point on the WL and a reference line.
• Line density (Heatmap).

As a final step, taking into account all the WLs and their distances from the reference line, the change rate product is calculated to illustrate the evolution of the coast under analysis over time (erosion or accretion). To accomplish this, a series of polygons have been drawn along the reference line. A change rate product is calculated for each polygon, taking into account only the WLs and their distances included in the polygon, which is defined by a width w and a length l across the reference line. A second filtering step is applied in order to eliminate possible outliers: the distances in each polygon are described statistically using a Gaussian Mixture Distribution (GMD) with k components using information derived from the Heatmap. For the purpose of filtering, the mean μ and standard deviation σ of the distances belonging to the component with the largest population are computed. In order to calculate the change rate product for that polygon, only distance values that meet the criteria |x_i-μ|≤σ are used.

After the second filtering, the remaining WLs distances are used to perform a linear regression analysis. The change rate product is defined as the slope of the linear relationship fitting the available data in this analysis. A polygon's slope indicates whether erosion has occurred (negative inclination) or accretion has occurred (positive inclination). The tool can be tuned according to the end user's request, and it is possible to provide the change rate in various ways:

• Time sampling (monthly, annual, etc.).
• Sampling of space (appropriately defining polygon widths).

Despite the fact that SAR images cannot directly be compared with optical images since they may be affected by speckle noise and geometry artifacts, the water lines produced by IsardSAT provide trends over time that are associated with seasonal events.

The tragic collapse of the Champlain South Condominium Tower in Surfside, Florida motivated the examining of the building’s stability and coastal subsidence using InSAR. The 2016-2021 Sentinel-1 InSAR data of the city of Surfside in Miami Beach, FL, reveals several subsidence hotspots with subsidence rate of up to 1 cm/yr velocity in the radar line-of-sight (LOS) direction (corresponding to 1.4 cm/yr vertical velocity). The subsidence is centered in newly constructed high-rise buildings that suggests the construction could have been the causative factor. Two major subsidence hotspots are: (1) Surf Club hotel and (2) Oceana residences. For the Surf Club hotel, the temporal correlation of subsidence with nearby construction activity indicates that the subsidence could have been related to the construction of the foundation. For the Oceana, the differential displacement of 3.5 mm/yr has not stopped by 2023. Using the geotechnical reports for these buildings and the history of soil’s condition before construction, we can compare the major differences between these sites that could have caused the diversity in the InSAR signal. We also aim to model the consolidation and secondary compression (creep) of South Florida’s young limestone under building loads and other construction activities such as pile-driving to understand the observed patten of subsidence. Defining the causes of subsidence in the South Florida’s coral rocks is important for the mitigation of possible hazards and providing better guidelines for future construction projects.

As a part of the Observational Products for End-Users from Remote Sensing Analysis (OPERA) project, NASA has tasked the Jet Propulsion Laboratory, California Institute of Technology to produce high-resolution (< 30m) line-of-sight (LOS) land-surface Displacement Products (DISP) over North America from Sentinel-1 and NISAR SAR data (see https://www.jpl.nasa.gov/go/opera for more detailed product information). In our research work, we focus on subsequent higher-level processing using the OPERA DISP product as an input source for generating decomposed quasi-horizontal and vertical displacements. To realize this, relative high-resolution LOS InSAR displacements need to be re-referenced and projected to a geodetic reference frame. This is commonly done by referencing InSAR with GNSS observations, and decomposing LOS displacement vectors into North-South, East-West, and Up-Down directions with defined apriori assumptions (e.g. negligible horizontal or North-South motion, or using models to constrain certain displacement components). However, it becomes challenging to perform these tasks at large scales due to multiple tracks of relative InSAR observations with different imaging geometries and noise levels, as well as various non-linear and long-wavelength ground motion signals.

Here we present a scalable approach to derive quasi-vertical land motion from relative LOS InSAR observations over large-scale areas, with a focus on SAR observations and ground motion settings over North America. The approach consists of two steps: 1. re-referencing InSAR displacement rates with a GNSS model projected in LOS, and 2. LOS decomposition with support of external ground motion data/models to solve the undetermined equations. Re-referencing is performed by estimating a surface between low-pass filtered InSAR displacement rates and a coarse GNSS velocity model (50 x 50 km), thereby constraining the short-wavelength and long-wavelength displacement signals with InSAR and GNSS, respectively. After re-referencing, we apply pixel-wise LOS decomposition of InSAR observations with additional external data (e.g. GNSS) providing horizontal ground motion. If InSAR displacements are available from both viewing SAR geometries, i.e spatially overlapping ascending and descending tracks, only external North-South ground motion is added to solve the rank deficiency. Measurement and model uncertainties are propagated to the final result, as associated product quality metrics.

We demonstrate our approach on multiple case studies within the North American scope, that cover most of the expected scenarios in terms of satellite SAR acquisition plan, land cover, and ground motion. In preparation for the release of OPERA DISP product, we leveraged JPL's Advanced Rapid Imaging and Analysis (ARIA) open-access archive of Sentinel-1 Geocoded Unwrapped interferograms (S1-GUNWs, 90m-posting) to produce InSAR time series over the large-scale case studies. We applied additional corrections to the InSAR time series by utilizing the ARIA S1-GUNW correction layers for solid-earth tides, and ionospheric and tropospheric phase delays embedded in the product.

In the last few decades, InSAR has been used to identify ground deformation related to slope instability and to retrieve time series of landslide displacements. In some cases, retrospective retrieval of time series revealed acceleration patterns precursory to failure. Although the higher temporal and spatial resolution of new-generation satellites may offer the opportunity to detect precursory motion with viable lead time, to rely entirely on the possibility of retrieving continuous time series of displacements over landslides is a limiting strategy. This is because successful phase unwrapping is impaired by factors such as unfavourable orientation, landcover and high deformation gradients over relatively small areas, all common on landslides.

We generated and analysed 112 Sentinel-1 interferograms, covering the period between April 2015 and June 2020, at medium spatial resolution (8 and 2 looks in range and azimuth respectively) over the Achoma landslide in the Colca valley, Peru. This large, deep-seated landslide, covering an area of about 40 hectares, previously unidentified, failed catastrophically on 18^th June 2020, damming the Rio Colca and giving origin to a lake. We explored a methodology to retrieve precursory signs of destabilisation of landslides with characteristics unfavourable to unwrapping and time series inversion based on the investigation of spatial and temporal patterns of coherence loss within the landslide and in the surrounding area and on the extraction of a relative measure of incremental displacements through time obtained from the wrapped phase.

We observed significant, local interferometric coherence loss outlining the scarp and the southeastern flank of the landslide, intermittently in the years before failure. Moreover, we observe a sharp decrease in the ratio between the coherence within the landslide and in the surrounding area, roughly six months before the failure which is interpreted as a sign of critical landslide activity and a precursor. The wrapped interferometric phase also revealed a sequence of acceleration phases, each characterised by increasing annual rates. We observe a behaviour that recalls progressive failure, with no clear evidence for response to one particular trigger and two acceleration phases followed by a more stable period and the last leading to failure.

This type of approach is promising with respect to the extraction of relevant information from interferometric data when the generation of accurate and continuous time series of displacements is hindered by the nature of landcover or of the landslide studied, such in the case of the Achoma landslide. The combination of key, relevant parameters and their changes through time obtained with this methodology may prove necessary for the identification of precursors over a wider range of landslides than with InSAR time series generation alone.

Lava flows deform even after the mechanical flux stops. During the post-emplacement phase, there are several physical processes that are responsible for these phenomena. In the initial stages after deposition, degasification may cause a cooling lava body to rapidly expand [1]. Crust sinking and lava tube collapse [2] might produce rapid movements that can occur since lava deposition. Poroelastic deformation or viscoelastic relaxation of the substrate caused by the lava flow gravity load can produce downward surface movement [3,4]. Horizontal continuous displacements have also been detected by residual shearing of the lava on the flank [5]. Thermal cooling of lavas produces contraction and consolidation, being the main driving mechanism of surface subsidence in lava fields and in correlation to lava thickness [6]. InSAR represents a valuable tool to monitor lava fields deformation, as coherence is well preserved in time and allows to retrieve information in inaccessible areas. Modelling the physical mechanisms allows to differentiate the potential causes of the observed displacements.

The most recent eruption in the western flank Cumbre Vieja Volcano (La Palma, Spain) lasted for 85 days, from the 19^th of September to the 13^th of December 2021 [7]. It was a fissure strombolian eruption with phreatomagmatic pulses which emitted an estimated volume of more than 200Mm³ of volcanic materials and emplacing a lava field that covered more than 12 km². The lava flows followed an East to West direction, reaching the sea and forming two lava deltas. Lava composition is mostly basaltic (basanite and tephrite) and the type of lava flows is largely a'ā. The lava field covered 1,676 edifications, 37 km² of agricultural lands and affected 73,805 km of roads, blocking the transit between the NW to the SW part of the island. Reconstruction works started soon after the end of the eruption and a provisional trail was habilitated for traffic in August 2022 crossing the lava field. The government intends to declare part of the lava fields a geological heritage protected area, but there is a great interest and funding resources to start the reconstruction of roads and other infrastructures.

In this work we present and discuss the preliminary InSAR deformation results of post-emplaced lavas in La Palma. We have processed 33 ascending and 36 descending orbit Sentinel-1A SAR images covering the entire island from the end of eruption (mid-December) to February 2023. We used the software SNAP and StaMPS with a Single Reference approach and a linear tropospheric correction using TRAIN. Our preliminary results show a clear deformation pattern within the lava field borders, with LOS rates up to 23 cm/year and 32 cm/year in ascending and descending orbit respectively. The LOS velocity standard deviation of PS outside the lava field is high (~2cm/year) which highlights the strong turbulent atmospheric contribution in the island. PS density within the lava field is around 400 PS/km². Next steps will consist of refining the InSAR processing by adopting a SBAS approach with short time baselines, decompose the ascending and descending geometries into vertical and horizontal displacements and examine the relation between lava thickness and deformation. Our final goal is to investigate the physical mechanisms producing deformation, which will provide useful data for the reconstruction.

[1] Peck, D. L. (1978). Cooling and vesiculation of Alae lava lake, Hawaii (No. 935-B). US Govt. Print. Off. doi:10.3133/pp935B

[2] Borgia, Andrea, et al. "Dynamics of lava flow fronts, Arenal volcano, Costa Rica." Journal of volcanology and geothermal research 19.3-4 (1983): 303-329. doi:10.1080/01431160051060246

[3] Stevens et al. (2001). Post-emplacement lava subsidence and the accuracy of ERS InSAR digital elevation models of volcanoes. International Journal of Remote Sensing, 22(5), 819-828.

[4] Lu, Z. et al. (2005). Interferometric synthetic aperture radar study of Okmok volcano, Alaska, 1992–2003: Magma supply dynamics and postemplacement lava flow deformation. Journal of Geophysical Research: Solid Earth, 110(B2). doi: 10.1029/2004JB003148

[5] Carrara, A. et al. (2019). Post-emplacement dynamics of andesitic lava flows at Volcán de Colima, Mexico, revealed by radar and optical remote sensing data. Journal of Volcanology and Geothermal Research, 381, 1-15. doi: 10.1016/j.jvolgeores.2019.05.019

[6] Ebmeier, S. et al. (2012). Measuring large topographic change with InSAR: Lava thicknesses, extrusion rate and subsidence rate at Santiaguito volcano, Guatemala. Earth and Planetary Science Letters, 335, 216-225, doi:10.1016/j.epsl.2012.04.027

[7] González P.J., (2022) Volcano-tectonic control of Cumbre Vieja. Science, 375(6587), 1348-1349, doi:10.1126/science.abn5148

The European Ground Motion Service (EGMS) is the first-ever service to provide pan-European ground motion data, fully free and available to everyone. It is based on full-resolution Sentinel-1 imagery and can be used for monitoring the deformation of infrastructure as well as geohazards such as landslides, volcanoes and mining effects.

The EGMS is a new addition to the Copernicus Land Monitoring Service (CLMS) portfolio and is implemented by the European Environment Agency.

The scope of this presentation is to provide an update of the production, validation, and user uptake activities.

The EGMS provides three product levels: Basic and Calibrated, which are Line-of-Sight (LoS) measurements, and Ortho, in which a decomposition of all Calibrated measurements yield the vertical and East-West motion components.

The first product release took place in May 2022 and was based on imagery from the period 2015 – 2020. The first annual updates were published early and mid-2023 and were based on 2015 – 2021 and 2015 – 2022 imagery, respectively. The first update alone contained approximately 10 billion measurement points, provided in roughly 15,400 deliverables for the Basic and Calibrated products and 1,600 deliverables for the Ortho product.

Validation is performed independently from production. The goals are to a) verify the usability of the data with respect to the expected applications and b) perform a quality assessment of the products relative to the requirements. This is done through seven activities such as comparisons with GNSS and in-situ data, landslide inventories, and other ground motion services. The activities are carried out over approximately 50 locations in 16 countries with e.g., different climates, topographies, and ground motion phenomena.

Finally, we will share insights into EGMS user uptake activities. The first-time provision of free and open, wide-area deformation maps yields numerous application potentials, largely relevant for new and non-expert users. Hence, great efforts are put into reaching those users, e.g. via webinars and bilateral, national-level meetings with public and private entities from member states. Here, we wish to present an overview of our efforts and the first results from fostering the uptake amongst new and non-expert users.

The EGMS data can be viewed and downloaded from the EGMS Explorer (https://egms.land.copernicus.eu/), while all supporting material is available here: https://land.copernicus.eu/pan-european/european-ground-motion-service.

Deep-Seated Gravitational Slope Deformations (DSGSD) comprise a collection of slow and complex deformational processes driven by gravity, which involve entire slopes over long time intervals [1]. These phenomena occur in various morpho-structural conditions and are characterized by typical morphological features such as double ridges, ridge-top depressions, trenches, scarps, counterscarps, and tension cracks, generally distributed along the entire ridge-slope-valley floor system. Although DSGSD rarely claim lives, they can cause significant damage to infrastructures and sometimes fail catastrophically [2].

The Pisciotta DSGSD represents a noteworthy example. Located along the coast of the Tyrrhenian Sea in the south of Italy, the DSGSD has been known since the 1960s. Its westward movement towards the Fiumicello riverbed manifested from the second half of the eighties [3], with mean rates of approximately 1m/year. Significant movements affected the SS447 road, connecting the Ascea and Pisciotta municipalities and crossing the DSGSD mass at its middle height, which suffered continuous planimetric and altimetric distortions, cracking and bulging of the pavement, and tilting of guardrails and retaining walls. The progressive sliding also affected the Salerno-Reggio Calabria railway tunnel, running on two distinct sediments and crossing the Fiumicello torrent.

The kinematics, spatial extent, and temporal behavior of the Pisciotta DSGSD were partly investigated by a few studies [3]–[5]. Therefore, we collected and analyzed data of different nature to assess the long and short-term spatial and temporal behavior of the Pisciotta DSGSD and its interaction with nearby infrastructures. We first collected geomorphological information such as structural data, high-resolution orthomosaics, and Digital Surface Models (DSM) employing Drone investigations. We then exploited high-resolution optical imagery and Synthetic Aperture Radar (SAR) satellite data from the Sentinel-1 satellite mission to assess the long- and short-term kinematics of the DSGSD body. Optical data from 1943 to 2022 were exploited by means of digital stereoscopy and Digital Image Correlation (DIC) analysis. SAR data were processed through the Small Baseline Subset (SBAS) multi-temporal method of Differential SAR Interferometry [6] to obtain ground displacement maps and displacement time series from September 2016 to October 2021. The interpretation of such data has been assisted by ancillary information consisting of topographic maps at different scales, airborne Lidar data, and ground-based measurements such as rainfall data, boreholes, and inclinometric measurements. All these data were exploited by analytical approaches to provide the best estimate of the DSGSD failure surface(s) and volume and assess its current kinematics.

All these data and analyses fully described the long- and short-term DSGSD evolution and kinematics. The in-situ surveys and the morphological analysis of historical aerial images allow inferring the onset of the DSGSD movement at approximately the middle of the second quarter of the twentieth century. The causes of the triggering of the movement are ascribable to the progressive weathering of flyschoid rocks with interbedded clay-rich layers composing the DSGSD mass, which produced a progressive movement of the slope towards the Fiumicello torrent, often accelerated by strong rainfall events. River erosion is excluded since the DSGSD is very close to the Fiumicello mouth, as well as anthropogenic forcings can be excluded since the even railway line was built before the onset of the slope movement approximately in 1889, while the odd railway track was built between 1955 and 1960 when the slope movement was still active.

From then on, we identify a first period during which the DSGSD experienced a gradual increase in displacement rate as observed by the analysis of the deformations suffered by the SR447 road. During this stage, the DSGSD expanded mainly to the southwest and developed several discrete structures, such as primary and secondary scarps, counterscarps, and linear cracks with strike-slip kinematics. The DSGSD reached maximum displacement rates in the 2006-2011 period, with mean horizontal displacement rates up to 150 cm/y as testified by inclinometric measurements performed at the end of 2009, but without undergoing a rapid collapse. Instead, the progressive stress redistribution and change of relief energy caused a gradual decrease in the displacement rate from 2006 to 2022, as testified by DIC-derived horizontal displacements, vertical displacements computed from height difference of the available Digital Elevation Models (DEM) between 1990 and 2021, and InSAR-derived vertical and horizontal (E-W) displacement rates. If such a trend is confirmed, we should expect a gradual decrease in the displacement rate until the DSGSD can eventually stop.

From a spatial point of view, the observed vertical and horizontal displacement patterns are often associated with rotational sliding. Still, translational sliding can also produce similar patterns when the slip surface is less inclined than the slope. In the latter case, the apparent vertical collapse at the landslide head relates to the opening of the landslide trench, while the uplift at the toe results from lateral slope motion. Our case is in between. The DSGSD head is affected by vertical movements, probably caused by rotational sliding. Otherwise, the uplift measured at the toe should correspond to the prevalent horizontal motion of the DSGSD. Therefore, we argue that the slope moves mainly along a roto-translational deep detachment, with several secondary shallow discrete surfaces acting as secondary detachments, as testified by inclinometric measurements.

To quantitatively understand the DSGSD behavior and its potential effects on the adjacent infrastructures, we interpreted the observed displacements through analytical approaches to reconstruct the DSGSD deep basal shear surface and volume, according to the procedure proposed by Prajapati and Jaboyedoff [7]. The obtained basal shear surface shows that the DSGSD mass reaches a maximum thickness of approximately 85 m and a volume of roughly 6.2x10⁶ m³, which is consistent with surface area-volume empirical estimates from the literature [8], [9]. Furthermore, an apparent interference is observed with the odd railway tunnel, which intercepts the DSGSD toe for approximately 60-80 meters.

References

[1] M. E. Discenza e C. Esposito, «State-of-Art and Remarks on Some Open Questions About Dsgsds: Hints from a Review of the Scientific Literature on Related Topics», Italian Journal of Engineering Geology and Environment, vol. 1, 2021, doi: 10.2139/ssrn.3935750.

[2] P. Lacroix, A. L. Handwerger, e G. Bièvre, «Life and death of slow-moving landslides», Nature Reviews Earth and Environment, vol. 1, fasc. 8, pp. 404–419, 2020, doi: 10.1038/s43017-020-0072-8.

[3] P. De Vita, M. T. Carratù, G. L. Barbera, e S. Santoro, «Kinematics and geological constraints of the slow-moving Pisciotta rock slide (southern Italy)», Geomorphology, vol. 201, pp. 415–429, nov. 2013, doi: 10.1016/J.GEOMORPH.2013.07.015.

[4] P. De Vita, D. Cusano, e G. La Barbera, «Complex Rainfall-Driven Kinematics of the Slow-Moving Pisciotta Rock-Slide (Cilento, Southern Italy)», in Advancing Culture of Living with Landslides, M. Mikoš, N. Casagli, Y. Yin, e K. Sassa, A c. di Cham: Springer International Publishing, 2017, pp. 547–556. doi: 10.1007/978-3-319-53485-5_64.

[5] M. Barbarella, M. Fiani, e A. Lugli, «Landslide monitoring using multitemporal terrestrial laser scanning for ground displacement analysis», Geomatics, Natural Hazards and Risk, vol. 6, fasc. 5–7, pp. 398–418, lug. 2015, doi: 10.1080/19475705.2013.863808.

[6] P. Berardino, G. Fornaro, R. Lanari, e E. Sansosti, «A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms», IEEE Transactions on Geoscience and Remote Sensing, vol. 40, fasc. 11, pp. 2375–2383, nov. 2002, doi: 10.1109/TGRS.2002.803792.

[7] G. Prajapati e M. Jaboyedoff, «Method to estimate the initial landslide failure surface and volumes using grid points and spline curves in MATLAB», Landslides, vol. 19, fasc. 12, pp. 2997–3008, dic. 2022, doi: 10.1007/s10346-022-01940-5.

[8] F. Guzzetti, F. Ardizzone, M. Cardinali, M. Rossi, e D. Valigi, «Landslide volumes and landslide mobilization rates in Umbria, central Italy», Earth and Planetary Science Letters, vol. 279, fasc. 3–4, pp. 222–229, mar. 2009, doi: 10.1016/J.EPSL.2009.01.005.

[9] M. Jaboyedoff, D. Carrea, M.-H. Derron, T. Oppikofer, I. M. Penna, e B. Rudaz, «A review of methods used to estimate initial landslide failure surface depths and volumes», Engineering Geology, vol. 267, p. 105478, mar. 2020, doi: 10.1016/j.enggeo.2020.105478.

Volcanic eruptions can damage or destroy surrounding forest, with the potential to alter its characteristics in the long term. The impact of eruptions on forest has not been systematically studied with satellite data, although individual studies have demonstrated that explosive eruptions in particular produce an impact measureable from satellites. The impact of an eruption and the rate of forest recovery both depend on eruption characteristics, such as temperature, volume and spatial distribution of ejected material, as well as the ecological setting. Here, we explore the use of radar and optical satellite data from Sentinel-1, Sentinel-2 and Landsat 8, to study the forest impact and recovery following two volcanic eruptions: the 2015 eruption of Calbuco volcano and the 2008 eruption of Chaiten volcano.

The nature of damage to vegetation caused by a volcanic eruption depends on the eruption style, magnitude and duration. Large explosive eruptions cause intense damage in the near-field through mechanisms including pyroclastic density currents and lahars, while more extensive but less destructive impacts are caused by distal tephra fall deposits. The most recent eruptions of Calbuco and Chaiten provide examples of such processes. The 2015 eruption of Calbuco started on the 22^nd of April and consisted of three explosive episodes between the 22^nd and 23^rd of April producing large buoyant ash plumes, pyroclastic flows and lahars. These damaged the temperate broadleaf forests around Calbuco up to 15 km away from the eruption centre. We use Sentinel-1, Sentinel-2 and Landsat 8 imagery that spans the eruption onset and recovery period to identify the satellite signature of forest damage and how this signature changes with time. The 2008 eruption of Chaiten began in May and continued for the next three years, producing pyroclastic flows, lahars and an ash plume. In particular, the tephra fall damaged the surrounding temperate broadleaf forest. We use this case study primarily to study the recovery of the surrounding forest.

A drop in the normalised difference vegetation index (NDVI) value is detected in both the Landsat 8 and Sentinel-2 imagery, which correlates with areas of both flow deposits and ash fall. In the NDVI some areas show steady recovery, although the most damaged areas have not yet returned to pre-eruption values. In the Sentinel-1 backscatter data, which is not restricted by cloud coverage, there is an initial increase in the backscatter following the eruption, and areas of flow deposits are clearly identifiable and yet to return to pre-eruption values. In the Sentinel-1 coherence data there is an initial drop in coherence immediately after the eruption, followed by an increase in coherence particularly in areas of flow deposits. We will develop approaches to track the impact of volcanic eruptions on forests with remote sensing data that can be applied globally using freely available data, in different ecosystems and for different styles of eruption. Our eventual aim is to develop a toolkit for identifying the footprint of past volcanic eruptions on forested environments.

The rate at which the Antarctic Ice Sheet flows from the interior of the continent into the ocean is a key indicator of its stability. When the ice enters the ocean it contributes to sea level rise, and satellite observations show that ice loss is currently trending at rates which match the worst-case scenarios in the IPCC’s Fifth Assessment Report. Ice loss in Antarctica is dominated by dynamic imbalance, where the ice accelerates and subsequently thins, and along with grounding line retreat this has been recorded in the Amundsen Sea Embayment of West Antarctica since the 1940’s.

Ice velocity observations can be used in conjunction with measurements of ice thickness and surface mass balance to determine ice sheet mass balance, the measure of the ice sheet’s net gain or loss of ice. Quantifying mass loss is essential as the ice sheet contribution to the global sea level budget remains the greatest uncertainty in future projections of sea level rise. Both long term and emerging signals must be accurately measured to better understand how the Antarctic Ice Sheet will change in the future, with consistent records from satellite platforms required to separate natural variability from anthropogenic signals. The Sentinel-1 constellation is the most recent in a series of C-band SAR platforms to observe Antarctica, allowing for the construction of a record of ice velocity observations from the early 1990s to the present day.

We present measurements of speed change of outlet glaciers in the Amundsen Sea Embayment of Antarctica, covering the whole operational period of Sentinel 1, from 2014 onwards. Velocities are determined through intensity feature tracking 6 and 12 day pairs of Level 1 Interferometric Wide swath mode Single Look Complex images from both Sentinel-1A and 1B satellites. Intensity feature tracking is performed using patch intensity cross-correlation optimization to derive displacement estimates and associated errors. The data are filtered and then posted at 100m on a common grid before a Bayesian smoother is applied to the time series for each grid cell. We present maps of ice speed and acceleration across the Amundsen Sea Embayment, as well as time series and flow lines for notable outlet glaciers.

In western Poland, the town of Wapno has experienced dangerous land deformation due to a salt mine collapse in 1977. The town center has faced ongoing subsidence, with rates reaching up to 5 mm/year. The most significant risks stem from unstable geological conditions, causing periodic sinkholes, faults, and cracks in the terrain. After the mine's closure, no organization was responsible for monitoring deformation until the Geohazards Center of PGI-NRI was enlisted in 2013 to create an affordable remote sensing system.

Using PSI processing of archived ERS and Envisat data, radar corner reflectors (CR) were deployed at seven locations for SAR (Synthetic Aperture Radar) interferometric measurements, where natural radar reflecting objects were lacking. These specially designed corner reflectors enabled ascending and descending TerraSAR-X and Sentinel-1 observations, as well as GNSS and optical leveling measurements for validation. From 2014 to 2015, 40 TSX acquisitions were completed, followed by continuous S1 data. In March 2021, a sinkhole emerged in one problematic location, prompting monitoring via terrestrial laser scanning and UAV photogrammetry.

By carefully processing and decomposing Line of Sight data from all available TSX and Sentinel-1 A satellite tracks, near-daily CR displacement records were reconstructed and validated with leveling and GNSS. The CR displacement data verified the subsidence velocity obtained through PSI processing. The long-term CRInSAR observations (nearly 8 years) also identified seasonal effects and subsidence anomalies linked to sinkhole development. Corner reflectors have proven crucial for detailed scientific monitoring and sinkhole hazard mitigation. In 2022, the monitoring system was expanded with four additional corner reflectors to address spatial gaps in problematic areas.

SAR interferometry has been routinely used for surface deformation monitoring with a high impact on the geoscience community. The accuracy of the estimated deformation depends on several factors such as the atmospheric delay, the unwrapping errors and the phase decorrelation. Different approaches and techniques have been proposed to mitigate these effects and improve the accuracy of InSAR surface deformation. The most successful technique is the Persistent Scatterers (PS) (Ferreti et al., 2001) technique aimed to explore the phase stable of some particular pixels, the Persistent Scatters, within a time series of interferograms. The atmospheric effects are mitigated and the phase decorrelation is considerably reduced. A complementary technique, Distributed Scatterers (DS), has been proposed for rural areas with low PS density (Ferreti et al., 2011). This technique explores partially decorrelated areas in the time series and recovers natural scatters that are spatially correlated. To reduce the noise of the natural scatters a spatial filtering or multilook is applied to the interferogram. According to Maghsoudi et al. (2022), the multilooked interferograms reveal a systematic signal that interferes with the accuracy of the estimated deformation. They call it a fading signal with a short-living signal that could be due to soil moisture change or biomass growth or both.

In this work, we present the results of an experiment aimed to analyse the relationship between the phase bias and the time-varying soil moisture and vegetation water content. We show that the decorrelation phases are related to the variability of the vegetation water content computed using the Normalized Difference Water Index (NDWI) from Sentinel-2 images and to a less extent with the soil moisture change. We were able to improve surface deformation estimates after the removal of the soil moisture and vegetation water content.

Recently, Michaelides and Zebker (2020) have proposed a new approach for the estimation of the decorrelation phases based on the single value decomposition (SVD) solution of a system of equations with all phase triplets combinations within a time series of interferograms. Applying the methodology, Mira et al. (2022) have estimated the phase decorrelation and evaluated the relation between decorrelation phases and in-situ observed soil moisture. They report a scale effect of 10% between the in situ soil moisture variation and the decorrelation phase-derived soil moisture. Although some approaches have been proposed for t removing or mitigating the fading signal, the physical phenomenon is not fully understood.

To answer this question, we made an experiment on a rural area close to Lisbon, Portugal, where a soil moisture sensor was continuously operating during the experiment and the land cover is known. Ascending and descending Sentinel-1 SAR images were interferometrically processed using all possible pair combinations of SAR images in both polarizations (VV and VH). The deformation was estimated using the temporal small baseline approach. The phase was properly mutlilooked, unwrapped and calibrated. The resulting unwrapped phase time series was converted into cumulative surface deformation. The decorrelation phase was estimated with the single-value decomposition methodology proposed by Michaelides and Zebker (2020). The Normalized Difference Water Index (NDWI) was used to compute the vegetation water content with Sentinel-2 multispectral images acquired over the same area and during the same period. The estimated decorrelation phases, in situ soil moisture changes and the NDWI variability during the time series, were analysed in the study area. The results show that there is a spatial correlation between the NDWI variability and the decorrelation phases, that is, higher values of phase decorrelation correspond to higher values of NDWI variability. These areas correspond to intense agricultural practices. The linear regression between the decorrelation phase and the soil moisture shows for VV polarization an R² value of 0.76 and 0.86 for ascending and descending tracks respectively. It means that a large component of the descorrelation phase can be physically explained by the variability of vegetation water content within the analysed time interval.We have also observed that the phase bias can be removed using the decorrelation pahses or equivalently the vegetation water content variability.

This work was supported in part by Academia Militar, Portugal, under PhD Grant to Nuno Cirne Mira and by Fundação para a Ciência e Tecnologia (FCT) – project UIDB/50019/2020

References

Maghsoudi, Y., Hooper, A.J., Wright, T.J., Lazecky, M., Ansari, H., Characterizing and correcting phase biases in short-term, multilooked interferograms, Remote Sensing of Environment, 275, 113022, 2022.

A Ferretti, A., Prati, C., Rocca, F., Permanent scatterers in SAR interferometry, IEEE Transactions on geoscience and remote sensing 39 (1), 8-20, 2001.

Ferretti, A., Fumagalli, A., Novali, F., Prati, C., Rocca, F., Rucci, A., A new algorithm for processing interferometric data-stacks: SqueeSAR, IEEE transactions on geoscience and remote sensing 49 (9), 3460-347. 2011.

Michaelides, R., & Zebker, H. (2020). Feasibility of Retrieving Soil Moisture from InSAR Decorrelation Phase and Closure Phase. IGARSS 2020 - 2020 IEEE International Geoscience and Remote Sensing Symposium, 12–15. https://doi.org/10.1109/IGARSS39084.2020.9323833

Mira, N. C., Catalão, J., & Nico, G. (2022). Soil Moisture Variation Impact on Decorrelation Phase Estimated by Sentinel-1 Insar Data. IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium, 5792–5795. https://doi.org/10.1109/IGARSS46834.2022.9883817

Tillage farming in Ireland is a large industry with a valuation of roughly €650M of farm gate value to the rural economy, with its main market being for animal feed. As a result the primary crops grown are cereals (barley, wheat, oats), potatoes, and break crops. Despite its large portion of arable land, the country relies heavily on fodder imports due to the relative size of the national bovine herd. This mismatch in production to import ratio has put pressure on policymakers, who aim to increase tillage production to 1% per annum by 2027 [1]. Teagasc, the national agricultural research body, recommends that Ireland “maximizes crop yield potential by developing our understanding of the soil, crop, management and climate factors that limit crop yield” and “develop precision farming approaches” [1] to that end. One way to achieve this is by utilizing the suite of remote sensing instruments provided by ESA. However, it is difficult to monitor crops using traditional optical-based remote sensing methods due to the extensive number of overcast days for the majority of the island of Ireland, particularly during the winter seasons.

Using Sentinel-1 synthetic aperture radar (SAR), we demonstrate an alternative, more robust method, for both crop monitoring and climate shock detection, particularly during extended periods of cloud cover. We achieve this by building on previously determined relationships between colocated Sentinel-1 SAR and Normalized Difference Vegetation Index (NDVI) data derived from both Sentinel-2 and MODIS. We present a case-study for using this method on a small tillage and pasture family farm in Enniscorthy, Co. Wexford, located in the south-eastern area of the country where 50% of agricultural activity takes place, and where 80% of cereals are grown nationally [2]. We find that we can detect the drought year in crop yields of barley in 2018, which was of national importance due to a national fodder shortage at that time [3]. These events are predicted to increase as the precipitation seasons are altered due to climate change [4]. Our approach has several advantages, such as increased temporal monitoring of agricultural land, the ability to identify specific areas under cultivation that require in-situ examination and potential intervention regardless of cloud cover conditions, and a means of quantifying changes at a national level in the tillage farming calendar. This works for farmers, policymakers, and researchers interested in improving the sustainability and productivity of tillage farming in Ireland. SAR can provide information about the production status of national crops in near real-time, giving farmers on the ground, and policymakers advance warning of such shortages in the future.

[1] - https://www.teagasc.ie/media/website/publications/2020/2027-Sectoral-Road-Map---Tillage.pdf

[2] - https://www.cso.ie/en/releasesandpublications/ep/p-fss/farmstructuresurvey2016/da/lu/

[3] - https://hydrologyireland.ie/wp-content/uploads/2021/12/03-Paul-Leahy-NHC_ClimAg_A0_Poster_Leahy.pdf

[4] - https://www.epa.ie/publications/research/climate-change/research-339-high-resolution-climate-projections-for-ireland-.php

In recent years, the availability of freely available Sentinel-1 images with continuous and regular acquisition, the development of Advanced Differential InSAR (A-DInSAR) techniques, and the increase in computational resources have allowed the implementation of Sentinel-1-derived satellite interferometric products that facilitate in monitoring over large areas. In fact, services at various scales have lately been established at continental, national and regional levels (Crosetto et al., 2020) with the purpose of giving an overview of the ground deformation active in the area of interest. The resulting deliverables are generally velocity maps and displacement time series for each Measurement Point (MP). In Europe, the European Ground Motion Service (EGMS) was recently activated under the supervision of the European Environment Agency. The service, which currently spans from 2015 to 2021, comprises Sentinel-1 SLC imagery processed by A-DInSAR. The main accessible products are divided into three levels: full resolution deformation maps with measurements along the radar Line-Of-Sight (LOS) (Level 2A), InSAR outcomes combined with the GNSS network (Level 2B), and horizontal (east-west) and vertical (up-down) component deformation maps at reduced spatial resolution (Level 3) (Crosetto et al., 2020). In Italy, in the regions of Tuscany (central Italy), Valle d′Aosta (northwestern Italy), and Veneto (northeastern Italy), a continuous monitoring program based on Sentinel-1 satellite interferometry has been deployed. The principal derived products include velocity maps with displacement time series for ascending and descending orbits from 2015 to the present, and an anomaly detection database (Confuorto et al., 2021). Both EGMS and regional products cannot be utilized to provide early warning systems or to forecast potential deformations. To this end, a site-specific analysis is required for a detailed investigation.

In this work, we investigated ascending and descending data from both the EGMS and the regional monitoring services available in the Veneto Region (NE Italy). In particular, we focused our interest on the detection of landslides in the province of Belluno (Veneto Region) with the help of the Inventory of Landslide Phenomena in Italy (IFFI) in order to identify their state of activity. The density of points coverage was taken into account for a spatial analysis, as well as the displacement time series for a temporal analysis. Moreover, for a more detailed analysis, a site-specific study was conducted by processing data from several multi-sensor satellites, such as Sentinel-1 and COSMO-SkyMed, using the most common A-DInSAR techniques. The results show the potentiality and the advantages of having three distinct services working at different investigative scales. Additionally, the use of site-specific processing potentially allows for an update of the time period of study, an improvement of the coverage area and an enhancement of the precision of the interpretation. Moreover, a more detailed investigation could lead to the development of an early warning system and the assessment of future landslide evolution scenarios.

Confuorto, P., Del Soldato, M., Solari, L., Festa, D., Bianchini, S., Raspini, F., & Casagli, N. (2021). Sentinel-1-based monitoring services at regional scale in Italy: State of the art and main findings. International Journal of Applied Earth Observation and Geoinformation, 102(July), 102448. https://doi.org/10.1016/j.jag.2021.102448

Crosetto, M., Solari, L., Mróz, M., Balasis-Levinsen, J., Casagli, N., Frei, M., Oyen, A., Moldestad, D. A., Bateson, L., Guerrieri, L., Comerci, V., & Andersen, H. S. (2020). The evolution of wide-area DInSAR: From regional and national services to the European ground motion service. Remote Sensing, 12(12), 1–20. https://doi.org/10.3390/RS12122043

Within the framework of the EGMS validation project - funded by the European Environment Agency in the framework of the Copernicus program - the activity that we present here aims at comparing results from the EGMS service with pre-existing ground motion databases (called “inventories”) providing information on the position and geometry of known ground motion phenomena. Inventories are generally provided in the form of a polygon delimiting a given phenomenon or just in the shape of a point located at its center. The rationale of this evaluation is that a specific interest for geo-risk management end-users is the possibility to use EGMS data to complete (or even to build) new inventories of phenomena, because existing inventories are rarely exhaustive. Moreover, sometimes inventories do not exist at all over specific areas.

For the cross comparison, we propose the following approach. On the one hand, we will verify that the EGMS products (level 2b) located inside a polygon of an inventory have a significant movement compared to its neighborhood. Secondly, we will evaluate whether polynomials generated - automatically following an ADA (Active Deformation Areas) approach or by visual delimitation - from EGMS products have similar geometric characteristics to those contained in the databases. Finally, when the information is in the form of points, we will try to evaluate the number of phenomena identified in the inventory that coincide in terms of position with the polygons obtained from the EGMS products and those that do not. Contrary to the comparison with geodetic type measurements, we are comparing information of very different natures. Also, due to the partly qualitative nature of this exercise, the interpretation of the results will be very important.

Among all the sites selected for the validation of EGMS, we will present here an analysis applied to post-mining and landslide sites located in France and Spain. These two types of phenomena have very distinct geometric (extent) and movement (velocity) characteristics. They will be representative of a wide variety of phenomena observable from EGMS products. The results presented here will be used as a reference assessment of the EGMS in the future to come.

References

Solari, L., Barra, A., Herrera, G., Bianchini, S., Monserrat, O., Béjar-Pizarro, M., et al. (2018). Fast detection of ground motions on vulnerable features using Sentinel-1 InSAR data. Geomatics, Natural Hazards and Risk, 9(1), 152-174

Abstract: The Wilkes Subglacial Basin is one of the largest marine-based drainage basins in East Antarctica, which contains the ice equivalent of 3 to 4 m of mean sea level rise. It is essential to determine the grounding line migration of Cook Glacier, which has two outlets called Cook East Glacier and Cook West Glacier, as it acts as a key indicator of ice discharge from the Wilkes Subglacial Basin and instability of the marine ice sheets in the region. In this study, we identified the location of the grounding line of Cook Glacier by applying double-differential interferometric SAR (DDInSAR) to 8 InSAR pairs with a temporal baseline of 1-day acquired by the COSMO-SkyMed satellite constellation from 2020 to 2021. The DDInSAR is a technique for differentiating two differential interferograms. If the ice velocity of a floating glacier is constant, the DDInSAR technique can remove the flow-induced displacement and produce only the difference in the tidal deflection of the glacier. In the DDInSAR image, the equi-displacement line of zero can be defined as the grounding line. We identified the location of the grounding line of Cook East and Cook West Glaciers from the COSMO-SkyMed DDInSAR images and compared it with the grounding line detected from European Remote-Sensing Satellite-1/2 (ERS-1/2) DDInSAR images in 1996. The grounding line showed a spatially different migration. On the Cook East Glacier, the position of grounding line has changed little over the past 25 years, except in a few areas where the grounding line has advanced by ~4.5 km. The observed grounding line advance is possibly due to the inaccuracy of the grounding line position determined from the 1996 ERA-1/2 DDInSAR. Meanwhile, the grounding line of Cook West Glacier has retreated about 7 km, probably due to the ocean-induced basal melting of the glacier. The grounding line retreat of Cook West Glacier has the potential to significantly destabilize the marine ice sheet in the region. The bed elevation at the grounding line of Cook West Glacier is several hundred meters below sea level, and the elevation decreases rapidly upstream. This suggests that the rate of grounding line recession at Cook West Glacier may accelerate in the future.

The ice ridge is a linear pile-up of sea ice fragments, which has different sizes and shapes, on the upper and lower surface of the sea ice. The formation of ice ridges is caused by the breaking of sea ice under the action of wind, current and other environmental dynamics, accompanied by compression and overlapping. It is mainly composed of the ridge sail and keel. Ice ridges change the shape of sea ice surface, which is a potential danger for ships to navigate. Generally, the salinity and density of the ice ridge are lower than the surrounding level ice. Due to the dominant role of volume scattering, the backscattering signal of the ice ridge is higher than that of the surrounding level ice. At present, the extraction methods of ice ridges in SAR images are mostly based on their bright linear features, including direct threshold method and detection algorithm based on structure tensor. However, due to the interference of other backscattering characteristics similar to the ice ridge in the sea ice, such as the edge of floating ice and wind-induced rough lead, the traditional extraction methods based on backscattering intensity usually are not ideal.

Considering the height characteristics of ice ridges, they are extracted by interferometric synthetic aperture radar (InSAR) technology in this research. The extraction method is based on the assumption that the ridge height is greater than 1 meter and the width is less than 100 meters. Single-pass InSAR is an effective technique for sea ice topographic retrieval because the target motion between two received signals could be ignored. The interferometric phase includes information about terrain and noise. The phase noise caused by surface and volume scattering effects and radar system noise can be ignored under ideal conditions. Therefore, the sea ice surface height could be obtained from the interferometric phase by the single-pass InSAR technology.

According to the height difference between the ice ridge and the surrounding sea ice, an appropriate height threshold is set to extract the area with high sea ice terrain. Finally, using the curve characteristics of the ice ridge, the preliminary extraction results are processed by morphology. Simulation results show the effectiveness of this method. Besides, the method is tested with TanDEM-X data. The results show that the proposed method has good performance on ice ridges extraction.

This research was supported by the National Natural Science Foundation of China (No. 62231024).

Phase inconsistency exists in interferometric synthetic aperture radar (InSAR) processing when multilooking is used for suppressing the speckle noise ^[1]. Phase inconsistency had been ignored for a long time in multi-temporal InSAR (MT-InSAR) until researchers revealed closure phase, non-zero redundancy in a loop of interferograms of distributed scatterers ^{[2, 3]}. The phase inconsistency is reported to be related to ground physical changes, such as soil moisture and vegetation ^[4-6]. Moreover, current phase estimators are primarily based on the assumption of Gaussian circular noises. Phase inconsistency breaks this assumption; therefore, bias can exist in the restored time-series phase, leading to bias in the land deformation results. Recently, more and more attention has been paid to the inconsistent phase in SAR community ^[7-9]. It has been proposed that combination of different closure phases can be used to restore the inconsistent phase series of MT-InSAR. Currently, there are several studies focusing on sequential closure phase with a regular time interval. For examples, Maghsoudi et al. proposed to use closure phase from triple and quadra interferograms to restore the inconsistent phase ^[10], and Zheng et al. analysed the sequential closure phase in detail with respect to the inconsistent phase and proposed a workflow to calculate the inconsistent phase ^[11]. However, after experiment we found that the restored inconsistent phase results differ with different selection of the time interval. Practically, the regular time interval of closure phase is hardly to be meet in many applications due to extra limitation of spatial baseline, such as baseline selection in small baseline subset (SBAS) processing. In this study, we demonstrate the impact of different closure combinations with different time intervals on the inconsistent phase correction. In addition, we propose a practical combination based on temporal and spatial baseline selection results in SBAS. Finally, the results derived from different strategies for closure phase combination are compared with simulation and real data experiments.

References

[1] Ansari, H., De Zan, F. and Parizzi, A., 2020. Study of systematic bias in measuring surface deformation with SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, 59(2), pp.1285-1301.

[2] Morrison, K., Bennett, J.C., Nolan, M. and Menon, R., 2011. Laboratory measurement of the DInSAR response to spatiotemporal variations in soil moisture. IEEE Transactions on Geoscience and Remote Sensing, 49(10), pp. 3815–3823.

[3] Hensley, S., Michel, T., Van Zyl, J., Muellerschoen, R., Chapman, B., Oveisgharan, S., Haddad, Z.S., Jackson, T. and Mladenova, I., 2011. Effect of soil moisture on polarimetric-interferometric repeat pass observations by UAVSAR during 2010 Canadian soil moisture campaign. In 2011 IEEE International Geoscience and Remote Sensing Symposium (pp. 1063–1066).

[4] Zwieback, S., Hensley, S. and Hajnsek, I., 2015. Assessment of soil moisture effects on L-band radar interferometry. Remote Sensing of Environment, 164, pp. 77–89.

[5] De Zan, F., Parizzi, A., Prats-Iraola, P. and López-Dekker, P., 2014. A SAR interferometric model for soil moisture. IEEE Transactions on Geoscience and Remote Sensing, 52(1), pp. 418–425.

[6] Eshqi Molan, Y., Lu, Z., 2020. Modeling InSAR Phase and SAR Intensity Changes Induced by Soil Moisture. IEEE Trans. Geosci. Remote Sensing 58(7), pp. 4967–4975. https://doi.org/10.1109/TGRS.2020.2970841

[7] Jiang, M., 2014. InSAR coherence estimation and applications to earth observation, The Hong Kong Polytechnic University.

[8] De Zan, F., Zonno, M. and Lopez-Dekker, P., 2015. Phase inconsistencies and multiple scattering in SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, 53(12), pp. 6608–6616.

[9] Liang, H., Zhang, L., Ding, X., Lu, Z., Li, X., Hu, J., Wu, S., 2021. Suppression of Coherence Matrix Bias for Phase Linking and Ambiguity Detection in MTInSAR. IEEE Transactions on Geoscience and Remote Sensing, 59(2), pp. 1263–1274.

[10] Maghsoudi, Y., Hooper, A.J., Wright, T.J., Lazecky, M., Ansari, H., 2022. Characterizing and correcting phase biases in short-term, multilooked interferograms. Remote Sensing of Environment 275, 113022. https://doi.org/10.1016/j.rse.2022.113022

Zheng, Y., Fattahi, H., Agram, P., Simons, M., Rosen, P., 2022. On Closure Phase and Systematic Bias in Multilooked SAR Interferometry. IEEE Trans. Geosci. Remote Sensing 60, pp. 1–11. https://doi.org/10.1109/TGRS.2022.3167648

We study the Earth’s surface displacement field that was induced by the Mw 7.8 and Mw 7.5 seismic events occurred on 6^th February 2023 in South-East Turkey. We applied both the Differential SAR Interferometry (DInSAR) and the Pixel Offset (PO) techniques to a large set of spaceborne SAR images acquired by different satellite constellations.

DInSAR has widely demonstrated to be an effective tool to detect ground deformation at large spatial scale and with centimeter accuracy. Due to the wide diffusion of open access SAR datasets, DInSAR is nowadays used in operational services to retrieve the co-seismic surface displacements induced by an earthquake. One of this service is the EPOSAR one [1] that, within the framework of EPOS (European Plate Observing System) [2] and by exploiting the Copernicus Sentinel-1 data, allows producing co-seismic displacement maps at global scale and in an automatic way, immediately after the availability of a post-event acquisition. However, in case of large magnitude earthquakes like those under study, the expected displacement can reach up several meters, i.e., can be on the order of the SAR pixel size. Hence, particularly in the near-field event, it can be experienced a loss of coherence, thus making DInSAR not suitable to retrieve the actual displacement. Nonetheless, when the deformation introduces geometric distortions without significantly disturbing the SAR image reflectivity, displacements can be observed by comparing the amplitudes of SAR image pairs acquired before and after an event [3]. Based on this principle, the PO technique allows measuring, although with reduced accuracy with respect to DInSAR, ground deformation on the order of the SAR pixel size. Accordingly, to reach better accuracies small pixel sizes are preferable. Moreover, by jointly considering DInSAR and PO estimated on ascending and descending acquisitions over the same area of study, it is possible to retrieve the full three-dimensional deformation field [3].

In this work, to study the ground displacement induced by the South-East Turkey earthquakes, we exploit SAR datasets consisting of several co-seismic data pairs that have been collected by different satellite constellations. First of all, we exploited C-band (5.6 cm of wavelength) SAR data acquired by the Sentinel-1A sensor (pixel size: 4.5m along range and 14m along azimuth) from both ascending (Track 14) and descending (Track 94 and 21) orbits. By applying the PO technique, Sentinel-1 data allows to retrieve, with a good accuracy, the displacement along the range direction, while are less accurate along the azimuth one, due to the larger pixel size. To overcome this limitation, we also benefitted from the availability of a number of L-band (23 cm of wavelength) SAR images acquired by the twin satellites of the Argentine SAOCOM-1 constellation, programmed in collaboration with the Italian and Argentine Space Agencies. SAOCOM-1 data are acquired in Stripmap mode, with a pixel size of about 5m by 4m along range and azimuth, respectively, and completely cover the area interested by the earthquakes with 6 ascending and 5 descending tracks. Figure 1 shows an example of interferogram (Figure 1a), as well as of range (Figure 1b) and azimuth (Figure 1c) Pixel Offsets calculated from a SAOCOM-1 data pair spanning the earthquakes.

By jointly exploiting DInSAR and PO measurements that are retrieved from the described rich SAR dataset, we finally generate a detailed 3D co-seismic deformation field that may allow to effectively model the co-seismic sources of the earthquakes.

This work is supported by: the 2022-2024 IREA-CNR and Italian Civil Protection Department agreement, and by the H2020 EPOS-SP (GA 871121) and Geo-INQUIRE (GA 101058518) projects. The authors also acknowledge ASI for providing the SAOCOM-1 data under the ASI-CONAE SAOCOM-1 License to Use Agreement. Sentinel-1 data were provided through the European Copernicus program.

References

1. Monterroso, M. et al., 2020, A Global Archive of Coseismic DInSAR Products Obtained Through Unsupervised Sentinel-1 Data Processing, Remote Sens., vol. 12, no. 3189, pp. 1–21. https://doi.org/10.3390/rs12193189
2. EPOS-RI – www.epos-eu.org
3. Fialko, Y. et al., 2001, The complete (3-D) surface displacement field in the epicentral area of the 1999 MW7.1 Hector Mine Earthquake, California, from space geodetic observations: Geophysical Research Letters, v. 28, p. 3063–3066, doi:10.1029 /2001GL013174.

In the framework of the new French National Service of Observation “ISDeform”, dedicated to assist scientists in their usage of satellite imagery for monitoring surface deformation, we proposed a specific processing of Sentinel-1 data over the French metropolitan territory, in complement to the already available products of the European Ground Motion Service. The products will be made freely available to the scientific community. The goals are to provide : (1) a large-scale motion map in ITRF or EUREF reference frame with limited inputs from GNSS to preserve independence of observed large-scale motions from GNSS data; (2) time series of measured LOS phase delay as a function of time, devoid of any temporal filtering or model assumption; (3) different time-series products with different applied spatial filters; (4) associated products allowing scientists to assess the quality of the processing and the uncertainty of the obtained displacement maps.
To do so, we start with an automated processing by the FLATSIM service (Thollard et al., 2021) operated by CNES for the french ForM@TeR pole for data and service for the solid earth. The coverage of the french territory was divided into 28 segments, 14 ascending and 14 descending, with along-track overlaps of about 200 km. All archived Sentinel-1 data completely covering the segments from end of 2014 to April 2021 have been processed using a small-baseline strategy and the NSBAS processing chain (Doin et al., 2011). The number of retained acquisitions per segment is 291 on average. On average 1244 interferograms have been constructed per track, with a network including the n/n+1, n/n+2, n/n+3, n/n+2months and n/n+1year pairs for each acquisition “n”. The FLATSIM service provides wrapped differential interferograms in radar geometry, corrected from phase delay using the ERA-5 ECMWF atmospheric variables, that have been multilooked by a factor 8 in range and 2 in azimuth, and are here referred to as 2-looks interferograms. A further multilooking by a factor 4 is done before filtering and unwrapping. Spatial unwrapping stops where it must cross areas of low coherence. The time series is inverted using all available unwrapped phase values for a given pixel and the results are provided in terrain geometry with a spatial resolution of 120m. The default FLATSIM processing has been validated for all tracks. Despite drawbacks in the automated processing, the time series present an interesting and consistent seasonal behavior over France. However, unwrapping of one year interferograms is strongly impeded by low coherence in vegetated areas covering most France, and the velocity maps are dominated by apparent subsidence due to fading signals over crop areas (Ansari et al., 2021).
In order to overcome the limits of the FLATSIM processing and reduce the impact of fading signals, we implemented a new processing strategy that starts with the 2-looks products available in radar geometry: wrapped interferograms, temporal coherence proxy based on triplet inconsistencies, and the dispersion of radar backscatter amplitude. We analyse the signature of fading signals and devise a proxy in 2-looks for their potential bias impact (Cheaib and Doin, Fringe meeting, 2023). The bias proxy is used in the multilooking and filtering steps to avoid contamination of bias-free pixels by others. A new spatial filter and an improved unwrapping strategy are implemented, resulting in large unwrapped fractions of the image footprint, even for one-year interferograms. Unwrapping errors are detected by network misclosure during the temporal inversion step, and corrected iteratively starting from short baseline interferograms. For a few dates, especially including snow cover, unwrapping errors over a given area are too numerous for unambiguous correction of the phase. These areas are masked on the interferograms affected by the problem (mostly snow effects).
We will present the final time series and associated velocity maps. When including one-year interferograms, they present very limited bias, mostly restricted to areas where one-year interferograms cannot be unwrapped. Different time series are computed with different spatial filters applied. “PS-DS” like results can be obtained when we do not apply any low-pass filtering, and with or without high-pass filtering. For resolving large-scale deformation patterns, solutions with spatial filtering for extracting a continuous displacement map even in low-coherence areas are interesting. Provided quality maps quantify the property of a given pixel (coherence, bias proxy, network misclosure, network misclosure of 6 to 12 days interferograms, ...) or the adjustment of the displacement model (linear and seasonal) to the phase time series that includes residual atmospheric phase screens. A first quantitative comparison with EGMS products will be presented on specific sites of interest, especially those used for EGMS validation.

References

H. Ansari, F. De Zan, and A. Parizzi. Study of systematic bias in measuring surface deformation with sar interferometry. IEEE Transactions on Geoscience and Remote Sensing, 59(2):1285–1301, 2021.
M.P. Doin, F. Lodge, S. Guillaso, R. Jolivet, C. Lasserre, G. Ducret, R. Grandin, E. Pathier, and V. Pinel. Presentation of the small baseline nsbas processing chain on a case example: the etna deformation monitoring from 2003 to 2010 using envisat data. Proceedings of the ESA Fringe 2011 Workshop, Frascati, Italy, (19-23 September 2011), 2011:19–23, 2011.
Thollard, F., Clesse, D., Doin, M.-P., Donadieu, J., Durand, P., Grandin, R., Lasserre, C., Laurent, C., Deschamps-Ostanciaux, E., Pathier, E., Pointal, E., Proy, C., Specht, B., FLATSIM: The ForM@Ter LArge-Scale Multi-Temporal Sentinel-1 InterferoMetry Service, Remote Sensing, 13, 2021,18, 10.3390/rs13183734

Vertical total electron content (VTEC) and three-dimensional electron density are two important parameters to characterize the ionospheric spatial structure and variations. Several methods and models have been developed to obtain these two parameters, such as the global navigation satellite system (GNSS), ionosonde, incoherent scattering radar (ISR), coherent scattering radar, and the international reference ionosphere (IRI) model. A challenge to these methods and models is the low spatial resolution, leaving it difficult to analyze the ionospheric spatial variations. As an advanced space observation technique, space-borne synthetic aperture radar (SAR) has demonstrated potential for mapping high-spatial-resolution VTEC and three-dimensional electron density. However, the precision of SAR-based method is limited by the SAR imaging geometry. In this context, the improved method is proposed to map the high-spatial-resolution VTEC and three-dimensional electron density. The VTEC is estimated by combing of azimuth shift and split range-spectrum methods. The azimuth shift method is based on the phenomenon that ionosphere is sensitive to the pixel changes in azimuth direction and therefore can estimate the large-scale ionosphere. Split range-spectrum method exploits the dispersive nature of radar signals in estimating the ionospheric signals and is sensitive to the small-scale ionosphere. Once the VTEC is estimated, the initial three-dimensional electron density is calculated by ingesting the SAR-derived VTEC into an international reference ionosphere (IRI) model. In this process, the ionospheric global (IG) index is updated by minimizing the difference between the SAR-derived and IRI-derived VTECs, and the initial high-spatial-resolution electron density is reconstructed by exploiting the monotonic relationship between the electron density and the IG index. The initial electron density is further optimized by computerized ionospheric tomography (CIT) method. For a performance test of the proposed method, L-band Advanced Land Observation Satellite (ALOS) Phase Array L-band SAR (PALSAR) SAR images over Alaska regions are processed. The result shows that it is consistent between SAR-derived VTEC and international global navigation satellite system service (IGS) VTEC, demonstrating the reliability of the estimated VTEC. When comparing with the constellation observing system for meteorology, ionosphere, and climate (COSMIC) observations, the IRI-derived electron density profile is obviously corrected by the SAR-derived VTEC. The ionospheric variation in horizontal and vertical direction is analyzed and discussed over the study area. Our results prove that it is possible to map the high-spatial-resolution VTEC and three-dimensional ionospheric distribution from SAR images.

Open pit mines are mines that are exposed on a large scale to the surface. Open pit mining has problems such as environmental pollution due to mining activities and degradation of slope stability due to waste rock dumping. Therefore, systematic and continuous analysis for open pit mines is required. The Musan mine, located in Hamkyungbukdo Province, North Korea, is the most representative mine and the largest open pit mine in North and South Korea. The storage of tailings, where dumping has been completed, in open pit mines has the land cover with little vegetation. Hence, the application of InSAR technology to open pit mining has benefits to analyze the surface accurately and also can be powerful way for land subsidence monitoring. Among InSAR technologies recently used to observe surface deformation, Persistent Scatterer InSAR (PSInSAR) technology is widely recognized for its reliability and applicability. PSInSAR derives time-series displacements in millimeters using a persistent sactterer (PS) with a stable backscattered signal within a pixel. Using PSInSAR with Sentinel-1A/B SAR images and Stanford Method for Persistent Scatterers (StaMPS), we observed the surface displacement of the Musan mine about 5-year period from March 2017 to December 2021. We processed long-term PSInSAR using all images from a period of 5 years and we found that there is a continuous surface subsidence. However, the high deformation rate resulted in unwrapping errors. And long temporal coverage led the decorrelation of coherence so, there was a slight amount of PS. In order to ease the unwrapping error and increase the quantity of PS, we conducted several additional experiments. First, we re-derive PSInSAR the results by adjusting the unwrapping time window in the StaMPS process. In the study area, which exhibited fast deformation rates, we found that the smaller the unwrapping time window, the less frequently unwrapping errors occurred. And then, we decided to perform PSInSAR by dividing the time intervals into 1-year in order to obtain sufficient and high-quality PS. We found vertical displacements of up to around 220 mm/yr in the tailings storage area. We also found that east-west horizontal displacements occur on each side of the slope towards the valley. In this study, surface displacement derived from PSInSAR results was comprehensively analyzed using InSAR coherence and multi-temporal Digital Elevation Model (DEM).

The availability of Copernicus Sentinel-1 data, which is systematically acquired with global coverage, has led to the development of new applications in Remote Sensing. The vast amount of generated data allows for the use of scalable deep learning methods that can efficiently and accurately automate the extraction of information from these extensive data archives [1]. This automation can be used to monitor key earth processes, including geohazards. Volcanic hazards, in particular, are critical for reducing disaster risk, especially in urban areas where more than 800 million people live within 100km of an active volcano [2]. Such hazards pose a valid threat to the population, while volcanic eruptions may disrupt airspace operations.

Despite initiatives such as the Geohazard Supersites and Natural Laboratories, less than 10% of active volcanoes are monitored systematically. However, early detection of volcanic activity is crucial to mobilise scientific teams promptly, deploy ground sensing equipment, and alert civil protection authorities. Interferometric Synthetic Aperture Radar (InSAR) products provide a rich source of information for detecting ground deformation associated with volcanic unrest [3], which is statistically linked to an eruption [4]. Such deformation appears in the wrapped InSAR data as interferometric fringes. Unfortunately, atmospheric signals can produce similar fringe patterns, mainly due to vertical stratification that is correlated with topography, making it challenging to automatically detect interferograms with fringes attributed to volcanic ground deformation.

Recent studies have highlighted the potential of using Sentinel-1 InSAR data and supervised deep learning methods to detect volcanic deformation signals, with the aim of mitigating global volcanic hazards. However, detection accuracy is hindered by the lack of labeled data and class imbalance. Moreover, transfer learning approaches and heavy data augmentation techniques often result in models that fail to generalize well to previously unseen test samples. In this work, we introduce Pluto, an end-to-end early warning system for the global, automatic, detection and classification of volcanic activity based on deep learning with Sentinel-1 InSAR data. Pluto is based on Hephaestus [5], the InSAR dataset that we manually annotated to train deep models and on two modeling approaches that concentrate on self-supervised learning and domain adaptation methods.

Hephaestus is a curated wrapped InSAR dataset based on Sentinel-1 data, which enables the deployment of various services, such as automatic InSAR interpretation, volcanic activity detection, classification, and localization, as well as the identification and categorization of atmospheric contributions and processing errors. It contains annotations for roughly 20,000 InSAR frames from COMET-LiCS [6], covering the 44 most active volcanoes globally. This is the first publicly available large-scale InSAR dataset. Annotating such a dataset was a non-trivial task that required a team of InSAR experts to examine and manually annotate each frame individually. However, even with such a dataset, class imbalance poses a significant challenge to modeling volcanic activity, as the vast majority of available samples are not positive. In other words, natural hazards are rare yet destructive phenomena. To mitigate this, we provide over 100,000 unlabeled InSAR frames with Hephaestus (resulting in millions of 224x224 cropped patches) for global large-scale self-supervised learning.

In our work, we proceed to train deep learning models for InSAR binary classification (volcanic deformation or not), semantic segmentation of ground deformation, volcano state classification (unrest, rebound, rest) and classification of magmatic source (Mogi, Sill, Dyke). To address the issue of class imbalance, we have adopted two distinct modelling strategies. In the first strategy, we utilize self-supervised learning to train global, task-agnostic models that can handle distribution shifts caused by spatio-temporal variability, as well as major class imbalances [7]. In the second approach, we have introduced a novel framework for domain adaptation [8], in which we learn class prototypes from synthetically generated InSAR data [9], which we can generate in abundance, using vision transformers. Our approach can generalize well to the real InSAR data domain, without requiring additional human annotations. These models are currently the state-of-the-art for the InSAR binary classification task, with classification accuracy exceeding 95%.

The models are then fine-tuned to the labeled part of Hephaestus to create the foundation for a global early warning system for volcanic activity, called Pluto. Pluto continuously updates its database by synchronising with the COMET-LiCS Sentinel-1 InSAR portal, receiving new InSAR data collected over volcanic regions worldwide. This data is automatically fed into the trained models for detection of volcanic activity. If volcanic activity is detected, Pluto sends an email alert to users, containing all necessary information such as the InSAR metadata, the intensity of the event, and the exact location of the activity. To improve the service, a pipeline is implemented to collect misclassified samples in production and use them to further train and improve the existing models. This approach ensures the robustness and continual enhancement of the Pluto service.

In conclusion, Pluto is an end-to-end artificial intelligence based system for the detection and mitigation of volcanic hazards. It provides volcano observatories and civil protection stakeholders with early warnings and critical information to seamlessly and timely assess volcanic hazard associated with ground deformation on a global scale.

References

[1] Zhu et al., “Deep learning meets sar: Concepts, models, pitfalls, and perspectives,” IEEE Geoscience and Remote Sensing Magazine, vol. 9, no. 4, pp. 143–172, 2021.

[2] Brown et al, “Volcanic fatalities database: analysis of volcanic threat with distance and victim classification,” Journal of Applied Volcanology, vol. 6, no. 1, pp. 1–20, 2017.

[3] Papoutsis et al., “Mapping inflation at Santorini volcano, Greece, using GPS and InSAR”. Geophysical Research Letters, 40(2), pp.267-272. 2013.

[4] Biggs et al., “Global link between deformation and volcanic eruption quantified by satellite imagery,” Nature communications, vol. 5, no. 1, pp. 1–7, 2014

[5] Bountos et al., "Hephaestus: A large scale multitask dataset towards InSAR understanding." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, EarthVision Workshop, 2022.

[6] Lazecký et al. "LiCSAR: An automatic InSAR tool for measuring and monitoring tectonic and volcanic activity." Remote Sensing 12.15, 2430, 2020.

[7] Bountos et al. "Self-supervised contrastive learning for volcanic unrest detection." IEEE Geoscience and Remote Sensing Letters 19, 1-5, 2021.

[8] Bountos et al., "Learning from Synthetic InSAR with Vision Transformers: The case of volcanic unrest detection." IEEE Transactions on Geoscience and Remote Sensing, 2022.

[9] Gaddes et al., “Using machine learning to automatically detect volcanic unrest in a time series of interferograms,” Journal of Geophysical Research: Solid Earth, vol. 124, no. 11, pp. 12304–12322, 2019.

The development of innovative monitoring approaches based on the critical condition of infrastructure assets has triggered new demand for the use of novel technologies to be applied with non-destructive testing (NDT) methods and on-site inspections [1]. In this framework, satellite remote sensing data and multi-temporal processing techniques, have proven to be effective in monitoring ground displacements of transport infrastructure by MT-InSAR, including roads, railways and airfields, with a much higher temporal frequency of investigation and the capability to cover wider areas [2,3]. In addition, the integration of information provided by several satellite missions, including optical, multispectral and SAR data, can be effectively used for routine monitoring purposes, reaching very high standards for data quality and accuracy. On the other hand, the stand-alone implementation of these data do not allow to investigate about the causes of the detected damages associated to transport infrastructure (i.e. displacements, road damages). To overcome these limitations, an integrated investigative approach was proposed based on satellite information and data coming from ground-based non-destructive testing methods (NDTs) and on-site inspections. Several experimental applications, including satellite data, have been conducted for the provision of continuous and faster measurements to replace existing non-destructive technologies based on discrete methods of data collection. This approach was effectively applied in a variety of infrastructure categories, related to the higher requirements for the frequency of testing (e.g., bridges, railways, airfields), as well as the essential configuration of linear transport structures. Several applications were performed integrating information derived by multi-source satellite data, including SAR, optical, multispectral data, with ground-based NDTs (i.e. ground penetrating radar, levelling, mobile and terrestrial laser scanners). Furthermore, recent advances, main challenges and future perspectives arising from data integration for transport infrastructure monitoring were investigated, showing the high potential of satellite information, to be included in the next generation of infrastructure management systems.

Keywords – Satellite Remote Sensing, Non-Destructive Testing Methods, Laser Scanners, Ground Penetrating Radar (GPR), Integrated Health Monitoring, Railway monitoring, Transport Infrastructure Maintenance

Acknowledgments

The authors want to acknowledge the Italian Space Agency (ASI) for providing the COSMO-SkyMed Products® (©ASI). The Sentinel 1A products are provided by ESA (European Space Agency) under the license to use. This research is supported by the Italian Ministry of Education, University and Research (MIUR) under the National Project “EXTRA TN”, PRIN 2017 and the Projects “VAGARE (GDR 2020)” and “M.LAZIO”, accepted and funded by the Lazio Region, Italy.

References

[1] Chang, P.C.; Flatau, A.; Liu, S.C. Review Paper: Health Monitoring of Civil Infrastructure. Struct. Health Monit. 2003, 2, 257–267

[2] Tosti, F.; Gagliardi, V.; D’Amico, F.; Alani, A.M. Transport infrastructure monitoring by data fusion of GPR and SAR imagery information. Transp. Res. Procedia 2020, 45, 771–778

[3] Gagliardi, V.; Tosti, F.; Ciampoli, L.B.; Battagliere, M.L.; Tapete, D.; D’Amico, F.; Threader, S.; Alani, A.M.; Benedetto, A. Spaceborne Remote Sensing for Transport Infrastructure Monitoring: A Case Study of the Rochester Bridge, UK. In Proceedings of the IGARSS 2022—2022 IEEE International Geoscience and Remote Sensing Symposium, Kuala Lumpur, Malaysia, 17–22 July 2022; pp. 4762–4765

[4] Bianchini Ciampoli, L.; Gagliardi, V.; Ferrante, C.; Calvi, A.; D’Amico, F.; Tosti, F. Displacement Monitoring in Airport Runways by Persistent Scatterers SAR Interferometry. Remote Sens. 2020, 12, 3564.

Subglacial lakes beneath the Antarctic Ice Sheet were first identified using airborne radio-echo sounding (RES) surveys in 1970 (Robin, G. de Q. 2000). Since then, studies have identified subglacial lake locations and extent using RES and active lakes using satellite altimetry. Overall, of the 773 subglacial lakes identified globally, 675 of these are located in Antarctica, 20% of which exhibit surface elevation change suggestive of lake draining and filling cycles (Livingstone et al. 2022). Clusters of these “active” subglacial lakes are often located along subglacial hydrological pathways, enabling transfer of water within connected lake networks (Fricker et al. 2007, Stearns et al. 2008, Fricker et al. 2009). Despite efforts to characterise this understudied component of ice sheet mechanics, identifying the location and extent of subglacial lakes remains a work in progress, and observational studies of ice dynamic change connected to subglacial lake activity remain limited. Furthermore, triggers of lake drainage events, as well as drainage mechanisms themselves, are unresolved.

Here, we present the first Antarctic-wide analysis of subglacial lake activity and ice dynamic change. We use a new subglacial lake location dataset to assess whether changes in ice speed can be observed around periods of subglacial lake activity in Antarctica. Intensity feature tracking of 6/12-day repeat pass Single Look Complex (SLC) Synthetic Aperture Radar (SAR) images from the ESA-EC Sentinel-1 satellite mission, acquired in Interferometric Wide (IW) swath mode, coincident in time with CryoSat-2 swath-mode elevation change data, is used to measure a six-year record of ice velocity variations around subglacial lake activity. We investigate speed anomalies on active subglacial lakes beneath the Antarctic Ice Sheet, by separating radar scattering horizon changes due to drainage-associated surface elevation change between image pair acquisitions, from glaciologically physical speed change, thereby measuring the residual ice dynamic signal for each cycle of lake activity. These results improve ice velocity datasets derived from SAR satellite imagery, which are vital for monitoring changes in ice flow in Antarctica and quantifying the size and timing of the ice sheet’s contribution to global sea level rise. This work also improves our understanding of currently unresolved subglacial mechanisms and their impact on Antarctic Ice Sheet stability.

ABSTRACT:

The southeastern states are prone to frequent thunderstorms, which can produce damaging winds, hail, and tornadoes. According to the National Oceanic and Atmospheric Administration (NOAA), the southeastern states experience the highest frequency of thunderstorms in the US, and these storms have been increasing in frequency and intensity in recent decades. Additionally, the southeastern states are also vulnerable to hurricanes and tropical storms, which have become more frequent and severe in recent years due to warmer ocean temperatures. The increased frequency and intensity of severe storms in the southeast of the US pose significant risks to public safety, infrastructure, and the economy. It is essential to continue monitoring these trends and taking notes on the impacts of severe weather events. We propose a methodology that combines satellite-based proxy indicators in any weather condition even under thick cloud cover to detect damages. In particular, this study demonstrates the potential application of advanced technology using satellite Interferometric Synthetic Aperture Radar (InSAR) for mapping storm-induced floods and damages during a period of October 2019 to August 2021. One of the major storms, Hurricane Sally, happened during this period and made landfall in Alabama on September 16, 2020, causing notable damage to the state. We will use satellite images taken before and after a hurricane to identify areas that have been affected by the storm and to assess the damage to buildings, roads, and other infrastructure caused by hurricanes. In order to achieve the goal, The study identifies vulnerable areas using Sentinel-1 InSAR data before and after the storm and utilizes the interferometric radar coherence feature to detect the presence of floods in urbanized areas.Sentinel-1 InSAR data generated by the COMET-LiCSAR system was processed by the LiCSBAS processing package to obtain a surface deformation time series. Also, optical images are used to investigate soil moisture parameters and other climate changes with a time series of displacement and radar coherence extracted from SAR images. The research reports, classifies, and discusses the consequences of the hurricane for structures and highways in terms of various types of damage and warnings.Results of this research is expected to provide new techniques that can help emergency responders prioritize their efforts and resources to the areas that need help the most. Also, this technology can help in planning for repairs and reconstruction.

Keywords: Hurricane, Sentinel-1, Coherence, Alabama, Structures, InSAR

How phase information has become lost

Sentinel-1 (S1) is a Synthetic Aperture Radar (SAR) satellite that operates on routine bases both day and night independently of cloud cover, which makes it an excellent data source for monitoring changes in Earth’s surface. Nevertheless, SAR data is used by relatively small user segments, including university researchers and specific geographic information system (GIS) or earth observation (EO) companies. The use is limited because SAR data requires significant pre-processing, based on expert knowledge, before the data becomes ready for information extraction. While the Google Earth Engine (GEE) has become a key platform for large area analysis with pre-processed S1 backscatter imagery, additional pre-processing steps are recommended for many applications even there (Mullissa et a. 2021). However, pre-processing needed for SAR phase products is considerably more complicated and demands significant processing and storage capabilities. Therefore, majority of EO platforms like GEE or Sentinel Hub ignore Single Look Complex (SLC) data and consequently interferometric phase and coherence products. This is a crucial limitation for data users as one of the valuable parts of S1 data are simply ignored even though such data would benefit users globally (Kellndorfer et al. 2022).

Making repeat pass interferometric coherence data accessible to everyone

KappaZeta Ltd is dedicated to make SAR backscatter and repeat pass interferometric coherence information accessible and easy to use for a long list of expert and non-expert SAR data users. We have established the KappaOne service (KappaZeta 2023) where fully processed S1 data are prepared for users in analysis ready data (ARD) format. For both SAR backscatter and coherence imagery, the specifications for ARD are not rigorously defined and can vary by applications. Therefore, we have concentrated our effort on the configurations that suffice the widest range of applications and users. However, users who are highly aware of their specific needs regarding the SAR data set can interact with the KappaOne service to define the processing parameters that best suits to the application they aim. We have built an accurate SAR processing chain, which outputs ARD raster imagery and timeseries of parcel-based aggregated statistics. Users can access the KappaOne products via an Application Programming Interface (API), a Web Map Service (WMS) or a web-based user interface.

The ARD layers contain calibrated, noise corrected and speckle-supressed high-resolution backscatter and 6- or 12-day repeat pass interferometric coherence raster imagery in both polarisations (VH and VV), synthetic Normalized Difference Vegetation Index (sNDVI, modelled from S1 and Sentinel-2 data), and timeseries of parcel-level statistics. Both backscatter and coherence imagery are fully processed and orthorectified. To achieve the highest possible spatial resolution, the images are up-sampled to 5 m square pixels from their original 5 m x 20 m (range x azimuth) resolution.

To optimize the output raster layers and make them suitable for a wide range of applications, advanced custom filtering is used in determining a coherence estimation window and supressing speckle. A custom filter for KappaOne service is designed via combination and modification of multiple published filtering methods (Lee et al. 1999, Deledalle et al. 2014, Fracastoro et al. 2021). As a result, we can produce imagery with fine details and low speckle. This improvement in retaining the level of detail becomes especially apparent in coherence imagery in comparison with the products from standard processing with the European Space Agency’s Sentinel Application Platform (SNAP). The edges of the objects in imagery are much sharper and footprints of relatively tiny highly coherent objects in the landscape correspond better to their actual size.

Synthetic Normalized Difference Vegetation Index (sNDVI)

The most innovative among the ARD raster layers is the sNDVI, which is synthesised from S1 backscatter and repeat pass coherence timeseries and historical (within the 30-day limit) Sentinel-2 (S2) NDVI data via Artificial Intelligence (AI) modelling. Repeat-pass interferometric coherence is known to be inversely correlated to amount of vegetation and optical NDVI. Therefore, establishing a coherence derived proxy to NDVI has been proposed to fill gaps in NDVI timeseries caused by cloud cover (Bai et al. 2020). Our sNDVI model can produce promising results but it is still in experimental state. Historical S2 NDVI imagery, which serves as input to the model, is produced using our own AI-based S2 cloud mask – KappaMask. Our free and open source cloud mask is ranking at the top of the most reliable S2 cloud masks (Domnich et al. 2021, Aybar et al. 2022).

Timeseries of parcel-based aggregated backscatter and coherence statistics

In addition to, or alternative to, the ARD raster layers, timeseries of parcel-level statistics (incl. intraparcel min, max, mean, median, standard deviation) for VH and VV backscatter, VH/VV backscatter ratio, VH and VV 6- or 12-day repeat pass coherence are available. Usefulness of S1 parcel-level timeseries has been shown in various applications (Tamm et al. 2016, Tampuu et al. 2021), whereas production of a database of parcel-level temporal signatures instead of an image stack saves the data users from the burden of processing, extraction and storage of large volume of SAR data (Kumar et al. 2022). While many applications just do not need a pixel-based approach, there are others where aggregation of SAR pixels aimed to representing the target as a whole and reducing the influence of randomness of individual pixel values is advisable (Millard 2016).

KappaOne: advanced EO platform

The KappaOne service is based on the expert knowledge on SAR image processing, interferometry and AI. The KappaOne processing chain is built on SNAP, integrated with the customised functionalities as noise correction, calibration, advanced speckle filtering and coherence estimation. Fully processed SAR ARD products are made available to disseminate usage of SAR data among various user groups. Coherence ARD products save the users from the burden of processing, allowing easy adoption of interferometric products in any application or by any user. The capability of KappaOne to output parcel-level timeseries of statistics may significantly benefit various applications. The solid physical bases of the processing ensure the KappaOne output products are highly accurate and of the best value to the expert or non-expert data user.

References

Aybar, C., Ysuhuaylas, L., Loja, J., Gonzales, K., Herrera, F., Bautista, L., ... & Gómez-Chova, L. (2022). CloudSEN12, a global dataset for semantic understanding of cloud and cloud shadow in Sentinel-2. Scientific data, 9(1), 782.

Bai, Z., Fang, S., Gao, J., Zhang, Y., Jin, G., Wang, S., ... & Xu, J. (2020). Could vegetation index be derive from synthetic aperture radar?–the linear relationship between interferometric coherence and NDVI. Scientific Reports, 10(1), 1-9.

Deledalle, C. A., Denis, L., Poggi, G., Tupin, F., & Verdoliva, L. (2014). Exploiting patch similarity for SAR image processing: The nonlocal paradigm. IEEE Signal Processing Magazine, 31(4), 69-78.

Domnich, M., Sünter, I., Trofimov, H., Wold, O., Harun, F., Kostiukhin, A., ... & Cadau, E. G. (2021). KappaMask: Ai-based cloudmask processor for sentinel-2. Remote Sensing, 13(20), 4100.

Fracastoro, G., Magli, E., Poggi, G., Scarpa, G., Valsesia, D., & Verdoliva, L. (2021). Deep learning methods for synthetic aperture radar image despeckling: An overview of trends and perspectives. IEEE Geoscience and Remote Sensing Magazine, 9(2), 29-51.

KappaZeta Ltd (2023). KappaOne: Sentinel-1 Analysis Ready Data. https://kappaone.eu/ard_landing/ (Accessed 16.03.2023).

Kellndorfer, J., Cartus, O., Lavalle, M., Magnard, C., Milillo, P., Oveisgharan, S., ... & Wegmüller, U. (2022). Global seasonal Sentinel-1 interferometric coherence and backscatter data set. Scientific Data, 9(1), 73.

Kumar, V., Huber, M., Rommen, B., & Steele-Dunne, S. C. (2022). Agricultural SandboxNL: A national-scale database of parcel-level processed Sentinel-1 SAR data. Scientific Data, 9(1), 402.

Lee, J. S., Grunes, M. R., & De Grandi, G. (1999). Polarimetric SAR speckle filtering and its implication for classification. IEEE Transactions on Geoscience and remote sensing, 37(5), 2363-2373.

Millard, K. (2016) Development of methods to map and monitor peatland ecosystems and hydrologic conditions using Radarsat-2 Synthetic Aperture Radar (Doctoral dissertation, Carleton University).

Mullissa, A., Vollrath, A., Odongo-Braun, C., Slagter, B., Balling, J., Gou, Y., ... & Reiche, J. (2021). Sentinel-1 sar backscatter analysis ready data preparation in google earth engine. Remote Sensing, 13(10), 1954.

Tamm, T., Zalite, K., Voormansik, K., & Talgre, L. (2016). Relating Sentinel-1 interferometric coherence to mowing events on grasslands. Remote Sensing, 8(10), 802.

Tampuu, T., Praks, J., Kull, A., Uiboupin, R., Tamm, T., & Voormansik, K. (2021). Detecting peat extraction related activity with multi-temporal Sentinel-1 InSAR coherence time series. International Journal of Applied Earth Observation and Geoinformation, 98, 102309.

KappaZeta Ltd is dedicated to make SAR backscatter and repeat pass interferometric coherence information accessible and easy to use for a long list of expert and non-expert SAR data users. We have established the KappaOne service (KappaZeta 2023) where fully processed S1 data are prepared for users in analysis ready data (ARD) format. For both SAR backscatter and coherence imagery, the specifications for ARD are not rigorously defined and can vary by applications. Therefore, we have concentrated our effort on the configurations that suffice the widest range of applications and users. However, users who are highly aware of their specific needs regarding the SAR data set can interact with the KappaOne service to define the processing parameters that best suits to the application they aim. We have built an accurate SAR processing chain, which outputs ARD raster imagery and timeseries of parcel-based aggregated statistics. Users can access the KappaOne products via an Application Programming Interface (API), a Web Map Service (WMS) or a web-based user interface.

The Luțca bridge is a cable-stayed bridge in Neamț county, Romania, which collapsed on 9th of June 2022, only half a year after it was reopened in November 2021. In August 2020, the Luțca bridge over Siret River underwent major repairs after 30 years of operation. From persistent scatterer points still visible after the collapse, we notice that after the start of repair work some points started to subside, then the coherence of the time series decreases. This shows that along with a substantial change in the linear displacement, a change in the coherence of the time-series might be a sign that something is wrong.

In this work we present a methodology for detecting deformation profiles with deformation characteristics like the ones at the Luțca bridge collapse, i.e., a substantial change in the deformation slope, and/or a decrease of the time series coherence.

The proposed methodology is as follows. In the first step we remove the relevant harmonic components from the deformation profile using a zero-phase infinite impulse response filter. Then we fit a piecewise linear model with maximum four breaks. From the piecewise linear model, we extract the local deformation rate, the derivative of the deformation rate, the time series coherence, and the derivative of the coherence. We consider only the segments with deformation less than 42.6 mm/year (maximum measurable deformation rate with Sentinel-1 [1]) and on a time interval bigger than 200 days. In the last step we apply a heuristically determined decision equation. This methodology was applied to a small test are around the Luțca bridge. The result is a map depicting points with possible problems. Currently we are investigating different machine learning based algorithms for automatically finding the decision threshold and reducing the number of false alarms.

So far, in this work, we analyzed independent deformation profiles.

Anomaly detection for infrastructure monitoring using PSInSAR is not a new problem, however there is still room for improvement. Methods used so far include detection of substantial changes in liner deformation in the final part of the deformation profile, clustering profiles with similar behavior and analyzing them with statistical methods, classification (i.e., supervised learning) and so on.

The contribution in this study describes the procedure followed to validate EGMS products with Corner Reflectors (CR) deployed within the time frame of the EGMS products (2015-2021). This work is performed within the framework contract supporting the European Environment Agency’s (EEA) in the validation of the Copernicus European Ground Motion Service.

CR are one of the best ways to validate the EGMS products. CR with additional measurements, allow the evaluation of three parameters: height, location, and time-series displacements. Ideally, estimating these three parameters would be performed in a controlled environment where the CR are deployed and continuously measured with other techniques to validate Satellite interferometry derived measurements. Since there was no dedicated experiment to perform such a task in a controlled environment, the feasibility of the methodology is demonstrated with case studies where different in-situ measurements were performed.

Following the EEA requirements, we validate the EGMS products as follows:

i. Height of the MPs around the CR location: For this requirement, we use the CR with known heights derived by the levelling campaigns or Global Navigation Satellite Systems (GNSS) if levelling is not performed as ‘ground truth’. We then estimate the differences between the ‘ground truth’ (the CR) and the EGMS Measurement Point (MP) estimated heights at the location of the CR. The MPs around the CR are used to perform statistics. We assume that the differences between orthometric and geometric heights are negligible, given the small distances between CR (Marinkovic et al., 2007).

ii. Geopositioning accuracy by XY offset estimation: For this requirement, we use the measured location of the CR usually performed by GNSS at the date of the CR installation. With the accurate position of the CR, we compute the distance (offset) between the CR and the closest MP.

iii. Quality of the EGMS time-series displacements: To evaluate the quality of the EGMS time-series displacements, we use the GNSS station measurements, which are placed close to the CR. The methodology for this validation requirement is the same used for the validation of EGMS with GNSS. First, we perform temporal and spatial interpolation between the GNSS and EGMS MPs around each corresponding GNSS station. We ensure we use the same reference date for both datasets and estimate the resultant spatial interpolation error. Then we project the GNSS data to the radar line-of-sight and perform double differences for L2a products and single differences for L2b products. Finally, we perform the GNSS-InSAR comparison through time series and deformation model using the Best Linear Unbiased Estimator (BLUE).

We applied this methodology in different locations covering different deformation processes. This contribution presents the outcomes of the validation process applied to:

- subsidence due to soil consolidation and water extraction over the Thyborøn area on the west coast of Denmark;

- landslides at Jettan, Indre Nordnes and Gamanjunni regions, Norway;

- engineering works (seasonal hydraulic loads) at Calern’s multi-technical geodetic observatory, France;

- no significant ground displacements: a controlled experiment in the Netherlands.

We validate the three requirements qualitatively (by figures of time-series comparison, and offset distances) and quantitatively (by statistical testing for time-series comparison, offset estimation and corresponding accuracies [Teunissen, 2000]). The validation generates key performance indicators to evaluate the results.

Acknowledgements: The authors would like to acknolwedge Hans van der Marel (TUDelft) for providing the coordinates, heights and accuracies of the corner reflectors deployed in the Netherlands.

Marinkovic, P., G. Ketelaar, F. van Leijen and R. Hanssen (2007). InSAR quality control: Analysis of five years of corner reflector time series. Proceedings of Fringe 2007 Workshop (ESA SP-649), Frascati, Italy.

Teunissen, P. J. G. (2000). Testing theory; an introduction (1 ed.). Delft: Delft University Press.

In recent years’ groundwater over-exploitation and groundwater level decline damage humans and environment and causes land subsidence as well, which has been a problematic issue in arid and semi-arid areas such as Iran. Remote sensing technique have advantage over filed inspection measurement duo to low cost, time consuming and large scale coverage. The purpose of this study is to quantify the land subsidence in Qazvin province by using synthetic aperture radar interferometry and evaluating the effect of the groundwater depletion on this phenomenon. Qazvin plain as one of the largest agricultural areas in Iran was selected as a case study, since its experience both groundwater declines as well as subsidence. In this study the Interferometric Synthetic Aperture Radar (InSAR) technique used to estimate subsidence by using Envisat, Alos palsar-1, and Sentinel-1 satellite data between 2003 to 2017. Water table variation of Qazvin’s aquifer was studied using 180 data points of the pizometric wells. Annually averaged land-subsidence in this years was obtained as 39.9 mm/year for aquifer zone and this value was 33 mm/year for Qazvin province. According to the land-subsidence zone in Qazvin province it was revealed that most of the land-subsidence occur in the region of the aquifer whose fine-grained layer thickness would be larger than other areas. The maximum of Land subsidence was obtained at the northern parts of Buin-Zahra and near the Takestan borderline. This area has the highest cultivated area and groundwater depletion. The results of this study showed a strong correlation between the groundwater water table variations and land subsidence values in Qazvin province.

Agroecosystems are complex ecological systems that involve agricultural practices and the environment. One of the key components of a healthy agroecosystem is crop diversity, as it helps increase soil fertility, improve soil health, and reduce the risk of crop failure. However, crop diversity can be negatively impacted by soil erosion, which is a major challenge facing Romanian agricultural communities.

The purpose of this study is to analyze multi-temporal Sentinel-1 data to evaluate the agroecosystem status in Timis County, particularly at Emiliana Farm. The test site is located in the western part of Romania and has a moderately continental temperate climate with Mediterranean influence, characterized by weak mild winters and hot summers, with an average annual temperature of 10.8 °C and mean yearly rainfall of 550 mm. From a morphological point of view, the relief is flat with a uniform appearance but heterogeneous in terms of lithology and soil. Flat surfaces are frequently separated by abandoned meanders. Previous studies have shown that villages and road infrastructure are prone to subsidence phenomena induced by water infiltration.

The coherence of a time series of dual-polarized Sentinel-1 imagery is investigated for vegetation state monitoring based on land use land cover classes. The Synthetic Aperture Radar (SAR) data have been acquired in ascending mode between March 2018 to September 2021, with VV polarization, 103 orbit cycle, 102 relative orbit, at an incidence angle of 38⁰. The test site contains maize, wheat, sugar beet, sunflower and successive crops. Interferograms and coherence images were generated using single and dual-polarimetric data. Polarimetric interferometry (PolInSAR) coherence describes physical properties of various targets: man-made targets (villages) show high coherence magnitude while agricultural areas suffer from temporal and volume decorrelation due to seasonal changes and exhibit lower coherence. We also investigated the sensitivity of the radar information to the classification methods like Support Vector Machine and Random Forest. The results highlight that a small improvement in the classification accuracy can be achieved by using the coherence in addition to the backscatter intensity and by combining co-polarized (VV) and cross-polarized (VH) information. It is shown that the largest contribution to class discrimination is observed during winter when dry vegetation and bare soils are present.

The study demonstrated that the Sentinel-1 data can help monitor agroecosystems in Timis County and support decision-making for improving crop yields and reducing soil erosion. The study also highlighted the importance of crop diversity and soil conservation techniques in promoting healthy agroecosystems.

The Tianshan orogenic belt (TSOB) is one of the most active regions in Eurasia. The far-range effect of the collision between the Indian and the Eurasian plates in the late Cenozoic led to the reactivation of the TSOB and the occurrence of intracontinental orogeny. At the same time, the TSOB expanded to the foreland basins on its both sides, forming multiple rows of décollement- and fault-related fold belts in the basin-mountain boundary zone. Global Positioning System (GPS) observations show that the shortening rate in the north-south direction across the TSOB gradually decreases from ~ 20 mm/yr in the west to ~ 8 mm/yr in the east. However, how the deformation is distributed inside the TSOB is controversial. Here, we determine the present-day kinematics of the major structural belts based on the Interferometric Synthetic Aperture Radar (InSAR) data of the Sentinel-1 satellites.

We process Synthetic Aperture Radar (SAR) data from 5 ascending tracks (T27;T129;T56;T158;T85) and 4 descending tracks (T107;T34;T136;T63) of the Sentinel-1A/1B satellites recorded between November 2014 and December 2020. We constructed a total of 1074 single-reference single-look interferometric pairs based on Gamma software covering a 790-km-length and 520-km-width area of the TSOB. Finally, the InSAR time series are processed using the StaMPS software package. The long-wavelength and elevation-dependent atmospheric errors from each date are mitigated using the TRAIN package and ECWMF ERA5 models.

Combining InSAR and GPS measurements, we show that the tectonic deformation is not evenly distributed in the TSOB. The convergence across the Tianshan ranges is approximately 15–24 mm/yr; the deformation gradient in the junction area between South Tianshan and Pamir is the largest and adjusts ∼68% of the total convergence deformation. South Tianshan is relatively stable without sharp gradients, and the remaining deformation is distributed in the intermontane faults and basin systems in the north of South Tianshan. We also find that the Kashi fold-thrust belt is the most active unit in this area, and the deformation is mainly concentrated on a series of folds: the Mushi, Kashi, and Atushi folds, and the faults between the folds, such as the Kashi, Atushi, and Toth Goubaz faults. As the boundary fault between the South Tianshan and the Tarim basin, the Maidan fault shows a clear deformation gradient. In the Keping nappe, the deformation is mainly concentrated on the Keping hill and Kepingtag fault in the front of the nappe. There are several remarkable deformation zones in the Kuche foreland. The deformation in the north of South Tianshan is dispersed in a series of intermountain active structures and the depression basins, unlike in the south side, where the deformation is mainly concentrated on the thrust folds. Furthermore, our study can provide constraints for deformation and slip partitioning patterns associated with the ongoing India-Eurasia collision in the TOSB.

Over the last 20 years, the former freight station in Frankfurt am Main, Germany, has been developed into a new urban district: the Europaviertel. In 2017, construction of an extension to the existing U5 subway line began to connect the new neighborhood to the existing public transport network. The new tunnel includes sections built with both cut-and-cover and underground tunnel boring machine approaches, as well as underground stations.

The geology under Frankfurt is a mix of clay, sandstone and gravel, which often form lenses, as well as surface faults. A large part of the underground route runs through clay, overlaid with several quaternary layers of sandstone and gravel up to 2-10m thick. From January 2019, the area was dewatered and the groundwater lowered.

Here we present the results of a historical Sentinel-1-based analysis of the displacements that occurred during the tunneling activities and compare them to the dewatering levels as well as ground-based observations. We observe a clear correlation between the amount of dewatering that occurred for construction and the displacements observed in the InSAR results, as well as with the results of ground-based observations. Furthermore, local subsurface geological structures have a strong impact on the distribution of the surface displacements, enabling us to refine their presumed locations. Lastly, we also highlight a location that exhibited displacement patterns inconsistent with the temporal and spatial effects of dewatering.

Our results show that InSAR is a powerful complimentary tool for monitoring displacements associated with dewatering for tunneling activities and differentiating between pre-existing movement patterns and those resulting from construction. Combined with our understanding of the geological structures, we can map permeability distributions in the underground and guide dewatering activities while they are being performed to reduce structural damage

Understanding how the presence of fractured ice alters the dynamics, hydrology and surface energy balance of glaciers and ice shelves is important in determining the future evolution of the Antarctic Ice Sheet (AIS). However, these processes are not all well understood, and large-scale quantitative observations of fractures are sparse. Fortunately, the large amount of sythetic-aperture radar (SAR) data covering Antarctica gives us the opportunity to change this.

The Sentinel-1 satellite cluster has acquired SAR data over the AIS with a repeat period of 6-12 days for the last 8 years. Due to the coherence of scattered microwaves and their penetration through the upper snowpack, a broad range of crevasse types are visible in this imagery: rifts; surface crevasses (and some basal crevasses on ice shelves; and fine surface crevasses on grounded ice streams - even those bridged by snow or pixel-scale in width.

In this study, we use machine learning to automatically map crevasses directly from geocoded single-look-complex amplitude images, acquired using the interferometric-wideswath (IW) mode of Sentinel-1; producing monthly composite maps over the AIS at 50m resolution. We developed algorithms to partition crevasses into those on grounded and floating ice, and extract these features in parallel using a mixture of convolutional neural networks, trained in a weakly supervised way, and other computer vision techniques designed to exploit the spatial structure of the crevasse fields.

Having developed parallelisable routines for the large-scale batch processing of SAR data, we have processed every Sentinel-1 acquisition over the Antarctic Ice Sheet. The resulting dense timeseries of fracture maps allows us to assess the evolution of crevasses during the Sentinel-1 acquisition period. In particular, we developed methods to quantify changes to the structural integrity of floating ice shelves. This is done by measuring trends in the density of fractures, aided by the use of local statistical properties of the radar backscatter signal to remove contributions to the fracture density timeseries arising from the effect of surface ice conditions on crevasse visibility. On application of this method to the ice shelves of the Amundsen Sea Embayment, West Antarctica, we show an increase in crevassing over the last 8 years ­­in areas thought influential for the dynamical stability of the region.

Nepal has been subjected to a phenomenon of significant surface displacement due to natural as well as anthropogenic causes for a long time. The natural causes include the massive earthquake of 25^th April 2015 triggering a substantial uplift around Kathmandu and the tectonic movement of the Eurasian plate toward the Tibetan plate. However, even in absence of any such natural cause, the areas inside Kathmandu Valley have been exhibiting a perceptible magnitude of surface displacement. Previous studies till 2017 have demonstrated subsidence, with rates of several centimeters per year, occurring in the Kathmandu Valley indicating uncontrolled groundwater withdrawal as the major cause of subsidence.

This study aims at detecting the nature of surface displacement in Kathmandu and its surrounding for three years: 2015 (23^rd January to 8^th September), 2017 (18^th January to 26^th December), and 2019 (2^nd January to 28^th December) based on Persistent Scatterer Interferometry (PSI) technique using Synthetic Aperture Radar (SAR) datasets from Sentinel 1. The nature of displacement refers to whether the area is subsiding or uplifting, and what is the trend of the displacement that has been demonstrated in this study with time series plots. PSI is able to detect persistently backscattering targets and evaluate respective displacements from the backscattered signal. The study presents the abrupt displacement that occurred due to the massive earthquake of 2015 along with the other gradual surface displacements that occurred in the years 2015, 2017, and 2019. The results indicated that there was a significant uplift of up to 1.134 m along the Line of Sight (LOS) of radar in the study area for the year 2015. The results of 2017 and 2019 revealed significant displacement of -100.54mm and -129.19mm along the Line Of Sight (LOS) of radar during the study period at Baluwatar and Lazimpat area of Kathmandu district respectively. Likewise, New Baneshwor, Satdobato, Bode, and Imadol demonstrated displacements of -92.59 mm, -103.55 mm, and -125.62 mm respectively for the year 2017. Similarly in the year 2019, New Baneshwor, Bode, and Imadol exhibited a substantial displacement of -88.81mm,-103.55mm, -127.35mm respectively. Thus, this study was able to detect the displacement occurring in the Kathmandu Valley.

Glacier velocity is an important parameter that provides insight into the dynamic behavior of glaciers and their response to climate change. The NASA MEaSUREs Inter-mission Time Series of Land Ice Velocity and Elevation (ITS_LIVE) project provides global glacier surface velocities using Sentinel-1/2 and Landsat-8/9. However, the accuracy of glacier velocity obtained from ITS_LIVE V2.0 has yet to be fully validated for mountain glaciers. Therefore, it is important to compare it with ground-based measurements to assess its reliability. In this study, we intend to validate the ITS_LIVE V2.0 against publicly available in-situ GPS data for two typical locations: the Argentière and Mer de Glacier.

The Argentière Glacier (Figure 1), located in the Mont-Blanc mountain range of the French Alps, had a surface area of around 10.9 km² in 2018. It spans about 10 km in length and stretches from an altitude of approximately 3,400 m a.s.l. at the upper bergschrund down to 1,600 m a.s.l. at the snout. The GLACIOCLIM program, which is the French glacier-monitoring initiative, provided the field observations of the Argentière Glacier, including mass-balance, thickness variations, ice-flow velocities, and length fluctuations over the past 50 years. In addition to GPS data from four specific location points spanning from 1976 to 2020, we have also acquired Tour Noir GPS data from 2007 to 2020. The glacier velocity derived from ITS_LIVE V2.0 at Argentière Glacier (blue cross marker) was shown in Figure 2a. The Argentière glacier velocity is 0-300 m/yr with seasonal variations.

Mer de Glace (Figure 1) is the largest glacier in the French Alps, covering an area of 32 km². Its upper accumulation area rises to approximately 4300 m a.s.l. and feeds into the lower 7 km of the glacier, which descends rapidly through a narrow, steep icefall between 2700 and 2400 m, terminating at a front of about 1500 m. The glacier includes multiple tributaries, and it has been the subject of numerous glaciological and geodetic measurements. The GPS data is available from 2008 to 2020 at the Leschaux branch, from 1996 to 2020 at the Tacul-langue branch, and from 2008 to 2019 in the Talefre branch. The glacier velocity at Mer de Glace (red cross marker) was shown in Figure 2b. The Mer de Glace velocity ranges from 0 to 400 m/yr with seasonal variations.

We used GPS measurements to obtain precise displacements of the ground surface at various locations and time periods. The corresponding ITS_LIVE V2.0 data will be extracted for the same locations and time periods. The two datasets were compared using a variety of statistical metrics, including Root Mean Square Error (RMSE), mean bias, correlation coefficient, and scatter plots. If the ITS_LIVE V2.0 glacier velocity resolution of 120 m is insufficient for mountain glaciers, this work will rerun offset-tracking - autoRIFT with parameters setup to generate glacier velocity with higher spatial resolution.

Our study will provide valuable insights into the accuracy of ITS_LIVE V2.0 data over mountain glaciers with high topographic relief and its potential applications in cryosphere remote sensing. The GPS measurements are necessary for detecting minor and temporary changes in velocity, while remote sensing estimates are more beneficial for determining overall patterns in velocity trends. To ensure the reliability of the ITS_LIVE V2.0, we will expand the validation process for different locations and time periods in the future.

Improvements in the resolution of SAR images together with the development in multi-temporal InSAR methods such as PS and SBAS have extended the application of satellite-based remote sensing for monitoring traffic infrastructures such as bridges, railway tracks and highways. Nevertheless, monitoring linear infrastructures with multitemporal InSAR remains a challenging task due to the narrow spatial extent of the target. Linear infrastructures are long and narrow and they are only covered by a few pixels in width. As a general approach in MTI-InSAR to address atmospheric artefacts and phase unwrapping, large areas beyond the extent of the linear infrastructure are first needed to be processed to derive regional displacement field in the study area using all coherent pixels. However, most of the pixels within this area are not of interest in the context of linear infrastructure monitoring, as they correspond to e.g. urban areas. Therefore, the resulting displacement field needs to be intersected with a buffer zone around the linear infrastructure to discard all non-relevant pixels outside the buffer. This common approach has a high computational burden as all coherent pixels need to be unwrapped. Moreover, a major limitation in InSAR is the propagation of errors in the phase unwrapping step, which degrades the accuracy and reliability of the resulting deformation time series. Therefore, including pixels from outside the linear infrastructure can lead to the propagation of errors to the linear infrastructure. An obvious solution to these two drawbacks is limiting the InSAR time series analysis to the pixels on the linear infrastructure. But this is not feasible as a reliable estimation of the atmospheric phase contribution requires a homogeneous spatial sampling over the area of interest and is not necessarily given by merely the pixels on the linear infrastructure. Hence, monitoring linear infrastructures efficiently and reliably requires an InSAR time series method tailored to this task.
In this contribution, we address the above identified drawbacks in high computational time and error propagation by proposing a new InSAR time-series methodology that has been tailored to the monitoring of linear traffic infrastructures. Our time series approach is based on a stack of single-look interferograms from a redundant interferogram network with small temporal baselines. The phases are unwrapped in space per interferogram and the coherent pixels are selected using a fast a priori assessment of the phase noise from the interferogram stack by spatial filtering. We estimate the deformation time series in a two-step procedure. First, the atmospheric phase screen (APS) is estimated from a sparse set of first-order pixels with a high signal-to-noise ratio. These first-order pixels are selected carefully by removing outliers and are homogeneously distributed over the area of interest to ensure valid sampling of the APS. Second, we remove the APS from the final dense set of pixels and also unwrap their phase in space and invert the network of interferograms to retrieve the phase time series. Different to previous approaches, we select the final dense set of pixels merely among the pixels on the linear infrastructure. Due to the two-step approach, the final pixel density can easily be adapted in the second step by altering the threshold for the pixel selection.
We perform experiments with both real and simulated datasets to validate our approach and compare its performance with respect to standard methods implemented in SARScape and StaMPS in terms of computational time and difference in the resulting deformation map and time series. The experiments are performed on a stack of Sentinel-1 images from Jan. 2017 to Jan. 2019 over a study area in Germany covering the open-pit mine Hambach which shows strong subsidence also on the surrounding highway and railway tracks. First results show differences within the measurement noise between our approach tailored to linear infrastructure monitoring and the standard approach which processes all coherent pixels. However, the computational time of our approach is significantly reduced from a few hours to a few minutes processing time. Our experiments show the validity of our approach and, hence, our InSAR time series approach paves the way for continuous monitoring of linear infrastructures based on Sentinel-1 data.

Time series interferometric synthetic aperture radar (InSAR) can be significantly affected by the ionosphere, limiting its capability to measure long spatial wavelength deformation, especially for the L-band low-frequency SAR, such as ALOS-2, LuTan-1, and the forthcoming NISAR and ROSE-L. Due to the dispersive nature of the ionosphere with respect to the microwave signal, the propagation of the radar signal traveling through the ionosphere results in a group delay and a phase advance. The two ionospheric contributions are equal in magnitude but opposite in sign, based on which the group-phase delay difference method is proposed to measure the relative ionospheric phase via the combination of speckle tracking and interferometry (Meyer et al., 2006, GRSL; Brcic et al., 2011, IGARSS).

Compared with the range split-spectrum method, the group-phase delay difference method has the following advantages: 1) it’s more accurate theoretically; 2) it’s potentially more robust in practice since it does not need to unwrap the subband interferogram; 3) if coregistration was carried out using cross-correlation, the range offset can be re-used, thus, more computationally efficient. These advantages make the method desirable for operational big data processing.

Here I extend the existing group-phase delay difference method to the InSAR time series. I present an algorithm to estimate the time series of ionospheric phase delay, which can be used to correct the InSAR time series of deformation. Preliminary result shows a good agreement with the split-spectrum method (Liang et al., 2019, TGRS) using Sentinel-1 data over northern Chile.

Future work includes 1) testing Sentinel-1 data over southern California against independent GNSS network observations; 2) testing ALOS-2 data over Kyushu, Japan against the split-spectrum method (Fattahi et al., 2017, TGRS); 3) evaluating the performance of the even faster Global Ionospheric Maps (GIM) method (Gomba & De Zan, 2017, TGRS) for interseismic secular deformation mapping from InSAR time series.

The use of multi-temporal Interferometric techniques, and specifically of the Small BAseline Subset (SBAS) method for building a network of ultimate combination of interferograms, is widely known and adopted for the monitoring of slow surface displacements. On the other hand, applying the SBAS method for long-term monitoring is a challenging task in areas with intensive underground mining where the surface response has fast rate (0.5-1.5 m/year) and a pattern with multiple sparsely distributed patches of deformation at smaller scale (~200-300m). Such is the case of surface deformations in Southern Poland where one of the biggest European coal deposits is located in the so-called Upper Silesian Coal Basin (USCB). The coal extraction in USCB is done mainly by the usage of long-wall technology for which the deposit is exploited in parallel, in horizontal and vertical position prolonged galleries, as the works follow horizontal direction. In this way, the surface subsidence follows the pace of the works and the appearance of the subsidence bowls have non-linear spatial and temporal behaviour. Another complication related to the analysis of SAR data over this mining area is the mostly rural land cover, which could cause a signal temporal decorrelation. All these characteristics impose additional threats to the unwrapping and modelling processes. Several new functionalities included in the last version of the SARscape software as layover and shadow masking, as well as a reduction of the atmospheric noise by application of external water vapor data as GACOS, and automated selection of the appropriate inteferometric pairs based on the statistic parameters and presence of unwrapping discontinuity, improve the SBAS processing. The current study is based on 3-years ascending and descending Sentinel-1, C-band data (2018-2020) over USCB. Time series of deformation obtained from the SBAS workflow are additionally analysed to classify the regions with different behaviour – linear, periodical or quadratic – depending on the changes in the acceleration at the edges of the moving subsidence bowls. The gained knowledge aims to support the decision-making processes and infrastructure protection actions in the mining areas. Moreover, the displacement maps of the subsidence bowls are modelled through the analytical equations for a tensile dislocation in an elastic half-space for stacked period of 1 month, equal to the rhythm of panel extractions. The goal is to assist the prediction of the extraction influence, starting from the surface fields of deformation measured from Sentinel-1 data. The classical modelling and prediction procedure applied now by most of the mining companies rely on in-situ, mainly levelling, data with, in the best case, monthly frequency up to measurements twice per year, implemented in Knothe-Budryk prediction algorithm. We propose an improved approach that targets enhancement of the assessment of the hazard in the mining areas based on more frequent and spatially distributed input data.

A floating roof tank is a storage medium typically used for volatile liquids, such as crude oil. The roof on top of the tank moves vertically as the volume of oil changes to reduce evaporation loss. Since these storage tanks often have large dimensions, we can see them on freely available satellite imagery, such as the one acquired by the Sentinel-1 Synthetic Aperture Radar (SAR). We typically distinguish three bright pixels in the SAR amplitude of a Single Look Complex (SLC) image of a storage tank. They appear aligned on the same image row with increasing column index according to range (distance to the satellite):
- (A): The corner formed between the platform on top of the tank and the tank façade, i.e., the fixed roof corner.
- (B):The corner between the tank façade and its base, i.e., the fixed base corner.
- (C):The corner between the inner wall of the tank and the horizontal floating roof, i.e., the floating roof corner.

When looking at an aligned time series of SAR images, the fixed roof (A) and fixed base (B) corners remain in the same position. Conversely, the floating roof corner (C) moves by a few pixels from one date to another corresponding to a height change in meters. Therefore, previous methods were developed to convert the floating roof column index at a certain date into a crude oil volume (or a normalized "fill ratio" in [0,1]) for the storage tank. On the other hand, Interferometric Synthetic Aperture Radar (InSAR) techniques have demonstrated their efficacy in estimating millimetric surface deformation. Among the algorithmic developments throughout the years, we distinguish the Persistent Scatterer (PS) approach, which restricts the analysis to a group of stable reflectors. A double-phase difference on reflectors p and q for images i and j can be defined. During PS processing, it is estimated on two nearby reflectors to mitigate the atmospheric effects. Therefore, we test the same strategy to derive InSAR measurements between fixed reflectors on the tanks and assume that the deformation of the tank will be the predominant signal in the double-phase difference. Thus, we hope to measure small millimetric movements of the fixed reflectors between two dates, which may indicate crude oil volume change. In this article, our main contributions are establishing a correlation between InSAR measurements and tank fill ratio and presenting a novel InSAR use case which could motivate the development of adapted InSAR techniques.

Our Area Of Interest (AOI) contains NT = 19 tanks in the Juaymah tank farm in Saudi Arabia. (lon=49.987°, lat=26.819°). We selected orbit 101 of Sentinel-1 and dates between 2017-01-05 and 2021-12-22. In total, we recovered NI=151 images. We selected the first image as the primary and generated aligned crops of size 512 x 1024 around our AOI using a procedure based on the geolocation of a set of Digital Elevation Model (DEM) points from the Shuttle Radar Topography Mission (SRTM).
We also estimated an orbital phase and a topographic phase per image (relative to the primary image). Consequently, double-phase differences can be defined on compensated images.
An estimation of the fill ratio in [0, 1] for each tank k and each image i, was provided by the company Kayrros.
The double difference of the fill ratio between two tanks and two dates can also be defined. We compared the values of the double phase difference on the roof corners (A) on two tanks against the double difference of the fill ratio. The experiments were conducted on the set T of neighbouring tanks according to a distance threshold (here, 300 m). The image couples were also selected in a set S such their temporal separation is less than a temporal threshold (here, 90 days). The double-phase difference is taken on the roof for all tank couples in T and all image couples in S. It is plotted against the double difference of the fill ratio. We can see a trend suggesting a negative correlation between the two quantities. This trend is present in approximately half of the tank couples. The plot suggests that the double-phase difference is mostly already unwrapped. Therefore a tank filling up induces a fixed roof movement away from the satellite in the order of 1 cm. This relationship is not verified for the other half of tank couples.
Furthermore, no clear trend emerges when using the fixed base reflectors (B). We posit that this may be caused by the small reflections from the top of the floating roof, which often contaminate the base (layover effect). On the other hand, we observed several remarkable factors, such as a dependence of the double-phase difference on the orthogonal baseline for some tank couples, indicating an uncompensated topographic term, or an occasional dependence on time, with some seasonal effects. We also notice that the noise in the scatterplots increases when the corner is not a persistent scatter according to traditional metrics.

We conclude that we sometimes observe a correlation between the double difference of the fill ratio and the double-phase difference at the fixed roof corner of the tank. We listed some difficulties which suggest the need to develop further adapted InSAR techniques to this specific use case.

Slope movements are one of the most important geological hazards that affect infrastructures. The village of Castril, in the province of Granada (southern Spain), is located at an altitude of 890 m next to the Castril river talweg, on steep slopes affected by landslides. The village is built on Quaternary rocks that overlay a thrust sheet system made of Mesozoic and Cenozoic carbonates and marls. The hazard for slope movements is conditioned by abundant fault planes with fault gauges and breccias, and periodical heavy rains that affect the region. The Portillo dam, located just 800 m upstream of Castril, is a loose materials dam with a height of 80 m and a crest length of 370 m. It allows the storage of about 33 hm³ of water in the Portillo reservoir, with a surface area of 143 ha. The risk involved in the landslide of the slope on which Castril is located is significant both for the riverbed and for the dam itself. Firstly, there is a risk that the material on the hillside will displace towards the river, which could cause flooding and damage to homes as well as nearby infrastructure. Secondly, the slope movement observed in Castril village could become a major problem for the water supply and downstream evacuation infrastructures. Satellite radar interferometry (InSAR) allows the detection of horizontal and vertical ground displacements at the millimeter level, which is useful for monitoring geological hazards, including landslides. It is a less expensive and more efficient alternative to traditional ground-based monitoring techniques, which require the installation of a large number of sensors to cover large areas. Multi-temporal MT-InSAR techniques are able to monitor the temporal evolution of ground motion, especially useful in areas with continuous and slow movement over time. Using Sentinel-1 data, it can be seen that the Portillo dam, with almost 25 years of service, shows settlements of the structure with values in the order of 1 cm/year. On the other hand, the hillside where the village of Castril is located shows a continuous landslide in the direction of the river bed with values close to 1 cm/year, affecting half of the town. This case study from SIAGUA project highlights the importance and use of these satellite techniques for monitoring these infrastructures. It emphasizes the necessity of ensuring the safety of the dam and the population living downstream taking measures to stabilize the continuous movement of this slope for preventing future landslides.

The Nakdong River Deltaic Plain is composed of the thickest soft ground layer in South Korea. National land development plans have led to reclamation operations in this area, which are now used for various purposes including residential, commercial, and cultural, as well as industrial facilities such as ports and factories. Despite improvements in civil engineering to prevent soft ground subsidence through terrestrial surveys, soil testing, and subsidence calculations during the reclamation, subsidence continues due to the thick clay layer that can exceed 50 meters and the consolidation caused by heavy landfill loading. This subsidence causes great damage to human and material resources and costs a lot of infrastructure maintenance. Thus, continuous observation is essential to manage subsidence and mitigate possible damages. Traditional surveys such as continuous global navigation satellite system (GNSS) stations or terrestrial leveling surveys have been utilized. Although they have high temporal resolution and can observe surface deformation very precisely, it is difficult to observe subsidence occurring in a wide range due to their sparse spatial resolution. Exploiting Synthetic Aperture Radar Interferometry (InSAR), ground subsidence that occurs over a wide area can be monitored efficiently regardless of temporal and spatial constraints. The advanced InSAR technology, multi-temporal InSAR (MT-InSAR), is a method that can effectively separate the phases such as atmospheric phase delay, height error, and noise from the deformation phase. Persistent scatterer interferometry (PSI) is an approach using a spatiotemporally stable scatterer (persistent scatterer; PS) and is particularly effective in areas with lots of artificial structures or rocks. However, since subsidence due to consolidation in the soft ground often occurs non-linearly, there are limitations to the PSI technique which estimates surface deformation by linear fitting model. In this study, we aim to observe ground subsidence in the Busan coastal reclaimed land in South Korea from 2014 to 2021 using the PSI approach with multi-frequency SAR imagery acquired by the X-band COSMO-SkyMed, the C-band Sentinel-1, and the L-band ALOS-2 PALSAR-2 missions. To validate the results, we utilize GNSS station data and compare them with the PSI results obtained from ALOS PALSAR SAR acquisitions from 2007 to 2011 using the hyperbolic model of non-linear subsidence in soft ground.

More capable Sentinel-1

Sentinel-1 is a powerful data factory. No other current SAR mission produces data with systematic global coverage in such a large quantity. However, its information content is relatively limited – dual-polarisation backscatter and repeat-pass interferometry data. Across-track interferometry is not feasible with Sentinel-1 due to temporal decorrelation (6 or 12 days) and short interferometric baselines (<100 m) (Potin et al. 2019). The limited information content of Sentinel-1 sets an inherent limit to applications built on Sentinel-1.

There is a way for increasing the value of the Sentinel-1 mission significantly with an additional investment constituting just a fraction of what a new SAR satellite would cost. Currently, Sentinel-1 produces two-dimensional imagery about the Earth's surface. By adding relatively small and simple passive receiver satellites to the mission, it is possible to add the height dimension to the Sentinel-1 data and make it literally 3-dimensional (3D). Providing height information about the terrain, vegetation, and ice sheets at 12-day intervals would be unprecedented, all while cleverly augmenting the existing Copernicus program infrastructure and data.

Concept for safe low-cost implementation of Sentinel-1 passive companions

We propose a constellation of three passive receiver satellites forming a tandem with Sentinel­‑1. Having three passive receiver satellites allows simultaneous along- and across-track interferometry data takes with reconfigurable baselines from 400 m to 4 km (corresponding to the height of ambiguities from 50 m to 5 m).

The mission will be developed following low-cost New Space approach as much as possible, using only passive receivers and following the example set by Finnish Iceye (Ignatenko et al. 2020). Critical requirements for the success of the companion mission are a light deployable antenna, accurate orbit station keeping, Attitude Determination and Control System (ADCS), and enough electrical power for data downlink.

The innovation risk of the proposed companion mission is considered relatively low due to the comprehensive research heritage of the German Aerospace Center (DLR) and the rest of the scientific community. The success of TanDEM-X mission (Krieger et al. 2007) proved many of the technology necessary for implementing the mission we propose and the applications of its data. The applied research carried out to facilitate the planned TanDEM‑L mission (Moreira et al. 2015) is also, to a large extent, reusable. Many of the applications planned for TanDEM-L could also be possible with our proposed mission, but just with a fraction of its cost thanks to cleverly re-using and augmenting the existing Copernicus Sentinel-1 infrastructure. Significant synergies could be established with the planned ESA Harmony mission (López-Dekker et al. 2019), which has been recently approved as Earth Explorer-10. While Harmony is purely a cutting-edge EO science experiment, the goal of our proposed mission is to become a systematic data factory for operational applications for decades to come.

The collision risk with Sentinel-1 is minimised with a clever configuration of the orbit. The passive receivers will be placed behind the Sentinel-1 satellite following it at a safe distance. Fine baselines, necessary for the across-track and along-track interferometry applications will be built between the passive receiver satellites minimising the risk for Sentinel-1 – the main or parent satellite.

The specifications of the planned data products will be largely defined by the specifications of the Sentinel-1 main mission and the planned Sentinel-1 Next Generation (S1-NG) (Zonno et al. 2021).

Enablement of a variety of applications with improved accuracy and sensitivity

The aim of our proposed companion mission is to provide a rich bistatic SAR dataset on regular basis for direct applications and Artificial Intelligence (AI) modelling based Earth Observation (EO) solutions on a global scale at unprecedented temporal resolution. The mission would enable measurements of the terrain height, tree and other vegetation height, canopy density and structure with the same spatial resolution as Sentinel-1 (5 m x 20 m).

Physical remotely sensed height measurements would considerably increase the accuracy of above-ground biomass measurements compared to what is currently possible. This is essential for the global climate goals, the European Union (EU) Green Deal and for moving towards an accurate digital twin of the Earth. It is difficult to imagine a digital twin of the Earth without an up to date vegetation height data. A Sentinel-1 3D mission could to be one of the most feasible ways to achieve it with global coverage and timely updates.

The transition from an oil-based economy to bio-materials based circular economy has already started. Biomass will be an ever more critical resource for fuel and raw materials in the manufacturing industry. Mapping and monitoring of forest and agricultural biomass for the bio-based circular economy, recording early signs of crop damage and managing the associated risks could be made richer and more accurate with the planned data set. Additional linearly independent input features in the Sentinel-1 data package mean higher accuracy AI models and better predictability of the resource supply. TanDEM-X interferometric coherence or the digital surface models (DSM) extracted from the phase centre height have been previously proven useful for forest mapping (Schlund et al. 2014) and forest height estimation (Olesk et al. 2015, Bispo et al. 2019).

The addition of the height dimension to the SAR data would open avenues for EO-based carbon stock inventories in national states and for carbon trading projects. Voluntary carbon markets are already a multi-billion dollar business which is predicted to continue to grow rapidly. However, the sector is bogged in lack of transparency, and there is an increasing need to quantify carbon removals, i.e. biomass accumulation, in a relevant, accurate, complete, consistent, comparable and transparent manner (COM 2022). Single-pass across-track InSAR is probably the best compromise between price and accuracy to do the task. Airborne lidar is far more expensive for global use, whereas simple SAR backscatter and optical reflectance models have much lower accuracy. The bistatic SAR data would also enable a long list of other forestry, agricultural, ocean, and security applications.

The mission could provide important input into ocean models to improve their forecast accuracy by contributing to understanding equatorial and high-latitude ocean surface current fields, bulk wave parameters and wave density spectra estimates. It could complement the proposed European Space Agency’s (ESA) Sea surface KInematics Multiscale monitoring mission (SKIM) by improving forecast model parametrizations. Monitoring of ice dynamics in coastal zones and in the marginal ice zone would improve the safety of winter navigation and climate adaptation. Monitoring of surface currents in large rivers would enable a more accurate estimation of their flow rates and discharge to the coastal sea.

Experience gained with the C-band Sentinel-1 passive companions could be translated into a companion constellation mission to the planned ESA’s L-band mission ROSE-L in 2028 (Davidson & Furnell 2021). The 3D data factories with global coverage, simultaneous wide area footprints with high resolution and 12-day repeat cycle orbits and the possibility to compare nearly simultaneously obtained C- and L-band imagery would constitute a powerful tool to understand our planet, processes and phenomena occurring there in time and space. Such 3D data could help us better deal with the challenges such as food security and forest loss if to mention only two of the fields where we see the possible greatest contribution of the companion satellites mission described in this paper.

References

Bispo, P. D. C., Pardini, M., Papathanassiou, K. P., Kugler, F., Balzter, H., Rains, D., ... & Araujo, L. S. (2019). Mapping forest successional stages in the Brazilian Amazon using forest heights derived from TanDEM-X SAR interferometry. Remote Sensing of Environment, 232, 111194.

COM (2022). Proposal for a REGULATION OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL establishing a Union certification framework for carbon removals. COM(2022) 672 final, 2022/0394(COD).

Davidson, M. W., & Furnell, R. (2021, July). ROSE-L: Copernicus l-band SAR mission. In 2021 IEEE International Geoscience and Remote Sensing Symposium IGARSS (pp. 872-873). IEEE.

Ignatenko, V., Laurila, P., Radius, A., Lamentowski, L., Antropov, O., & Muff, D. (2020, September). ICEYE Microsatellite SAR Constellation Status Update: Evaluation of first commercial imaging modes. In IGARSS 2020-2020 IEEE International Geoscience and Remote Sensing Symposium (pp. 3581-3584). IEEE.

Krieger, G., Moreira, A., Fiedler, H., Hajnsek, I., Werner, M., Younis, M., & Zink, M. (2007). TanDEM-X: A satellite formation for high-resolution SAR interferometry. IEEE Transactions on Geoscience and Remote Sensing, 45(11), 3317-3341.

López-Dekker, P., Rott, H., Prats-Iraola, P., Chapron, B., Scipal, K., & De Witte, E. (2019, July). Harmony: An Earth explorer 10 mission candidate to observe land, ice, and ocean surface dynamics. In IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium (pp. 8381-8384). IEEE.

Moreira, A., Krieger, G., Hajnsek, I., Papathanassiou, K., Younis, M., Lopez-Dekker, P., ... & Parizzi, A. (2015). Tandem-L: A highly innovative bistatic SAR mission for global observation of dynamic processes on the Earth's surface. IEEE Geoscience and remote sensing magazine, 3(2), 8-23.

Olesk, A., Voormansik, K., Vain, A., Noorma, M., & Praks, J. (2015). Seasonal differences in forest height estimation from interferometric TanDEM-X coherence data. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 8(12), 5565-5572.

Potin, P., Rosich, B., Miranda, N., Grimont, P., Shurmer, I., O’Connell, A., ... & Gratadour, J. B. (2019, July). Copernicus Sentinel-1 constellation mission operations status. In IGARSS 2019-2019 IEEE international geoscience and remote sensing symposium (pp. 5385-5388). IEEE.

Schlund, M., von Poncet, F., Hoekman, D. H., Kuntz, S., & Schmullius, C. (2014). Importance of bistatic SAR features from TanDEM-X for forest mapping and monitoring. Remote Sensing of Environment, 151, 16-26.

Zonno, M., Matar, J., de Almeida, F. Q., Younis, M., Reimann, J., Rodriguez-Cassola, M., ... & Tossaint, M. (2021, March). Sentinel-1 Next Generation: main mission and instrument performance of the Phase 0. In EUSAR 2021; 13th European Conference on Synthetic Aperture Radar (pp. 1-5). VDE.

The growth of coastal megacities (those with populations of more than 8 million people) is concentrating populations in hazardous places, particularly in developing countries such as Pakistan. Similarly, more cities are expected to grow/develop along the coast of Pakistan such as the Baluchistan coast (Pasni, Omwara, Sumiani and Gwadar). These coastal areas are expected to be most vulnerable to seawater intrusion. The vulnerability of any coastal area increases with increasing land subsidence, deteriorating water drainage system, increase in sea level and local seismic activity (Elshinnawy & Almaliki, 2021).

Interferometric Synthetic Aperture Radar (InSAR) has become one of the most important and useful methods for the estimation of ground (Kumar et al., 2020; Ramzan et al., 2022). The enriched availability of new SAR tools and satellite collections has encouraged a solid development of processing procedures such as finding the small ground deformation signals linked to the different phases of the seismic cycle (Ali et al., 2021). InSAR is a radar technique that uses two or more SAR images to produce surface deformation maps. This technique can measure sub-cm changes in deformation over spans of days to years (Ali et al., 2018; Lu et al., 2020) over large areas with a high spatial resolution by using radar signals from Earth-orbiting satellites (Khan et al., 2020).

Figure 1 shows the study area, the Arabian Plate subducts beneath the Eurasian Plate and is associated with an accretionary wedge of sediments developed since the Cenozoic. The Makran Trench is connected by the Minab Fault system to the Zagros folds and thrust belt. The Makran Trench is bounded by the transgressional strike-slip Ornach-Nal and Chaman Faults, which connect to the Himalayan orogeny (Ali et al., 2021).

The objective of this study is the investigation of the potential significance of ground deformation for structural damage evaluation, by measuring the magnitude and extent of surface deformation in the Makran subduction zone (Pasni, Omwara, Sumiani and Gwadar) and the impact of Sea Water Intrusion on land subsidence along the coastal areas. The coastal area of Pakistan lies in a high-risk zone. Disasters related to drought, earthquake and tsunami can strike anywhere. Indus Delta is facing many problems due to the increasing seawater intrusion under prevailing climatic change, where land deformation can augment its vulnerability. Therefore, this study will be helpful for assessing the extreme changes in coastal dynamics.

In this study, open-access Sentinel-1 Interferometric Wide Swath (IW) C-band data is used, because of its considerable area coverage and high spatial resolution. SAR data were used in pairs of master and slave images to develop interferograms for the estimation of surface deformation. The unprecedented increase in prevailing surface deformation and its relationship with seawater intrusion can cause significant damage to the infrastructure and ecology of the region which needs immediate attention of the policymakers and scientific community, which will also help the community to mitigate the challenges of rising sea levels if any in future.

PDO considered to be a global leader in the field of Enhanced Oil Recovery (EOR) and has invested a great deal of time and money in ground-breaking EOR projects. EOR is a key factor contributor for the company’s hydrocarbon production sustainability. Currently, there are around 16 projects and field trails under execution by the company to devise and find the optimum EOR techniques for various production fields.

Yibal is one those fields where EOR techniques have been applied and is considered amongst PDO’s largest producing fields with vertically stacked carbonate reservoirs having gas from shallow Natih Formation and oil from lower Shuaiba formation with water flood recovery. Natih formation is a highly compacting formation as the reservoir pressure declines with production. Reservoir compaction of Natih A has induced noticeable damage surface facilities and several Shuaiba wells penetrating through the compacting layer. With significant facilities at Yibal stations (A, GGP), accurate predictions of surface subsidence and differential settlement (tilt) up to the end-of-field-life (EoFL) are critical to assess the design tolerance and adopt mitigations such as strengthening or modifications to ensure integrity and avoid any production deferment or HSE event.

Extensive surveillance methods such as synthetic aperture radar (InSAR) and Global Navigation Satellite Systems (GNSS) are in place to monitor surface deformation. Subsurface surveillance includes Compaction Monitoring Instrument (CMI) to measure subsurface compaction and micro-seismic to monitor fault reactivation and cap rock integrity.

Geomechanical model subsidence predictions calibrated with surveillance data provides reliable estimates of current subsidence (with maximum about 2.0 m) with EoFL maximum predicted to be around 2.5m.

Geomechanical modeling results integrated with surveillance data, provide key inputs for risk assessment and engineering design parameters. In terms of spatial resolution, InSAR data provides the best quality to plot and visualize spatial subsidence and derive associated tilt maps. However, InSAR data is not available since the beginning and provides estimates only in the time period the data is available. An approach by combining GNSS data, Geomechanics model and InSAR derived spatial subsidence ratio trends was developed to generate a synthetic total subsidence map at EoFL. Detailed maps of yet-to-expect subsidence can now be generated for assessing future risks and calibrated with new data as it comes in to improve accuracy.

The generated maps provide key inputs to engineering teams in assessing structural health of facilities and input in design or restoration of ageing facilities. The EoFL subsidence map can be combined with the surface topography map to support hydrology studies in assessing risks due to changes in water accumulations from surface runoff. In- addition, it provides reliable frequencies of building inspection and other surface infrastructure, minimize integrity issues and maintaining cap rock integrity.

And for a better analyzation and interpretation of the derived cumulative surface displacement map, a classified risk map was generated to highlight different severity risk into three zones (low --- > tilt less than 400 mm/km, medium --- > tilt between 400 – 800 mm/km and high ---- > tilt higher than 800 mm/km).

References

[1] Blanco, P., F. Pérez, A. Concha, J. Marturià, and V. Palà, 2012, Operational PS-DInSAR deformation monitoring project at a regional scale in Catalonia (Spain): IEEE International Geoscience and Remote Sensing Symposium, 1178–1181, https://doi.org/10.1109/ IGARSS.2012.6351338.

[2] Ferretti, A., C. Prati, and F. Rocca, 2001, Permanent scatterers in SAR interferometry: IEEE Transactions on Geoscience and Remote Sensing, 39, no. 1, 8–20, https://doi.org/ 10.1109/36.898661.

[3] Ferretti, A., G. Savio, R. Barzaghi, A. Borghi, S. Musazzi, F. Novoli, C. Prati, and F. Rocca 2007, Submillimeter accuracy of InSAR time series: Experimental validation: IEEE Transactions on Geoscience and Remote Sensing, 45, no. 5, 1142–1153, https://doi.org/10.1109/ TGRS.2007.894440.

[4] Ferretti, A., 2014, Satellite InSAR data: Reservoir monitoring from space: EAGE.

[5] Henschel, M. D., B. Deschamps, R. Rahmoune, and M. Sulaimani, 2014, Validation of operational surface movement at an enhanced oil recovery field: Presented at Geologic Remote Sensing Group Annual Meeting.

[6] Klemm, H., I. Quseimi, F. Novali, A. Ferretti, and A. Tamburini, 2010, Monitoring horizontal and vertical surface deformation over a hydrocarbon reservoir by PSInSAR: First Break, 28, no. 5, https://doi. org/10.3997/1365-2397.2010014.

[7] Rahmoune, Rachid & Sulaimani, Mohammed & Stammeijer, Jan & Azri, Saif & Gilst, Roeland & Mahruqi, Abir & Aghbari, Rawya & Belghache, Abdesslam. (2021). Multitemporal SAR interferometry for monitoring of ground deformations caused by hydrocarbon production in an arid environment: Case studies from the Sultanate of Oman. The Leading Edge. 40. 45-51. 10.1190/tle40010045.1.

Cultural property, as defined under Article 1 of the 1954 Hague Convention, is protected in the event of an armed conflict as well as in times of peace (UNESCO 2021). The exposure of cultural heritage to war damages in areas such as Iraq, Syria or currently, Ukraine makes it crucial to provide evidence of the condition of the sites, to be ready for recovery or to look into allegations of war crimes (EPRS 2022). Satellite imagery is particularly effective in monitoring and accurately assessing damage to cultural heritage in situations of armed conflict where the locations are not accessible and ground observation is inhibited (Casana & Laugier 2017).

This study focuses on utilising the integration of synthetic aperture radar (SAR) and optical Earth observation (EO) data for damage assessment in urban areas of Ukraine affected by the recent war. With space-borne SAR being able to acquire imagery independent of weather conditions, SAR is highly suitable to complement optical EO for monitoring and conservation of cultural heritage in crisis situations (Luo et al. 2019, Tapete & Cigna 2017). However, various approaches are based on commercial SAR satellite sensors, which provide very high-resolution on-demand imagery and fine-scale mapping (Tapete & Cigna 2015, Tapete & Cigna 2019). Using worldwide available, open-access and cost-effective data such as the Sentinel-1 SAR sensor from the Copernicus programme could overcome the disadvantages of lower spatial coverage. Several studies demonstrated SAR-based applications in conflict areas such as Raqqa (Syria), Mosul City (Iraq) or Kiev (Ukraine) and assessment of building damage by incorporating Sentinel-1 and interferometric coherence, permanent scatter techniques or intensity analysis (Boloorani et al. 2021, Braun 2018, Aimaiti et al. 2022).

Since the Russian invasion of Ukraine in February 2022, UNESCO has listed 241 cultural sites embedded within highly affected cities such as Kharkiv or Mariupol to be damaged or destroyed (UNESCO 2023). Damages are assessed based on field reports along with time- and cost-intense visual interpretation of commercial VHR imagery. The main objective of the present study is to determine the usability of freely available Sentinel-1 SAR and Sentinel-2 optical data for mapping damaged or destroyed cultural sites in the course of an ongoing war.

We used Sentinel-1 IW SLC products to generate coherency layers between pre-event data and pre-event to post-event data to approximate damage extent for the whole built-up area. Damage is assessed by detecting changes between the corresponding image pairs according to Serco Talia SPA (2020) workflow using SNAP 9.0.0 software. Post-images are selected from different dates as the war continues, to compare the situation before, during and after major reported battles. The results are complemented with structural damage identified by using multi-spectral optical imagery and pixel-wise differences in the spectral values of the near-infrared band (NIR) of pre- and post-event Sentinel-2 scenes. Integrating open-source GIS data, such as building footprints and point features, allows for spatially locating and identifying cultural and historical sites within the built-up areas. Results from Sentinel-1 and Sentinel-2 change detection are overlaid with the reference data to quantify the potential damage to cultural property. Limitations arise in differentiating damage levels or detecting changes related to smaller or single buildings as a result of the spatial resolution of Sentinel imagery. The lack of ground survey data only allows a qualitative accuracy assessment of the results using rapid damage maps published by the United Nations Institute for Training and Research (UNITAR) Operational Satellite Applications Programme (UNOSAT) and damaged cultural sites verified by UNESCO. However, the resulting damage maps can be used to highlight areas of major destruction and a rapid mapping of the potential impact on cultural heritage. A further investigation shall include texture features generated from the Grey Level Co-occurrence Matrices (GLCM) as recommended by Aimaiti et al. (2022) which may improve the workflow. If information reaches sufficient and acceptable accuracy, it can help to improve the efficiency of monitoring and damage assessment by focusing on more affected areas, e.g. during war crisis.

Aimaiti, Yusupujiang; Sanon, Christina; Koch, Magaly; Baise, Laurie G.; Moaveni, Babak (2022): War Related Building Damage Assessment in Kyiv, Ukraine, Using Sentinel-1 Radar and Sentinel-2 Optical Images. In: Remote Sensing 14 , 6239. DOI: 10.3390/rs14246239.

Boloorani, Ali Darvishi; Darvishi, Mehdi; Weng, Qihao; Liu, Xiangtong (2021): Post-War Urban Damage Mapping Using InSAR: The Case of Mosul City in Iraq. In: IJGI 10 (3), S. 140. DOI: 10.3390/ijgi10030140.

Braun, Andreas (2018): Assessment of Building Damage in Raqqa during the Syrian Civil War Using Time-Series of Radar Satellite Imagery. In: giforum 1, S. 228. DOI: 10.1553/giscience2018_01_s228.

Casana, Jesse; Laugier, Elise Jakoby (2017): Satellite imagery-based monitoring of archaeological site damage in the Syrian civil war. In: PloS one 12 (11), e0188589. DOI: 10.1371/journal.pone.0188589.

European Parliamentary Research Service EPRS (2022): Russia’s war on Ukraine cultural heritage. Available online: https://www.europarl.europa.eu/thinktank/en/document/EPRS_ATA(2022)729377, accessed on 24 February 2023.

Luo, Lei; Wang, Xinyuan; Guo, Huadong; Lasaponara, Rosa; Zong, Xin; Masini, Nicola et al. (2019): Airborne and spaceborne remote sensing for archaeological and cultural heritage applications: A review of the century (1907–2017). In: Remote Sensing of Environment 232. DOI: 10.1016/j.rse.2019.111280.

Serco Italia SPA (2020): Lebanon Damage Assessment with Sentinel-1 & Sentinel-2. (version 1.1). Retrieved from RUS Lectures at https://rus-copernicus.eu/portal/the-rus-library/train-with-rus/, accessed on 24 February 2023.

Tapete, Deodato; Cigna, Francesca (2017): Trends and perspectives of space-borne SAR remote sensing for archaeological landscape and cultural heritage applications. In: Journal of Archaeological Science: Reports 14, S. 716. DOI: 10.1016/j.jasrep.2016.07.017.

Tapete, Deodato; Cigna, Francesca (2019): COSMO-SkyMed SAR for Detection and Monitoring of Archaeological and Cultural Heritage Sites. In: Remote Sensing 11, 1326. DOI: 10.3390/rs11111326.

Tapete, Deodato; Cigna, Francesca; Donoghue, Daniel N.M.; Philip, Graham (2015): Mapping Changes and Damages in Areas of Conflict: from Archive C-band SAR Data to New HR X-band Imagery, towards the Sentinels. In: L. Ouwehand (Hg.): Proceedings of Fringe 2015: Advances in the Science and Applications of SAR Interferometry and Sentinel-1 InSAR Workshop. Frascati, Italy, 23-27 March: ESA Publication SP-731.

UNESCO (2021): The Hague Convention. 1954 Convention for the Protection of Cultural Property in the Event of Armed Conflict. Available online: https://en.unesco.org/protecting-heritage/convention-and-protocols/1954-convention, accessed on 24 February 2023.

UNESCO (2023): War in Ukraine. Damaged cultural sites in Ukraine verified by UNESCO. Available onnline: https://www.unesco.org/en/articles/damaged-cultural-sites-ukraine-verified-unesco?hub=66116, accessed on 23 February 2023.

Ground deformation is related to various geophysical and geological processes (GGPs) that act under Earth's surface (mainly in the Earth's crust), such as seismic events, volcanism, landslides, and subsidence, and it is characterized by surface displacements, highly variable in temporal and spatial scales. Surface displacement measurements contribute enormously to our understanding of the subsurface processes; knowledge of the surface displacement field and its spatial-temporal evolution is crucial for deciphering its causes, triggering factors, and mechanisms. During the last 30 years, InSAR technology has become a valuable tool in detecting and monitoring surface displacements associated with GGPs.

The study area, which comprises the Cerro Prieto pull-apart center and its surrounding, is located in the Mexicali Valley, northwestern Mexico. The study area lies within a highly active tectonic region, in the boundary between the Pacific and North American plates. The surface deformation in this area is caused by various natural processes, such as earthquakes, continuous tectonic deformation, sediment compaction, and human activity, primarily the fluid extraction in the Cerro Prieto Geothermal Field (CPGF) for energy production. Subsidence is a phenomenon common to the industrial development of geothermal energy fields, where in most cases, the extraction of fluids from geothermal systems occurs at a rate higher than the natural recharge and/or re-injection, inducing localized volumetric strain changes. Land subsidence (up to 18 cm/year) and related ground fissures are becoming a severe geological hazard in the study area damaging the local infrastructure and disturbing the social and economic development.

Surface deformation in Mexicali Valley has been studied using leveling and geological surveys, geotechnical instruments, and Differential SAR interferometry (DInSAR). Results obtained during the ESA C1P3508 project showed the importance of the DInSAR ground deformation monitoring in the Mexicali Valley (e.g., Glowacka et al., 2010; Sarychikhina et al., 2011, 2015, 2018). Moreover, they also highlighted the principal limitations of the DInSAR technique, mainly temporal decorrelation in highly vegetated areas surrounding the CPGF. However, since the launch of the Sentinel-1A (April 2014) and Sentinel-1B (April 2016) satellites, the provided data offer new opportunities to investigate surface deformation and create improved displacement time series in the area of study as a result of more frequent image acquisitions, every 6 or 12 days.

Here, the Sentinel-1 SAR images from 2015-2022 were used to infer surface deformation in the study area. The conventional DInSAR was applied to investigate the surface deformation caused by moderate sized earthquakes and creep events, whereas the advanced multitemporal DInSAR was applied to obtain the aseismic surface deformation rate and time series. Integration of results for 2015-2022 obtained here with results for early period (1993 – 2014), obtained in the previous studies, allows the surface deformation evolution analysis covering 30 years.

The random volume over ground (RVoG) model, based on the hypothesis of vertical homogeneous volume, utilizes an exponential function to depict the forest vertical structure. Specifically, in the RVoG model, the strongest backscatter is located at the top of the canopy, demonstrating high applicability to the relatively high-frequency polarimetric interferometric synthetic aperture radar (Pol-InSAR) systems. However, for P-band systems with remarkable penetration, the backscatter power is more likely to arise from the middle or lower layer of the canopy, implying the less effectiveness of the RVoG model in this situation. One solution is to establish a more complicated model to remedy the defect of the RVoG model. However, this technique brings high inversion complexity.

Due to the invalidity of the null ground-to-volume ratio assumption, one solution to P-band Pol-InSAR inversion based on the RVoG model is to increase observations, and yet, the inversion complexity is also compounded by its multi-baseline configuration. Fixing the extinction coefficient is often used to solve this problem. Nevertheless, the extinction varies drastically in the complex environment.

In terms of model improvement, Kugler et al. have extended the RVoG (called extended RVoG, i.e., E-RVoG in this letter) model with the negative extinction coefficient, which effectively takes the characteristics of P-band Pol-InSAR systems into account. Although the E-RVoG model retains the same parameters as the RVoG model, it has a stronger ability to describe the vertical structure.

On account of the fact that the vertical structure varies with forest species, age, shape, density, and so on, this paper puts forward a novel inversion scheme for single-baseline P-band Pol-InSAR, in which the extinction coefficient in the E-RVoG model is forecast by machine learning.

As correlations between each variable and the extinction coefficient are coupled jointly, it is of substantial difficulty to obtain the analytical expression of the inner relationship. Hence, the supervised machine learning is implemented to establish the potential correlations. The true extinction coefficient is acquired by the intersection of the solution space curve and the coherence line.

The feature extraction of the extinction coefficient depends on the incidence angle, terrain phase and the volume-only coherence. The machine learning adopts the random forest regression (The regressor is not unique.). Thus, the extinction coefficient can be forecast by the trained model.

The actual Pol-InSAR data verification illustrates that the inversion performance of the proposed scheme overmatches that of the traditional schemes.

This research was supported by the National Natural Science Foundation of China (No. 62231024).

The current study emphasizes the utility of the PS-InSAR technique for measuring tectonic and non-tectonic surface deformation towards the western part of the Indian plate. The matching of PS-InSAR time-series with GNSS time-series demonstrates the technique's mm level of accuracy. PS-InSAR is an advanced radar-based remote sensing method of InSAR technique applied for the periodic measurements of ground deformation. We have applied the technique for the measurements of tectonic deformation and non-tectonic (ground subsidence) deformation. For the tectonic deformation measurements, the crustal deformation estimation in the Kachchh and Saurashtra region of western India has been carried out, using Sentinel-1A images from 2014 to 2021. The results show an average LOS displacement of 4.3 ± 1.5 mm/yr towards the eastern part of Kachchh and show up to 5 ± 2.0 mm of annual LOS displacement within the Saurashtra. The time-series analysis using PS points matches with the GNSS-derived deformation rates. Further, for the non-tectonic deformation measurements, we applied the PS-InSAR technique in the city of Ahmedabad, western India using the Sentinel-1A dataset (2017 to 2020). The results based on the PS-InSAR data analysis reveal displacement (LoS) of up to 25 ± 2.5 mm/yr in several parts of the city, which corresponds to the GNSS vertical displacement. Furthermore, groundwater level data from 1960 to 2020 was simulated to estimate ground subsidence and results closely matched those of PSI and GNSS. As a result, we conclude that groundwater decline, as identified by PS-InSAR, GNSS, and water level datasets, is the primary cause of surface subsidence in the city.

Persistent Scatterer Interferometry (PSI) is a powerful tool to estimate ground deformations with millimeter-level precision. Due to the integrated processing of a large data stack, numerous errors and artifacts are eliminated and coherent Point Scatterers (PS) are detected for objects characterized by stable and high coherence in the analyzed period. In practice, most of these points, due to the nature of the reflection of a radar wave, will be located on buildings or infrastructure objects. Unfortunately, despite the millimeter precision of the estimates, the PS typically suffer from low geolocalisation accuracy, which makes it difficult to relate them to a real object in space and in consequence makes it hard to interpret the deformation pattern. Moreover, interpretation is also hampered by the 1-dimensional character of the InSAR results in the satellite line of sight (LOS). When multiple data stacks are available with different orbit geometries (ascending, descending) from regions of uniform motion (RUM), a decomposition into multiple displacement vectors can be made (Brouwer and Hanssen, 2022). With sufficiently dense data, such a decomposition could be made on object level. Hereby, linking the original PS to the correct object is crucial. To improve the accuracy of PS geolocation, the PSI – LiDAR point cloud linking algorithm (Dheenathayalan et al., 2016, van Natijne et al., 2018, Hu et al., 2019) can be used. The algorithm aims to find the nearest LiDAR point within the metric defined by the variance-covariance matrix of the PS position, conveniently visualized using a rotated 3D error ellipsoid.

However, in practice, the application of the algorithm reveals that the interpretation of the PS data does not necessarily become easier. Although the results after linking the PS look visually attractive, since they are obviously aligned with geo-objects, there is no more opportunity for human verification of the outcome. Whereas the original PS data show a certain spread in the PS locations, which can be interpreted by InSAR experts and expresses the uncertainty in the PS position, this information is lost after the linking step. Hence, the applied one-way linking process results in a loss of useful information. The actual correctness of the linking step can no longer be verified. To overcome this problem, in our contribution we present a methodology to enable the interpretation of both the original and the linked PS positions. The approach is based on a 3D visualization of the PS and LiDAR data, together with PS position error ellipsoids and linking vectors. This approach both enables verification of the linking process and improves the interpretation of the PS results.

The methodology is applied to study areas in the Upper Silesia Coal Basin (USCB), Poland, and Amsterdam, The Netherlands. In both cases, nationwide airborne LiDAR datasets and the results of PSI processing of C-band (Sentinel-1) and X-band (TerraSAR-X) data were used. The extraction and visualization made it possible both to notice differences in the quality of the geolocalisation data from the various sensors as well as to relate the observed deformations, especially in USCB, to the objects affected by them.

The PSI – LiDAR linking algorithm and 3D visualization tools for improved PS interpretation, both implemented in Python, are available as open-source repositories.

1. Brouwer, W.S., & Hanssen, R.F. (2022). A Treatise on InSAR Geometry and 3D Displacement Estimation, https://doi.org/10.31223/X55D37.
2. Dheenathayalan, P., Small, D., Schubert, A., & Hanssen, R. F. (2016). High-precision positioning of radar scatterers. Journal of Geodesy, 90(5), 403-422, https://doi.org/10.1007/s00190-015-0883-4.
3. Hu, F., Leijen, F. J. V., Chang, L., Wu, J., & Hanssen, R. F. (2019). Monitoring deformation along railway systems combining multi-temporal InSAR and LiDAR data. Remote sensing, 11(19), 2298, https://doi.org/10.3390/rs11192298.
4. Van Natijne, A. L., Lindenbergh, R. C., & Hanssen, R. F. (2018). Massive linking of PS-InSAR deformations to a national airborne laser point cloud. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci, 42(2), 1137-1144, https://doi.org/10.5194/isprs-archives-XLII-2-1137-2018.

The continuous analysis of ground deformation is essential for both the assessment of natural hazards and the monitoring of human-induced activities. In this study, we present the results of a PSI analysis of ground deformations in the region of Cluj-Napoca, Romania. Cluj-Napoca is the second most populous city in Romania, located in a hilly environment, built on the banks of the river Someșul Mic is ideal for such an assessment. The urbanization of the city has rapidly progressed in the recent decades, more than doubled the area of the city in 30 years, as the boundaries of the city reached the neighboring hills with slopes up to 26% steepness, which are prone to landslides.

The PSI was performed using more than 8 years of Sentinel-1 descending data via the Interferometric Point Target Analysis module of the Gamma software. For the interpretation, we used GIS to integrate the local geological information and include a geotechnical viewpoint as well. The thorough analysis is indeed necessary as many types of deformations are present, often superimposed, related to mass movements, groundwater pumping, sediment compaction, industrial operations, mining, earthworks related to road construction, etc. Results expected to show significant movements on recently built areas at the edges of the city, often caused by the combined effect of anthropogenic activities and geological conditions. This study is also a proof of the necessity of local studies, although country and continent-wide maps are useful tools for mapping of large areas: results are more up-to-date, processing details are more specifically tailored to the region and the user needs, e.g. by using locally selected reference and adjusting parameters to the goals of the research. Furthermore, our detailed analysis involving local knowledge, local experts and auxiliary data provides information regarding the risks, the interpretation, origin and characterization of the detected movements. By doing so, we demonstrate the necessity of collaboration between remote sensing and local geotechnical experts to maximize the potential and operative effectiveness of InSAR data.

The accurately mapped and quantified ground deformations can be used for the better understanding of the geological processes and assessing the risk of the urban development in the area. The detected slope instabilities, subsidence or uplift can have significant impacts on the built environment, and it is also important to take them into account in the planning and design of new buildings and infrastructure.

Trees are a critical component of the ecological balance in forests, parks, and urban areas, and monitoring their health is essential to maintaining their ecological and aesthetic value. However, trees are often subjected to various diseases and environmental stressors, which can lead to their decline and eventual death. Thus, timely and accurate detection of tree health problems is crucial for effective tree management and conservation. Within this context, traditional methods for tree health monitoring, such as visual inspections or destructive sampling, are time-consuming and fail to detect diseases in their early stage [1].

Satellite imaging technology has increasingly been utilised for forestry applications in recent years, as it can provide valuable information on the overall health of trees, including the leaf area, the photosynthetic activity, and the water stress [2]. This method can detect changes in tree health over time and across large areas, and can therefore inform forestry management decisions. This includes informing on which trees to prioritise for treatment or removal, thus helping to prevent the spread of diseases to other trees.

In terms of ground-based non-destructive testing (NDT) methods, recent studies have demonstrated the potential of Ground Penetrating Radar (GPR) for tree health monitoring. With regards to the investigation of tree root systems, GPR can provide valuable insights on tree roots’ distribution and mass density, as well as their interaction with the soil and the built environment [3]. As such, the use of GPR for tree health monitoring is gaining interest and attention from researchers and professionals in the field.

The aim of this study is therefore to assess the viability of integrating satellite imaging and GPR for tree health monitoring and diagnosing tree diseases. A diseased tree located in an urban park within the London Borough of Ealing, London, was selected for investigation purposes. Signs of decay in the tree have been analysed from the historical satellite radar data. Subsequent GPR investigations of the root area with a 600 MHz central frequency antenna system showed anomalies compatible with the presence of root damage. Excavations were carried out for validation purposes, and the evidence has confirmed an ongoing root disease.

Results of this preliminary study have proven the viability of integrating of satellite remote sensing and GPR. The combination of these techniques has the potential to improve the efficiency of monitoring, reduce the need for destructive sampling, and support sustainable forestry and urban green space management. Further research is needed to explore the application of these techniques to other tree species and environmental conditions.

Keywords

Multi-scale tree health monitoring; InSAR for tree management and conservation; Ground Penetrating Radar (GPR)

Acknowledgements

The Authors would like to express their sincere thanks and gratitude to the following trusts, charities, organisations and individuals for their generosity in supporting this project: Lord Faringdon Charitable Trust, The Schroder Foundation, Cazenove Charitable Trust, Ernest Cook Trust, Sir Henry Keswick, Ian Bond, P. F. Charitable Trust, Prospect Investment Management Limited, The Adrian Swire Charitable Trust, The John Swire 1989 Charitable Trust, The Sackler Trust, The Tanlaw Foundation, and The Wyfold Charitable Trust. The Authors would also like to thank the Ealing Council and the Walpole Park for facilitating this research.

References

[1] Alani, A.M., Lantini, L. Recent Advances in Tree Root Mapping and Assessment Using Non-destructive Testing Methods: A Focus on Ground Penetrating Radar. Surveys in Geophysics 41, 605–646 (2020).

[2] Lechner, A.M., Foody, G.M., Boyd, D.S. Applications in Remote Sensing to Forest Ecology and Management, One Earth 2(5), 405-412 (2020).

[3] Lantini, L.; Tosti, F.; Giannakis, I.; Zou, L.; Benedetto, A.; Alani, A.M. An Enhanced Data Processing Framework for Mapping Tree Root Systems Using Ground Penetrating Radar. Remote Sensing 12, 3417 (2020).

The region surrounding the city of Patras in the northwest of the Peloponnese peninsula in southern Greece is considered one of the most seismically active areas in the Mediterranean. The area is under the influence of the Hellenic subduction zone east of the area, a rift system bordering the region to the north, which consists of the Gulf of Corinth and Gulf of Patras, and numerous active faults within the area of interest (e.g. the Rion-Patras fault and the Aigia Triada fault), which increase the risk for ground deformation and earthquakes. The Greek mainland and the Peloponnese Peninsula diverge from each other by about 1.5 cm per year, while the African continental plate is subducted under the Aegean microplate at a rate of 0.5 - 3.5 cm per year about 100 km off the southwestern coast of Greece. The urban area of the city of Patras is additionally affected by subsidence, while the rural mountainous areas south and east of the city are affected by 137 known active landslides. Large infrastructure constructions such as the Parapeiros-Peiros dam south of Patras or the Rio–Antirrio Bridge connecting the region to the Greek mainland are affected by these surface deformations and therefore need to be monitored regularly.

In this study we analyzed a time series of Sentinel-1 SAR images using the Persistent Scatterer Interferometry algorithm Stanford Method for Persistent Scatterer, in order to document the described ground deformation. A spatial analysis of the deformation patterns was performed based on the resulting mean velocity maps. In addition, the dynamic of the different deformation patterns was considered. The Matlab-based software Persistent Scatterer Deformation Pattern Analysis Tool (PSDefoPAT) automatically assigns a suitable time series model to the displacement time series of each persistent scatterer. Time series models with and without seasonal components are considered, as well as a linear, quadratic, or piecewise linear long-term trend. By displaying different combinations of the estimated model parameters as an RGB triplet, PSDefoPAT enables the visual representation of the temporal deformation patterns in a spatial context and thus supports the analysis of Persistent Scatterer Interferometry results concerning the stability of infrastructure, such as dams, and the risk of geohazards, such as landslides.

Analysis of External DEM on Open-pit Mining Area Deformation Monitoring by Means of LuTan-1 SAR

LuTan-1 SAR satellite is the first bistatic spaceborne SAR constellation for multiple applications in China, which consists of two identical multi-polarimetric L-band SAR satellites. The twin satellites have been successfully launched from Jiuquan satellite launch center on 26 January and 27 February 2022, respectively. Due to the precise orbit control and two satellites operating in a common reference orbit with a 180-degree orbital phasing difference, the revisit cycle of LuTan-1 will be reduced from 8 days to 4 days with 350m orbital tube, which ensure the high temporal and spatial coherence for interferometric applications of LuTan-1 data. Thus, surface deformation monitoring with centimeter even millimeter accuracy may be achieved based on InSAR technique. The performance of LuTan-1 will be fully tested and verified for multiple applications during in orbit test. Then LuTan-1 will continually provide high-quality SAR data, which will support the world wide environmental monitoring, especially for disaster monitoring.

Geological disasters such as local ground subsidence, cracks and collapse in coalfield are induced by intensive and large-scale coal mining. InSAR has a capability of surface deformation monitoring with high accuracy, which can effectively support the mine ecological security monitoring and protection. A series of issues such as ground subsidence, landslides and damage of structures are existed over coal mining areas. Therefore, it is significant to monitor the surface deformation over coalmines. On 22 February 2023, a large area collapse of Xinjing strip mine in Inner Mongolia was happened, inducing heavy casualties and property losses. It is necessary to carry out high precision deformation monitoring in opencast mining area. In our research, the ability of LuTan-1 for open-pit mining area deformation monitoring was evaluated. Especially the influence of different external DEM for deformation monitoring was further discussed and analyzed. The results demonstrated that high accuracy and timeliness external DEM is necessary for open-pit mining deformation monitoring using InSAR techniques.

LuTan-1 SAR data are acquired on 25 December 2022 and 10 January 2023 over the open-pit mining area, shown as Figure 1. The configuration parameters of LuTan-1 SAR data was listed in Table 1.

Figure1. LuTan-1 SAR data over the open-pit mining area.

Table 1. Configuration parameters of LuTan-1 SAR data.

The topography of opencast coal mine area usually changes obviously with the mining of coal resources. Therefore, the high accuracy and timeliness external DEM has significant influence on deformation monitoring. In order to effectively reduce the deformation monitoring error caused by the external DSM, the DSM extracted by GaoFen-7 satellite was utilized in our research. And the GaoFen-7 data was acquired on 20 November 2022, which is closed to LuTan-1 SAR data acquisition time. The difference between GaoFen-7 derived DSM and SRTM was analyzed and discussed, shown as Figure 2.

Figure 2. DSM analysis over the open-pit mining area. (a) DSM derived by GaoFen-7, (b) SRTM DEM, (c) DSM difference between GaoFen-7 and SRTM, (d) Statistical histogram of DSM difference.

In the process of DInSAR strategy, reliable external DEM is crucial to obtain accurate deformation results. For open-pit coal mine, mining activities and dump have significant influence on the topography of the mining area. The comparison of SRTM DEM and GaoFen-7 DSM was shown as Figure 2, which revealed significant difference. The elevation difference is mainly distributed among -50 m to 50 m, and the maximum difference can reach to 328.04 m. The mining time of the open-pit coal mine is obviously later than SRTM production time, thus the SRTM cannot accurately characterize the topography of study area. On the contrary, GaoFen-7 data was obtained on November 2022, which is nearly to the acquisition time of LuTan-1 SAR data. That’s why the obvious difference between SRTM DEM and GaoFen-7 DSM was displayed.

Figure 3 shows the differential interferograms generated by SRTM DEM and GaoFen-7 DSM respectively. The differential interferogram based on SRTM DEM show relatively dense fringe, and the characteristic of the fringe is basically consistent with the intensity image of mining area. Therefore, it can be judged that the interference fringe is mainly caused by terrain error, and the deformation fringe is coupled with the terrain error fringe. From November 2022 to December 2022, the terrain of the mining area has little change, thus the differential interferogram based on GaoFen-7 DSM contain obvious deformation fringe. Due to the application of GaoFen-7 DSM, the terrain error for deformation monitoring can be greatly reduced.

Figure 3. (a) Differential interferogram generated by SRTM DEM; (b) Differential interferogram generated by GaoFen-7 DSM.

Furthermore, the deformation monitoring with different external DEM were compared and discussed over the study area.

1. Deformation monitoring of opencast mining area using SRTM

LuTan-1 SAR data covering the mining area were used for differential InSAR processing with SRTM as inputs. The vertical baseline of the interferometric image is 395.81 meters, and the corresponding height of ambiguity (HOA) is 54.23 meters. In other words, for differential InSAR processing when the DEM error is lager than 54.23 meters, it will cause more than one interference fringe error on the differential interferogram. And thus, a significant error of deformation monitoring may be derived due to the application of SRTM DEM.

Figure 4. Deformation monitoring results using SRTM (superimposed on optical image).

The deformation results using SRTM DEM are highly correlated with the topography difference of the open-pit mining area, so the deformation information is mainly caused by the error of external DEM, which further demonstrated the importance of high-precision and time-efficient external DEM for InSAR deformation monitoring.

2. Deformation monitoring of opencast mining area using GaoFen-7 DSM

Due to the extensive mining activities, the topography of opencast coal mine area generally changes obviously. In order to reduce the influence of external DSM error, GaoFen-7 derived DSM was applied in our research, and the GaoFen-7 data acquisition time is closed to LuTan-1.

Figure 5 shows the deformation results using LuTan-1 SAR data and GaoFen-7 DSM from 25 December 2022 to 10 January 2023.

Figure 5. Deformation monitoring results from 12 December 2022 to 10 January 2023 using GaoFen-7 DSM. (a) Collapse area on 22 February 2023; (b),(c)and(d) are three deformation areas.

The results preliminarily indicated that there are multiple obvious deformation areas in the open-pit mining area. Within the 3km×3km range of the mining area, four obvious subsidence areas were detected from 25 December 2022 to 10 January 2023. The maximum subsidence of the four areas (a), (b), (c) and (d) are 0.1m, 0.15m, 0.25m and 0.23m respectively. With high frequency SAR observations and timely processing, dynamic deformation over study area can be monitored. In combination with prior knowledge, geological basis and expert interpretation, the hazard monitoring and identification may be achieved owing to the multiple SAR observations.

Digital Twins allow to investigate and visualize multi-source data in a unique environment [1]. Amongst others, satellite imageries have been increasingly implemented due to the continuous growth of satellite missions. In this context, the use of the Multi-Temporal Interferometric Synthetic Aperture Radar (MT-InSAR) was significantly consolidated, for the continuous assessment of bridges and the health monitoring of transport infrastructures [2]. This research aims to investigate the viability of an experimental implementation of a Digital Twin of transport assets, based on multi-source and multi-scale information. To this purpose, satellite remote sensing and ground-based techniques provide accurate and updatable information useful for monitoring activities [3]. These crucial pieces of information were analyzed for the structural assessment of infrastructure assets, selected as case-studies in Rome (Italy), and the prevention of damages related to structural subsidence. To this purpose, C-Band SAR products of the mission Sentinel 1 of the Copernicus programme of the European Space Agency, and high-resolution X-Band SAR imageries were acquired and processed by MT-InSAR technique. The analyses were developed to identify and monitor the structural displacements associated to transport infrastructures. An algorithm was developed to create and import automatically an informative digital object integrated into the Digital Twin, starting from the Persistent Scatterers (PSs), including the historical time-series of deformation. On the other hand, several Non-destructive Testing methods were implemented including Ground Penetrating Radar (GPR) and Laser Scanner technologies. More specifically, several GPR frequencies were implemented for this purpose, with the aim to investigate the condition of the layers of the superstructures at different propagation lengths. Several PS data-points with coherent deformation trends were analyzed, and an integrated interpretation was proposed using the GPR tomography. A novel data interpretation approach is proposed, paving the way for the development of a Digital Twin of the inspected transport asset. The outcomes of this study demonstrate how multi-temporal InSAR remote sensing techniques can be applied to complement non-destructive ground-based analyses, for routine infrastructure inspections.

Keywords – Digital Twin, Persistent Scatterers Interferometry (PSI), Ground Penetrating Radar (GPR), Integrated Health Monitoring, Railway monitoring, Transport Infrastructure Maintenance

Acknowledgments

The authors want to acknowledge the Italian Space Agency (ASI) for providing the COSMO-SkyMed Products® (©ASI, 2016-2018). The Sentinel 1A products are provided by ESA (European Space Agency) under the license to use. This research is supported by the Italian Ministry of Education, University and Research (MIUR) under the National Project “EXTRA TN”, PRIN 2017 and the Project “M.LAZIO”, accepted and funded by the Lazio Region, Italy.

References

[1] Hidayat F., Supangkat S. H. and Hanafi K., "Digital Twin of Road and Bridge Construction Monitoring and Maintenance," 2022 IEEE International Smart Cities Conference (ISC2), Pafos, Cyprus, 2022, pp. 1-7, doi: 10.1109/ISC255366.2022.9922473.

[2] Gagliardi, V. Tosti, F. Bianchini Ciampoli, L. Battagliere, M.L. D’Amato, L. Alani, A.M. Benedetto, A. Satellite Remote Sensing and Non-Destructive Testing Methods for Transport Infrastructure Monitoring: Advances, Challenges and Perspectives. Remote Sens. 2023, 15, 418. https://doi.org/10.3390/rs15020418

[3] D'Amico F., Bertolini L., Napolitano A., Manalo D. R. J., Gagliardi V., and Bianchini Ciampoli L. "Implementation of an interoperable BIM platform integrating ground-based and remote sensing information for network-level infrastructures monitoring", Proc. SPIE 12268, Earth Resources and Environmental Remote Sensing/GIS Applications XIII, 122680I; https://doi.org/10.1117/12.2638108

Landslides pose a destructive geohazard to people and infrastructure that results in hundreds of deaths and billions of dollars in damages every year. China is one of the countries worst affected by landslides in the world, and great efforts have been made to detect potential landslides over wide regions. However, a recent government work report shows that 80% of the newly formed landslides occurred outside the areas labelled as potential landslides, and 80% of them occurred in remote rural areas with limited capability of disaster prevention and mitigation. In this presentation, a multi‐source remote sensing technical framework is demonstrated to detect potential landslides over wide regions.

The Subdirectorate General for Monitoring, Warning and Geophysical Surveys, belonging to the National Geographic Institute of Spain has among its responsibilities: Planning and management of systems for observation, monitoring and communication to institutions of volcanic activity and determination of associated hazards, as well as management of geomagnetism observation systems and related work and studies.

In this framework of responsibilities, observation systems are multidisciplinary, including deformation, seismology, gravimetry, geochemistry and geomagnetism techniques. In order to monitor ground deformations, Spaceborne SAR interferometry (InSAR) has been combined with other deformation measurement techniques, such as GNSS inclinometers or robotic total stations.

In this context, a fully automatic processing methodology which has been running for the last 5 years, has been developed to obtain interferograms with each new image acquired by the Sentinel 1 Satellites over the Canary Islands. Recently, images from other sensors such as PAZ, has been added to this processing. Due to the special atmospheric and topographical characteristics of the Canary Islands, it is possible to observe an important contribution of atmospheric artifacts in the displacement and interferometric phase maps that are obtained as final products. These atmospheric effects are also especially common on volcanic islands such as the Canaries where there are large changes in the distribution of water vapor with height and where the winds that bring moisture from the sea have dominant directions.

In this work we present the results of the application of different methodologies such as the GACOS products and the relation between topography and phase to mitigate the effect that variations in the state of the atmosphere has on the interferograms. For this purpose, the same methodologies have been applied on islands with different atmospheric and topographic characteristics, different expected patterns of deformation trying to find the most applicable methodology for each case. A comparison of the application of these methodologies to the products obtained with images from different sensors has also been made. With all this information, it is intended to incorporate the atmospheric correction to our automatic processing, establishing thresholds for the different parameters studied, which allow us to discern which type of correction is most appropriate in each case.

Ports play a crucial role in the global economy as they serve as vital gateways for international trade, facilitating the movement of goods and connecting businesses to markets around the world. The efficient functioning of ports is essential for global trade and economic growth, as it enables businesses to access new markets, source inputs, and reach customers worldwide.

However, port infrastructures are vulnerable to multiple natural agents that can lead to their deterioration, hindering their efficient operation and functionality. To address this complex environment, DInSAR technologies have proven to be highly effective, enabling the monitoring of surface deformations in near real-time across the entire port area.

DInSAR technology could have a positive impact on the port environment in the following topics:

i) the continuous and non-intrusive description of damage evolution in breakwaters slopes, protective walls, cumulative deformation on jetties, etc…, ii) millimetre-accurate detection of cumulative deformations caused, for instance, by soil consolidation, in esplanades, pavements, parapets or crown walls., iii) the control of the collection of permanent waste, or iv) support for the certification of works based on measurements.

Detecting and quantifying the deformation caused in each individual component of the port infrastructure structure can be of great use for the precise evaluation and prediction of different failure modes.

Therefore, the precise positioning of persistent scatterers is crucial in the analysis of MTInSAR data for effective monitoring to identify potential disruptions in port activity and failure modes for different structural typologies present on the harbour infrastructure.

In this work we evaluate the accuracy of DInSAR-generated height data from different Persistent Scatterers (PS), Small BAseline Subset (SBAS) and Persistent Scatterers Distributed Scatterers(PSDS) software. We attempt to estimate the real phase centre of the scatterer over multiple port infrastructures by registering the DInSAR point cloud with high-resolution LiDAR data from the Spanish National Orthophoto Program. Furthermore, we also evaluate the effect of different subpixel corrections on DInSAR scatterers to improve the accuracy of deformation measurements in port environments.

The use of DInSAR with precise positioning of PS in port infrastructures with the aim of evaluating and having the capability of predicting their different failure modes.

The Himalayan region of Uttarakhand in India is known for landslides triggered by earthquakes and rainfall. Recently, a higher concentration of extreme climatic scenarios in the form of concentrated rain has been observed in many places causing loss of lives and damage to private and public properties (Dobhal et al., 2013). Besides disastrous landslide events, phenomena in the form of the development of cracks, subsidence, small-scale debris wash, erosional features, etc., occur at many places and serve as primary indicators of slope instability that may intensify into landslides in the near future. Therefore, it is essential to map the areas of active landslide-related creep as well as slope instability for the disaster management strategy of a region.

The city of Nainital in India, lies between longitude 79°25′35 “E to 79°28′32 “E and latitude 29°24′28 “N to 29°20 “05”. The township is a famous hill station with a highly variable floating population during the peak tourist season in summer and winter in India. The city is known to have had occurrences of landslides in the past, and about half of the area of the Nainital is covered with debris generated by landslides (Valdiya 1988). The earliest record of landslides in the area dates back to 1867 and 1880. The area again witnessed landslides as recently as 2009 due to increased and concentrated rainfall (DMMC 2011; Gupta et al. 2017). Further, an intense rainfall event during 17-18 October 2021 reactivated an old landslide (Balianala Landslide, Roy et al. 2022b) south of the city, putting several important civil establishments of Nainital town, i.e., Government Inter College, etc. at peril.

Multi-temporal InSAR technologies (e.g., Persistent Scatterer Interferometry (PSI), Small Baseline Subset (SBAS)) use a large number of SAR images for computing displacement time series (Ferretti et al., 2001; Berardino et al., 2002). PSI and SBAS have acquired wide popularity in the last decade regarding deformation monitoring (Ferretti et al., 2001). PSI and SBAS methods are extensively used in landslide studies, such as landslide investigation and identification (Bonì et al., 2018; Tessari et al., 2021), landslide inventory mapping and activity assessment (Cigna et al., 2013), slow landslide displacement monitoring, mapping of landslide areas and understanding landslide kinematics (Schlogel et al., 2015; Rosi et al., 2018).

We have applied SBAS and PSI techniques to monitor the landslide-related creep on the slopes surrounding Nainital city. SBAS technique was used from October 2014 to September 2019 using more than 100 scenes of Sentinel-1 SAR images in ascending and descending passes (relative orbit: 129 and 63, respectively). The SBAS technique help in identifying the broad locales of slope movement. Further commensurate use of dual pass geometries helps resolve the slope motion to east and vertical components. Once the SBAS helped identify the broad locales, we further refined the observation using PSI technique over April 2020 – December 2021 using descending pass imagery. The PSI technique provides a more accurate estimate of the movement rate and helps identify exact locations of instability.

SBAS processing results show how the northeastern portion of the Nainital lakeside was affected by noticeable deformation characterised by a crucial westward component all along the slope, in accordance with the local morphology and a vertical component mainly affecting the upper part of the slope. Both the vertical and east-west deformation velocity reached a rate of 20 mm/year in the most destabilised sector of the slope.

In addition, the south-eastern zone of instability around the Nainital lake, then instability up the slope of the Balianala landslide, could be identified (Roy et al. 2022b). In this case, projected vertical and east-west deformation maps provided only limited spatial information related to this instability phenomenon, showing the crown area of an unstable slope, probably affected by fast deformation evolving in debris and rock falls, as it could be confirmed from an optical scene over the study area.

Observations from PSI results over a different time period compared to the SBAS further verify the later observations. Due to the general good coherence spread and location of houses, the PSI algorithm identified many point scatterers around the Nainital lake and on the slopes surrounding it. It is seen that the general area of instability, as specified by the SBAS method, is coincidental with the unstable PS locations on the northeastern part of Nainital lake. Herein the threshold value of velocity for which the PS points are considered to be unstable is kept at 5 mm/y. This threshold also ensures that the derived velocities are generally noise-free (Roy et al. 2022a). The cluster of unstable PS located on the northeastern slopes of the lake region records velocities as high as ~ 27 mm/y (along LOS). In addition to this, the upslope locations of the Balianala landslide also register high velocities consistent with the SBAS observations.

The commensurate use of SBAS and PS methods observes and records the stability of the slopes around the Nainital lake within the premises of the Nainital city. The methods complement and supplement each other in identifying the broader locales of the deformation and pinpointing locations of slope instability. Such observations are pertinent in towns located within the valleys of the Himalayas, where monitoring slopes around the urban settlements is paramount.

Acknowledgements

PR and TRM thank Deputy Director (RSA) and Director, NRSC, for their support and guidance. GT acknowledges the Swiss Development Cooperation (SDC) that supported SARMAP analyses in the framework of the projects implemented in India since 2015: “Strengthening State Strategies for Climate Action (3SCA)”. The authors also kindly acknowledge the European Space Agency (ESA) for making available the Sentinel-1 images in the framework of Copernicus activities.

References

Dobhal DP, Gupta AK, Manish M, & Khandelwal DD (2013). Kedarnath disaster: Facts and plausible causes. Current Science, 105(2), 171-174.

Valdiya KS (1988) Geology and natural environment of Nainital hills, Kumaun Himalaya, Gyanodaya Prakashan, Nainital, India 160.

DMMC (2011). Slope instability and geo-environmental issues of the area around Nainital. A Disaster Mitigation and Management Centre (DMMC) publication.

Gupta V, Bhasin RK, Kaynia AM, Tandon RS, Venkateshwarlu B (2016) Landslide hazard in the Nainital township, Kumaun Himalaya, India: the case of September 2014 Balia Nala landslide. Nat Hazards. 80(2):863–877

Roy P; Jain N; Martha TR; Kumar KV. (2022b) Reactivating Balia Nala landslide, Nainital, India—A disaster in waiting. Landslides, 19, 1531–1535

Roy P, Martha TR, Khanna K, Jain N, Kumar KV (2022a) Time and path prediction of landslides using InSAR and flow model. Remote Sens Environ 271:112899

Ferretti A., Prati C., Rocca F (2001). Permanent scatterers in SAR interferometry. IEEE Trans. Geosci. Remote Sens. 39, 8–20

Berardino P., Fornaro G., Lanari R. Sansosti E. (2002). A new algorithm for surface deformation monitoring based on small baseline differential SAR interferograms. IEEE Trans. Geosci. Remote Sens. 40, 2375–2383

Bonì, R, Bordoni M, Colombo, A, Lanteri L, Meisina C, 2018. Landslide state of activity maps by combining multi-temporal A-DInSAR (lambda). Remote Sens. Environ. 217, 172–190

Tessari, G, Kashyap, D, Holecz, F, 2021. Landslide Monitoring in the Main Municipalities of Sikkim Himalaya, India, Through Sentinel-1 SAR Data. In: Casagli, N, Tofani, V, Sassa, K, Bobrowsky, PT, Takara, K (eds) Understanding and Reducing Landslide Disaster Risk. WLF 2020. ICL Contribution to Landslide Disaster Risk Reduction. Springer, Cham. https://doi.org/10.1007/978-3-030-60311-3_19

Cigna F, Bianchini S, Casagli N (2013). How to assess landslide activity and intensity with persistent scatterer interferometry (psi): the psi-based matrix approach. Landslides 10, 267–283

Schlogel ¨ R, Doubre C, Malet JP, Masson F (2015). Landslide deformation monitoring with ALOS/PALSAR imagery: a D-InSAR geomorphological interpretation method. Geomorphology 15 (231), 314–330.

Rosi A, Tofani V, Tanteri L, Stefanelli CT, Agostini A, Catani F, Casagli N, 2018.The new landslide inventory of Tuscany (Italy) updated with PS-InSAR: geomorphological features and landslide distribution. Landslides 15, 5–19.

Abstract

The North-western Indian region is among the most groundwater-depleted areas globally due to the rapid 12-fold increase of bore wells during India's green revolution. The Built-up areas in these Himalayan piedmont fan regions are undergoing rapid urbanization and experiencing rapid groundwater depletion and water table drop. The rapid urbanization over groundwater-depleted areas triggers inelastic aquifer compaction, endangering future groundwater potential. We estimated the ground deformation over piedmont fans around urban areas of NW India during 2014-2022 using Interferometric Point Target Analysis (IPTA) using ascending and descending Sentinel-1 acquisition modes. The region is experiencing vertical subsidence up to ~50mm/yr with prominent hotspots. The analysis of the decadal groundwater level at these locations revealed that 55 percent of the tube-well indicated ~5-8m lowering during 2005-2018, leading to the abandonment of 5-10% of tube wells around Chandigarh each year. Global warming is exacerbating the situation, with the highest increase of heat wave events in NW India during the past five decades forcing overdependence on groundwater. The LULC change around the study region shows that the built-up areas have increased four times from 100 sq. km to 400 sq. km, with a 100% increase in population in the past four decades. Comparing the subsidence with the aquifer parameters from the bore wells suggests that the clay-confining aquifer level III and semiconfined level II are experiencing the highest subsidence. The stress-strain relationship of these hotspot regions reveals the inelastic compaction of the aquifers producing severe subsidence. This unsustainable groundwater exploitation often triggers The piedmont zones of the Himalayas with identical aquifer geometry and population growth facing similar challenges. The combined DInSAR-IPTA and observational groundwater data modeling could provide a robust assessment for effective groundwater-aquifer health monitoring and management.

We analyzed and discussed the formation of the decadal-scale ‘cone of depression’ in many parts of the Chandigarh piedmont region with respect to the aquifer profile and correlated it with the subsidence observed in DInSAR data. The time series DInSAR-derived ground subsidence was correlated with the hydraulic head to understand the aquifer deformation. We also correlated DInSAR-derived subsidence with groundwater overexploitation, aquifer characteristics, and urban-area recharge scenarios

Decadal groundwater level change vs. DInSAR subsidence

Overexploitation from the tube wells has an adverse effect on the water table in the piedmont zone around Chandigarh. The water table level decline is observed at 55 percent of the tube-well, with groundwater data indicating ~5-8m lowering during 2005-2018, leading to the abandonment of 10% of tube wells around Chandigarh each year. The groundwater level in the region dropped sharply from 2006-07. The precipitation pattern also declined sharply during the past 5-8 years, which may have aggravated the stress on the groundwater. The overexploitation of groundwater and the absence of recharge in an area led to the development of a groundwater depression cone on a regional scale, which could lead to ground subsidence. We compared ground subsidence and cone of depression along the five equal distance N-S profiles.

The region is experiencing a spatially varying static water table (SWT), showing a general decline in southern Chandigarh with peak values ranging from 0.5m/yr to 1.0m/yr with the distinct cone of depressions (Figs. 5b-f). The Kharar region is experiencing a sharp decline in SWT with a peak of >0.75m/yr, where the cone of depression coincides with the ~45 mm/yr subsidence along profile-01. The multiple cones of depressions of SWT with reducing WT decline rates with spatially coherent subdued cones of subsidence towards the distal part form a bowl of ~10km radius of influence. However, the cone of depression has a larger radius than the cone of subsidence. The maximum SWT decline of 0.8m/yr is observed along profile-2 with the cone of depression and subsidence (50 mm/yr) centered around Landran in the distal fan region. Although the declining SWT produces a wide bowl of depression with a>5km radius of influence, the cone of subsidence (~3km radius) remains confined to the peak SWT decline region around Landran. In the adjacent profile-03, the cone of SWT depression with ~0.65 m/yr peak decline and the cone of subsidence (> 60mm/yr) coincides at the Sohana region with ~3-5 km radius of influence. The cone of SWT depression shows a sharp decline to ~0.8 m/yr in the proximal fan region around Eastern Chandigarh, but the subsidence cone with >40 mm/yr peak value is observed further south along the profile-04. A localized cone of subsidence (~15 mm/yr) near the airport colony coincides with the ~0.4 m/yr SWT decline in the distal fan region. Further east along Profile -5, a localized cone of subsidence>30 mm/yr coinciding with the cone of SWT depression with a peak of 0.8 m/yr is observed in the distal fan region around Dera Bassi. However, the proximal part of the fan remained steady. The SWT decline and cone of depression-subsidence rates are spatially correlated, representing a sinkhole type of subsidence possibly due to the focused zone (akin to a single source) of groundwater overexploitation. The zone coincides with the expanding urban centers such as Kharar, Sohana, Landran, and Dera Bassi, which do not have any restrictions on constructing boreholes, unlike Chandigarh urban areas (located in the proximal part). The cross-correlation of SWT decline rate with the subsidence rate shows a good correlation (R=0.61) in the hotspot regions, though the subsidence depends on other aquifer parameters.

Aquifer characteristics and subsidence

Three aquifer zones are identified in the northern part of the Chandigarh piedmont fan, with the semiconfined Aquifer-I and II zones in the proximal part being dominated by boulders and gravels down to 150m depth, followed by the sand-silt interlayered with clay beds. The composition varies with decreasing grain size southwards. The confined Aquifer-III is composed of fine-grained sand with a 30m thick, soft clay confining bed with 1.5x10^-4 to 7.5x10^-4 storativity in the proximal part. Only Aquifer II and III extend southward towards the distal portion of the fan. The primary abstraction is from Aquifer-III (Pleistocene alluvium) at a depth of ~100 m, where the ~ 40m thick Holocene soft clay acts as the confining bed. In the proximal fan region, the pumping test suggests the discharge varies between 450-900 liters per minute (lpm) for a drawdown of 2.5-25m in the Aquifer-I. The discharge increases to ~1000 lpm in Aquifer-II and 2000 lpm from 30 thick zones at ~200 mbgl in Aquifer-III at the distal part. Due to composition and grain size, the semiconfined Aquifer-I and II experience better groundwater recharge. However, the groundwater level depth decreases southwards with an almost artesian condition in the distal part of the fan.

To understand the spatial relationship between subsidence and SWT decline with the piedmont aquifer characteristics, we plotted them along the NE-SW profile-xx' line. The profile extends from the Himalayan foothills at Khuda Alisher to the distal fan near Manakpur, south of Chandigarh, where the artesian type condition prevails. Along the profile- xx', two prominent cones of SWT depression and ground subsidence cones are observed in the proximal (East Chandigarh-s1) and distal (Sohana-s2) fan regions. The narrow (3-4km) cone of the SWT depression up to ~0.5m/yr corresponds well with >20 mm/yr subsidence cone around east Chandigarh region, where all three aquifers are present (Fig. 6a),whereas the confined aquifer in the distal part around Sohana region experienced >0.6 m/yr SWT depression corresponding to >50 mm/yr subsidence with a wider cone, which is higher by an order of the proximal part. The proximal part of the piedmont, such as Khuda Ali sher and Sector 23 experienced negligible subsidence or narrow SWT depression and subsidence cones, including the area around Kharar. This represents a point source over-exploitation in the unconfined and semi-confined Aquifer-I and II which has higher recharge potential. The shape of the ground subsidence curve corresponds linearly with the SWT decline curve in the distal part with a significantly larger spatial extent of >15 km (profile yy') across the confined artesian aquifer. In the distal part, multiple cones of SWT decline intersect, resulting in the combined effect on the drawdown which can lower the groundwater table rapidly, as observed elsewhere in piedmont zones. The cone of depression laterally proliferates in the artesian aquifers. The aquifer load is supported by artesian pressure pushing upward and downward against the confining beds. The over-exploitation decreases the artesian pressure profoundly, leading to the aquifer collapse, as observed in many artesian aquifers. The drastic increase of confining clay layer thickness in the distal fan region reduces the groundwater recharge in aquifers II & III. The reduced recharge is unable to compensate for the overall extraction in the Sohana and Landran area, leading to categorizing the region as over-. The confined artesian aquifer is possibly undergoing inelastic compaction due to unregulated over-exploitation, resulting in pronounced ground subsidence in the distal fan around Sohana and Landran. The stress-strain curve can be used to find the elastic and inelastic nature at different parts of the aquifer.

Elastic and inelastic compaction of aquifer

In the study area, the groundwater level variations are measured 2-3 times a year during the pre and post-monsoon periods (CGWB, 2022), whereas the DInSAR vertical deformations have a fortnightly frequency. Owing to limited time series data availability time series, we attempted to analyze the stress-strain relationship and S_k values for 4 locations, namely, Landran, East Chandigarh, Dera Bassi, and Manimajra. Of these locations, three sites are experiencing high ground subsidence (and overexploitation), and one site has no ground deformation. Many hysteresis loops in the stress-strain curve indicate an aquifer's elastic behavior and their absence indicates inelastic deformation. The hydraulic head of East Chandigarh, Landran, and Dera Bassi registered a lower hydraulic head than the pre-consolidation head (historical minimum hydraulic head), implying inelastic compaction. The Manimajra exhibit multiple hysteresis loops in the stress-strain relation curve, indicating elastic deformation, which registered a higher hydraulic head in December 2019 than in November 2014, suggesting optimal recharge. The inelastic compaction in the overexploited distal part is due to the lack of aquifer recharge associated with urbanization, such as decreasing rechargeable area, increasing water demand, etc. The same is analyzed using land cover changes with high-resolution satellite images.

Land Use and Land Cover (LULC) change: recharge potential vs. demand

The LULC change around the study region shows that the built-up areas have increased four times from 100 sq. km to 400 sq. km, with a 100% increase in population in the past four decades. We analyzed the impermeable (built-up) surface area change using satellite images for three hotspot regions, namely Sohana, Landran, and Kharar, for the period 2000-2020 experiencing severe >60 mm/yr subsidence. The current water usage in Chandigarh urban area is ~250 liters/person, far higher than the national average of 132 liters per person. The population of the Chandigarh municipality region (proximal part of the piedmont zone) increased to 1.2 million from 0.8 million, a rate of ~1.5 % per year between 2000-2020, whereas the population in the Chandigarh suburbs, including Landran, Sohana, Kharar, and Dera Bassi, has grown from 0.5 million to 1.0 million at the rate of ~3-4% per year during the same period. The two-fold population growth in the distal part is likely to increase similar groundwater demand and cause severe over-exploitation owing to unregulated groundwater exploitation compared to the regulated Chandigarh municipality area in the proximal fan.

Conclusions

The DInSAR-derived vertical subsidence in the Himalayan piedmont zone around the fast-growing urban center of Chandigarh was analyzed in a combination of spatial and temporal changes in groundwater extraction, aquifer property, and urbanization-driven LULC changes responsible for changing demand. The analysis depicted precarious overexploitation-driven ground subsidence, causing the inelastic compaction of confined aquifers in the Himalayan piedmont zone. The severity is aggravated by the increase in impermeable urban areas, which deprives the area of natural surface recharge. Further, the decline in precipitation during the last decade (which may be related to climate change) has worsened even in the otherwise artesian condition of the distal fan zones. These results have significant implications for aquifer management in growing urban centers in the Himalayan piedmont zones in the Indo-Gangetic region, which is one of the most over-exploited areas with fast-growing urban centers.

The city of Lisbon faces significant risk from geohazards such as earthquakes, floods, geotechnical risks, and landslides. This work focuses on the landslide risk for urban areas of Lisbon, using the example of retaining wall collapse in 2017, causing structural damage on the buildings downstream, injuring 1 person and dislodging 57 people.

The wall was constructed in 1955, embedded in the Santo André hill, covering a slope of approximately 20 m high. The causes of the collapse was related to rainfall, irrigation of the garden upstream, inefficiency of the wall draining system and the presence of clayey material.

Before the collapse, the wall movements were monitored using topographic targets. The topographical monitoring is now complemented with Sentinel-1 data prior to the event, from 2015 until the day of the collapse, using the PSI (Persistent Scattering Interferometry) processing service SNAPPING (Surface motion mAPPING) in the Geohazard Exploitation Platform (GEP).

The goal of this work is to analyze the ground and structural displacements prior to the wall collapse in the surrounding area of the case study, using the in-situ monitoring and the PSI time series acquired by Sentinel-1 from 2015 to 2017.

The overall LOS (Line Of Sight) displacement is ~11 mm and the average displacement velocity varies from 1 mm/year to 6 mm/year. These displacements could indicate a failure mechanism that needs to be understood to prevent future similar events and identify patterns and access the triggers of the ground displacement.

The in-situ data can be linked to the remote sensing data to establish the full picture of landslide trigger. Nevertheless, this type of analysis should be implemented to areas considered at risk, to constrain the long-term temporal evolution of motions and predict potential landslides.

The area of the Upper Silesian Coal Basin in Southern Poland is one of the biggest coal deposits in Europe, which is still under active underground exploitation. The land compaction in the areas of the works manifests with irregular in space and time subsidence processes, depending mainly on the mining schedule. This causes various environmental effects on the region but also affects significantly the local infrastructure due to the high rate and scale of the terrain changes triggered by the underground caving. The current study focuses on the aftermaths on the infrastructure – roads, community buildings, railways, bridges. For this purpose, we observed the deformations in the areas of interest by application of the conventional Differential Synthetic Aperture Radar Interferometry (DInSAR) for three sets of data – ascending and descending Sentinel-1 SAR images, and one series of ascending TerraSAR-X radar images, all of them covering similar period between November 2021 and April 2022. The DInSAR method is chosen over other advanced InSAR techniques like Persistent Scatterer Interferometry (PSI) and Small Baseline Subset (SBAS) due to their limitations in observing rapidly changing terrain with non-line character of deformation. We used the European Space Agency (ESA) processing tools within the Sentinel Application Platform (SNAP) with improved processing chain for masking out the low coherent pixels before the unwrapping stage. In addition, we performed statistical tests to ensure the proper threshold for defining the acceptable level of coherence.

The influence of the water vapor content in the atmosphere that affects the radar signal propagation is reduced at the post-processing stage. It is done by extracting a polynomial surface constructed for each interferogram on the basis of non-deforming pixels with stable coherence in time. During this procedure, also the reference point with highest coherence and lowest displacement is chosen and used for unifying the series of interferograms for each AOIs. The suggested approach significantly improves the statistical characteristics of the interferograms and brings the pixels distribution closer to the normal.

The results are validated by several methods – by comparison of each SAR data set with leveling data from two cycles of measurements performed in November 2021 and April 2022, and by comparison of the results from the two SAR sensors C-band of Sentinel-1 and X-band of TerraSAR-X at the points from the chosen infrastructure objects. The RMSE for Sentinel-1 results in comparison with the levelling data is estimated to 0.03 m, while for TerraSAR-X the RMSE is 0.12m as there were noticed bigger differences between the TerraSAR-X results and levelling in the range of the subsidence bowls, while for Sentinel-1 these differences are mostly constant. The latest finding supported the decision to adopt the Sentinel-1 values as reference for assessment of the Terra-SAR-X results for areas without available levelling measurements and constructing time series for the points from chosen infrastructure objects.

As a well-established technique, Differential interferometric synthetic radar (D-InSAR) for ground surface deformation monitoring has been shown in different case studies. However, temporal decorrelation and atmospheric phase (ATP) are major limitations for D-InSAR applications. Multi-temporal InSAR (MT-InSAR) is an effective tool to solve such limitations and to measure the displacements quickly and accurately. Nevertheless, all MT-InSAR algorithms can only obtain ground deformation in the case of enough SAR acquisitions. In recent years, more spaceborne sensors capable of collecting multi-polarization SAR images have been launched (e.g., Sentinel-1, ALOS PALSAR, GF-3), which allows us to use fewer InSAR pairs to obtain deformation. Based on the fact that the atmosphere delay and the deformation phase are independent of polarizations, in this study we propose a novel approach called wavelet decomposition multi-resolution correlation analysis (WDMCA), which can estimate deformation based on only two dual-polarization interferograms. The key idea of WDMCA is to extract common phase components between two interferograms in a wavelet domain based on feasible wavelet basis function and decomposition scale. The WDMCA method includes three steps, i.e., deformation area identification, atmosphere extraction and deformation estimation. Firstly, the ATP and deformation are common low-frequency signals in two interferograms, to separate them, the deformation is first masked in this research, and an automatic recognition algorithm of the deformation area based on the SAR signal spatiotemporal characteristics is further put forward. After that, the ATP and non-ATP signals in the two interferograms are separated based on wavelet transform, and the common ATP is subtracted from the original interferograms. Finally, the wavelet transform is reused to extract the common deformation signal from the residual phase. To illustrate the effectiveness of the proposed WDMCA method, a simulation test through ALOS PALSAR HH and HV polarization data is carried out. The results show that the accuracy of the deformation area recognition is 97.24%. The coefficient of determination (R²) between the extracted ATP and the simulated one is 0.960 and the root-mean-square error (RMSE) is 0.042 rad, in addition, the R² between the extracted deformation and the simulated one is 0.980 and the RMSE is 0.003 rad. To further validate the accuracy of the topographic residuals, we compare the remaining phase components with the simulated DEM residuals. The R² and RMSE are 0.871 and 0.011 rad in HH-polarized interferograms and 0.798, and 0.019 rad in HV-polarized interferograms, respectively. These results prove the validity and reliability of WDMCA method and indicate the great potential for deformation monitoring by using multi-polarization interferograms.

Landslides are caused by earthquakes, rainfall, snow melt and human intervention, resulting in significant casualties and property damage every year all over the world. Due to the influence of sampling strategy, the resulting probability of landslides using logistic regression (LR) can deviate considerably from the actual areal percentage of landslides. With the increasing threat of recurring landslides, susceptibility maps are expected to play a bigger role in promoting our understanding of future landslides and their magnitude.

A new method for estimating probable landslide volume and area is proposed, which combines empirical modeling with time series Interferometric Synthetic Aperture Radar (InSAR) data. The method was created to assess probable landslides in Hokkaido, where landslides can have a severe impact on people, damaging lives and livelihoods. A better understanding of potential landslide magnitude is required for developing effective landslide risk management. The ground displacement derived from InSAR ranges from -87 mm/y to -35 mm/y along the line of sight (LOS). As a result, a map depicting the scale of probable landslide activity might be created. This research provides valuable scientific knowledge to landslide hazard and risk management in the context of continuing terrain evolution. It also demonstrates that this methodology can be used to assess the magnitude of probable landslides and so give critical information to landslide risk management.

Landslides and mass movements are events that can be classified as catastrophic when they take human lives. In Colombia, given its geological and climatic context, it presents some areas susceptible to being affected by these dynamic temporary spaces. Monitoring and follow-up is an integral part of risk management, in order to mitigate and possibly prevent the loss of human lives and to be able to generate early warnings for possible evacuations and activation of emergency plans. There are worldwide methodologies for mapping areas susceptible to these events, based on cross-references of information at the level of thematic layers, in an environment of geographic information systems, which has an impact on the fact that areas or areas that are active may remain. due to instability and surface deformation and are vulnerable areas for life and civil infrastructure.

Worldwide, interferometric techniques with Radar images taken by satellite have positioned themselves as a novel and practical alternative to delimit active zones due to processes of instability and surface deformation. Due to the above, advanced DINSAR interferometry techniques have been used, in order to delimit and monitor areas, with some degree of instability, that can trigger large-scale processes due to landslides and rock and earth movements, in the municipality of Arbeláez. Cundinamarca with central project coordinates 74.4° west longitude and 4.1° north latitude and an area of 25,000 hectares.

Images from the Sentinel-1 program of the European Space Agency in sigle look Compex SLC format were used. The SBAS Small Base Line technique was applied to detect unstable zones in rural areas composed of vegetation and natural environments. On the other hand, the technique of permanent dispersers was applied, in order to evaluate and monitor urban areas and civil infrastructure of the municipality. A total of 27 images were used in descending mode, the ascending orbit was not used because the area does not have satellite information in this orbit.

As results, it was possible to identify, together with the municipal administration, areas that are active due to deformation processes that were unknown to them. It was also possible to map about fifteen areas affected by surface instability.

Background:

Seasonal alpine snow is affected by strongly varying meteorological conditions, with diurnal temperature cycles around the freezing point, snow and rain fall. Situations with pronounced vertical gradients of snow temperature interchange with periods of almost constant snow temperature profiles. As the snowpack develops over the season, it is repeatedly exposed to fresh snow accumulation, whereas older layers beneath contain snow at various stages of the metamorphosis often with intermediate melting and refreezing periods. As a result, the complexity of the snowpack increases throughout the course of the snow season with associated implications on the interaction of radar signals with the snowpack and the underlying ground. Typical traits of seasonal snow include (1) melt-freeze crusts at different snow depths leading to significant backscattering contribution at the their interfaces, (2) temporally and depth-varying anisotropy of snow microstructure, and (3) liquid water content that also varies with snow depth and time yielding fluctuating penetration depths of the radar signal as a function of time.

A number of spaceborne radar/SAR missions at various frequencies with mission objectives about snow parameter retrieval (snow mass / snow water equivalent and snow cover extent) are under investigation or being implement: the Copernicus Polar Ice and Snow Topography Altimeter (CRISTAL) [1] with altimeters at Ku-band (13.5 GHz / 500 MHz bandwidth) and Ka-band (35.75 GHz / 500MHz bandwidth) and the preparatory CRISTALair airborne instruments, the Terrestrial Snow Mass Mission (TSMM) [2] , the Copernicus Sentinel Expansion Mission ROSE-L [3] at L-band and the NASA-ISRO SAR (NISAR) mission [4] at L/S-band – and, previously, other mission concepts, such as Hydroterra (G-CLASS) [5] at C-band, and CoReH2O [6] at X/Ku-band. Consequently, in-depth knowledge on the temporal variation of the parameters, such as penetration depth and layer-wise scattering contributions, is required, as those play an essential role to retrieve temporal changes of snow parameters (snow mass, anisotropy, layering, liquid water content etc.) throughout a snow season [7].

Methods and Data:

Time series of tower-mounted rail-based tomographic radar measurements were acquired at daily intervals within the ESA SnowLab project at Davos Laret, Switzerland [8] over four snow seasons using the ESA SnowScat radar [9] and the ESA Wide-band Scatterometer (WBScat) [10-12] in SAR tomographic profiling mode. Fig. 1 contains an overview of the test site and the tower-mounted rail-based SAR tomography measurement setup at the test site Davos Laret, Switzerland. The radar measurements were accompanied by additional snow characterization (snow density, specific surface area, SWE from snow pits; SnowMicroPen [17] measurements, GNSS-derived SWE and LWC [18]) and meteorological data. In this contribution, we analyze several time series obtained with SAR tomographic profiling mode, which is a microwave imaging technique that allows to non-destructively probe the vertical layering of the snowpack by means of vertical profiles of radar backscatter, depth-resolved co-polar phase differences, and interferometric phase differences as sketched in [12,13]. The tomographic profiles are focused using a time-domain back-projection approach [14,15]. The time series of SAR tomographic profiles include frequency bands L/S/C-band, X/Ku-band and Ka-band, a complete set of which was acquired quasi-simultaneously during the season 2019/2020 with the WBScat radar.

Results:

In this contribution, we are going to present a comparison of time series of SAR tomographic profiles of snow of entire snow seasons measured at different frequency bands (including 1-6GHz, 12-18 GHz and 28-40 GHz) with time series of reference snow characterizations obtained nearby by means of snow pit and SnowMicroPen (SMP) measurements and with further auxiliary environmental parameters. As an example, in Fig. 2, a 2019/2020 time series of SAR tomographic profiles obtained at 28-40 GHz and auxiliary reference data are shown.

We also include further detailed analysis and comparisons on depth-resolved co-polar phase difference vs. anisotropy as well as analyses on the differential interferometric phase which can be linked to changes in delta SWE.

Discussion:

The high-resolution structural information contained in the time series of SAR tomographic profiles obtained during the ESA SnowLab campaigns allows to tackle important knowledge gaps on the interaction of microwaves with seasonal alpine snow: the time series of vertical profiles of radar backscatter retrieved from the three bands of the tower-mounted ESA WBScat radar instrument and the ESA SnowScat radar instrument provides insight into the relative change of location and intensity of radar backscatter within the snowpack (e.g. during melting and refreezing cycles) as a function of time and various parameters (e.g.: snow accumulation, snow mass (SWE), snow surface temperature, liquid water content).

The comprehensive time series of tomographic profiles allows one to compare the vertical distribution of radar backscatter versus total backscatter, backscatter trends perceived in the different polarization channels and their combination in the Pauli basis. The wide range of radar frequencies (1-40 GHz) covered with the WBScat-derived tomographic data show evidence of frequency-dependent backscatter trends including trends in the vertical distribution of backscatter over time. The results indicate that, except at the frequency band 1-6 GHz, substantial backscatter is contributed also by horizontal layers. For instance, it is found that, using the 9.2/12-18 GHz and 28-40 GHz bands, the tomographic profiles show substantial scattering at melt/freeze crust interfaces within the snowpack, depending on the snow conditions. The ground contribution is often not the strongest backscattering contribution also under completely frozen conditions. In addition, the tomographic data set also reveals layer-wise co-polar phase differences under dry snow conditions as an indicator of vertical stratification of the anisotropy of the snow microstructure. Depth-resolved co-polar phase differences show interesting spatiotemporally consistent patterns and variations for cold dry periods and refreezing periods mainly for the Ku-band and the Ka-band data. The co-polar phase profiles indicate clear variations correlated with fresh snow and its subsequent metamorphosis. Non-zero interferometric phase differences at the 1-4 GHz band coincide with periods of snow accumulation. For the higher frequency bands the interferometric signal is more challenging to interpret with phase wrapping being a contributing factor with increasing frequency. Coherence loss is evident for periods with wet snow, particularly, wet snow surface, when the signal hardly penetrates the uppermost layer of the snowpack, which can be tracked well in the time series of tomographic profiles.

Conclusions and relevance for future mission concepts:

We can conclude that main characteristic features found in seasonal snow – (1) multiple melt-freeze crusts at different snow depths leading to significant backscattering contribution at the interface with these crusts, (2) temporally varying penetration depths of active microwave signals due to liquid water content that changes with snow depth and time, and (3) depth- and temporally varying anisotropy of the snow microstructure – can be localized and tracked along the time axis. Their quantification and exploitation potential for snow mass and snow structure retrieval requires further in-depth mission-case-specific research. The high-resolution depth-resolved imaging of the interaction of the radar signal with the snowpack can be used to further develop and validate layered snowpack scattering models (see e.g. [19]) to advance the understanding of the scattering mechanisms in seasonal alpine snow. Due to the almost complete coverage of frequency bands relevant for spaceborne SAR missions – the WBScat tomographic data covers a spectrum from 1-40 GHz – and accompanied reference snow samples taken, the tomographic data sets provide a rich source of information to further study the interaction of active microwave with seasonal alpine snow with respect to specific spaceborne mission concepts at high spatial and temporal resolution. All relevant frequency bands such as L-band (ROSE-L, NISAR, ALOS2/4, SAOCOM) and C-band (Sentinel-1, Radarsat Constellation Mission, Hydroterra) are covered by the tomographic time series as well as the frequency bands of the dual-frequency mission concepts at Ku-band (low and high) (TSMM), X-band/Ku-band (CoReH₂O), and the Ku-band / Ka-band altimeter (CRISTAL). In addition, single-pass bi-static and multi-static mission concepts can also be studied with the wide-range of spatial baselines and quasi-simultaneous measurements available for each tomographic acquisition.

Acknowledgements:

This work was performed at Gamma Remote Sensing in collaboration with the WSL Institute for Snow and Avalanche Research SLF, Davos, Switzerland as part of the ESA-funded project: “Scientific Campaign Data Analysis Study for an Alpine Snow Regime SCANSAS (ESA SCANSAS), Contract No. 4000131140/20/NL/FF/ab. ESA SnowLab campaign and data processing: ESA/ESTEC Contract No. 4000117123/16/NL/FF/MG. Hardware extension (rail) to enable SAR tomographic profiling: ESA/ESTEC Contract No. 20716/06/NL/EL CCN3 and ESA Wide-Band Scatterometer (WBScat) development: ESA/ESTEC Contract No. 4000117123/16/NL/FF/mg.

References:

[1] Kern, M. et al. (2020): The Copernicus Polar Ice and Snow Topography Altimeter (CRISTAL) high-priority candidate mission, The Cryosphere, 14, pp. 2235–2251., https://doi.org/10.5194/tc-14-2235-2020, 2020.

[2] Derksen, C. et al. (2021): “Development of the Terrestrial Snow Mass Mission,” in Proc. IEEE Int. Geosci. Remote Sens. Symp., pp. 614–617. DOI: 10.1109/IGARSS47720.2021.9553496.

[3] Davidson, M., Chini, M., Dierking, W., Djavidnia, S., Haarpaintner, J., Hajduch, G. et al., "Copernicus L-band SAR Mission Requirements Document", European Space Agency ESA-EOPSM-CLIS-MRD-3371, no. 2, 2019.

[4] NISAR (2018): “NASA-ISRO SAR (NISAR) Mission Science Users’ Handbook,” NASA Jet Propulsion Laboratory. 261p.

[5] ESA (2022): “Report for Mission Assessment: Earth Explorer 10 Candidate Mission Hydroterra, European Space Agency, Noordwijk, The Netherlands, ESA-EOPSM-HYDRO-RP-3779, 131p.

[6] ESA (2012): “Report for Mission Selection: CoReH₂0,” ESA SP-1324/2 (3 volume series), European Space Agency, Noordwijk, The Netherlands.

[7] Tsang, L. et al. (2022): “Review Article: Global Monitoring of Snow Water Equivalent using High Frequency Radar Remote Sensing,” The Cryosphere, vol. 16, no. 9, pp. 3531–3573, Sep. 2022, DOI: 10.5194/tc-2021-295.

[8] Wiesmann, A.; Caduff, R.; Werner, C. L.; Frey, O.; Schneebeli, M.; Löwe, H.; Jaggi, M.; Schwank, M.; Naderpour, R. & Fehr, T. (2019): “ESA SnowLab Project: 4 Years of Wide Band Scatterometer Measurements of Seasonal Snow,” in Proc. IEEE Int. Geosci. Remote Sens. Symp., pp. 5745–5748. DOI: 10.1109/IGARSS.2019.8898961.

[9] Werner, C. L., Wiesmann, A., Strozzi, T., Schneebeli, M., Matzler, C., (2010): “The SnowScat ground-based polarimetric scatterometer: Calibration and initial measurements from Davos Switzerland,” in Proc. IEEE Int. Geosci. Remote Sens. Symp., pp. 2363–2366. DOI: 10.1109/IGARSS.2010.5649015.

[10] Werner, C. L.; Suess, M.; Frey, O. & Wiesmann, A (2019): “The ESA Wideband Microwave Scatterometer (WBSCAT): Design and Implementation”, in Proc. IEEE Int. Geosci. Remote Sens. Symp., pp. 8339–8342. DOI: 10.1109/IGARSS.2019.8900459.

[11] Werner, C.; Frey, O.; Naderpour, R.; Wiesmann, A.; Süss, M. & Wegmuller, U. (2021), “Aperture Synthesis and Calibration of the WBSCAT Ground-Based Scatterometer,” in Proc. IEEE Int. Geosci. Remote Sens. Symp., pp. 1947–1950. DOI: 10.1109/IGARSS47720.2021.9554592.

[12] Naderpour, R., Schwank, M., Houtz, D., Werner, C. L., and Mätzler, C. (2022): “Wideband Backscattering From Alpine Snow Cover: A Full-Season Study,” IEEE Trans. Geosci. Remote Sens., vol. 60, no. 4302215, pp. 1–15, 2022, DOI: 10.1109/TGRS.2021.3112772.

[13] Frey, O.; Werner, C. L.; Caduff, R. & Wiesmann, A. (2018): “Tomographic profiling with SnowScat within the ESA SnowLab Campaign: Time Series of Snow Profiles Over Three Snow Seasons”, in Proc. IEEE Int. Geosci. Remote Sens. Symp., pp. 6512–6515. DOI: 10.1109/IGARSS.2018.8517692.

[14] Frey, O., Werner, C. L., and Wiesmann, A. (2015): “Tomographic Profiling of the Structure of a Snowpack at X-/Ku-Band Using SnowScat in SAR Mode,” in Proc. EuRAD 2015 - 12th European Radar Conference, pp. 21–24. DOI: 10.1109/EuRAD.2015.7346227.

[15] Frey, O., Meier, E. (2011): “3-D Time-Domain SAR Imaging of a Forest Using Airborne Multibaseline Data at L- and P-Bands,” IEEE Trans. Geosci. Remote Sens., vol. 49, no. 10, pp. 3660–3664, DOI: 10.1109/TGRS.2011.2128875.

[16] Frey, O., Magnard, C., Rüegg, M., Meier, E. (2009): “Focusing of Airborne Synthetic Aperture Radar Data from Highly Nonlinear Flight Tracks,” IEEE Trans. Geosci. Remote Sens., vol. 47, no. 6, pp. 1844–1858, Jun. 2009, DOI: 10.1109/TGRS.2008.2007591.

[17] Schneebeli, M., Johnson, J. B. (1998): “A constant-speed penetrometer for high-resolution snow stratigraphy,” Annals of Glaciology, vol. 26, pp. 107-111, DOI:10.3189/1998AoG26-1-107-1.

[18] Capelli, A.; Koch, F.; Henkel, P.; Lamm, M.; Appel, F.; Marty, C. & Schweizer, J. (2022): “GNSS signal-based snow water equivalent determination for different snowpack conditions along a steep elevation gradient,” The Cryosphere, vol. 16, no. 2, pp. 505-531, DOI: 10.5194/tc-2021-235.

[19] Picard, G., Sandells, M., and Löwe, H.: “SMRT: an active–passive microwave radiative transfer model for snow with multiple microstructure and scattering formulations (v1.0),” Geoscientific Model Development, vol. 11, no. 7, pp. 2763–2788, 2018, DOI: 10.5194/gmd-11-2763-2018.

Multi-temporal Interferometric Synthetic Aperture Radar (MT-InSAR) is a powerful geodetic technique to monitor displacements of the Earth’s surface. It has developed into an operational technology in certain applications over time. However, challenging applications still exist, one of which is large scale displacement monitoring in regions with challenging atmospheric conditions, as latter lead to increased interferometric uncertainty over large distances.

Various approaches have been proposed to integrate ground truth into MT-InSAR, like Global Navigation Satellite System (GNSS) measurements, to correct for spatially correlated errors which are mainly caused by insufficient modelling of atmospheric disturbances. A set of these approaches is based on sampling spatially correlated errors in each interferogram at reference points with known displacements, interpolating the sampled error onto all other pixels and removing it from the interferograms. We here present a modification of this approach by taking the variance-covariance of the sampled error into account, which is comprised by the variance of the ground truth, the variance of the MT-InSAR displacement estimate as well as the covariance of the spatially correlated error. For this purpose, the mean covariance of the spatially correlated error is estimated in small-baseline interferograms to reduce the impact of displacements in the interferograms. Error cokriging is finally applied for the interpolation.

We compare the proposed method with alternative approaches in a simulation study and a real data study applying the Persistent Scatterer Interferometry (PSI) technique. For the simulation study, we simulated interferograms which mainly consist of spatially correlated atmospheric delays and to a much smaller degree of individual pixel noise. We compare the different integration methods for different numbers of randomly selected ground truth pixels and different ground truth variance scenarios.

The real data study was carried out with Sentinel-1 data-stacks acquired between 2016 and 2022 over the Vietnamese Mekong Delta (VMD) in descending and ascending orbits. The VMD has been subsiding for more than a decade with rates of up to several cm per year, but absolute reference points such as permanent GNSS stations are rare. We investigated two different application scenarios of the proposed method. In a first study, we concentrated on the north-western part of the VMD where several solid rock outcrops are embedded in the sedimentary delta. We assumed that these outcrops are stable reference areas in the considered time series and selected pixels located on them as ground truth points with presumably zero displacements. Finally, we expanded the study to the whole extend of the VMD. In this scenario, reference points from outcrops are only distributed in the north-western part of the study area. As the land subsidence in the VMD is mainly driven by compaction in the upper sediment layers, we used large bridges with very deep foundations as additional reference points throughout the VMD, whose stability we previously tested in a triangulation network.

In all studies, our method shows superior performance in reducing uncertainty at large distances compared to the other applied ground truth integration methods. We show how adding bridges with deep foundations as additional reference points in the second real data study further reduces uncertainties significantly. We finally discuss how the decrease in displacement uncertainty helps to analyze PSI displacement time series and the causes of land subsidence.

Monitoring of flood events with high resolution in both the spatial and the temporal domain is becoming more and more feasible thanks to the availability of long time series of images acquired by both synthetic aperture radar (SAR) and optical sensors [1]. Many approaches have been proposed; among the most promising, those which cast the problem of flood water detection into a Bayesian probabilistic framework [2, 3] allow to treat in a flexible way a variety of heterogeneous information, and give as output a probability value for the presence of water in each considered image sample, which can be easily interpreted in terms of confidence.
SAR temporal image stacks represent an ideal tool to monitor the presence of water over large areas and with high temporal frequency in a systematic way, given the relative insensitivity of microwave signals to the presence of clouds and other atmospheric phenomena, and the active nature of SAR sensors. Recent international initiatives aim at operational provision of this kind of maps globally [4].

We independently developed a procedure which exploits the high-frequency characteristics of sensors such as the European Sentinel-1 (S1) constellation to account for slow backscatter changes on land areas, based on the assumption that floods are temporally impulsive events lasting for a single, or a few consecutive acquisitions [5]. The Bayesian framework also allows to consider ancillary information such as topography and satellite acquisition geometry, which can be cast into prior probability distributions which taper to zero for locations unlikely to be flooded.

In this contribution, we expand the treatment to the modeling of InSAR coherence temporal stacks. We limit our analysis to SAR interferograms obtained combining subsequent acquisitions with the shortest temporal baseline, which in the case of the S1 sensor is of 6 days for most of the sensor lifetime (thanks to the availability of the twin sensors S1-A/B from 2016 up to December 2021), or 12 days for the remaining periods. This choice allows for the maximum contrast between flooded and non-flooded areas, as on the latter temporal decorrelation is minimized.

As in the analysis of backscatter intensities, we can express the posterior probability p(F|g) for the presence of floodwater (F) given the coherence g at a certain pixel and at a certain time t (assuming coherence between times t and t+1) as a function of prior absolute and conditioned probabilities, through Bayes' equation:

p(F|g) = p(g|F)p(F) / (p(g|F)p(F) + p(g|NF)p(NF)),

with p(F) and p(NF) = 1 − p(F) indicating the a priori probability of flood or no flood, respectively, while p(g|F) and p(g|NF) are the likelihoods for the coherence values, given the two events.
The flood likelihood can be estimated over permanent water areas, whereas, to estimate the likelihood of non-permanent water areas potentially interested by flood events, we consider the residuals of the time series with respect to a temporal model trend, assumed to be a smooth function, relying on the above mentioned assumption that flood events
appear as (negative) anomalies in a temporal coherence trend.
Proper care must be paid in these modeling efforts to take into account the intrinsic coherence statistics, which generally differs from that of SAR intensity signals [6]. Nevertheless, S1 coherence time series have been recently shown to exhibit smooth, periodic trends over agricultural areas in southern Italy in non-flooded conditions [7].
We use Gaussian processes (GPs) [8] to fit the time series. GPs are viable alternatives to parametric models, in which the trends of the data are modeled by "learning" their stochastic behaviour through optimization of some “hyperparameters” of an assigned autocorrelation function (kernel). Residuals with respect to such model can be used to derive conditioned probabilities and thus inserted into Bayes' equation.
We present some results of an analysis exploiting both SAR intensity and coherence S1 time series over an agricultural area near the town of Vercelli (Northern Italy), characterized by the presence of widespread rice paddies, and hit by at least a large flood from the Sesia river in October 2020. The test site appears particularly challenging for the temporal modeling, as rice paddies are periodically inundated for normal agricultural practices, causing variability in both SAR intensity and InSAR coherence.
Acknowledgements
Work performed in the framework of the RiPARTI project "Monitoring of extreme hydrometeorological events from high-resolution remotely sensed data (Monitoraggio di eventi estremi idrometeorologici da dati telerilevati ad alta risoluzione)", funded by Regione Puglia, Italy. Sentinel-1 data are provided by the European Space Agency.
References
[1] A. Refice, A. D'Addabbo, and D. Capolongo, eds., Flood Monitoring through Remote Sensing. Springer Remote Sensing/Photogrammetry, Cham: Springer International Publishing, 2018.
[2] A. D'Addabbo, A. Refice, G. Pasquariello, F. P. Lovergine, D. Capolongo, and S. Manfreda, "A Bayesian Network for Flood Detection Combining SAR Imagery and Ancillary Data," IEEE Transactions on Geosci. Remote. Sens., vol. 54, pp. 3612–3625, jun 2016.
[3] A. D'Addabbo, A. Refice, F. P. Lovergine, and G. Pasquariello, "DAFNE: A Matlab toolbox for Bayesian multi-source remote sensing and ancillary data fusion, with application to flood
mapping," Comput. & Geosci., vol. 112, pp. 64–75, mar 2018.
[4] B. Bauer-Marschallinger et al., "Satellite-Based Flood Mapping through Bayesian Inference from a Sentinel-1 SAR Datacube," Remote Sensing, vol. 14, no. 15, p. 3673, Jul. 2022.
[5] A. Refice, A. D'Addabbo, F. P. Lovergine, F. Bovenga, R. Nutricato, and D. O. Nitti, "Improving Flood Monitoring Through Advanced Modeling of Sentinel-1 Multi-Temporal Stacks," in IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium, Jul. 2022, pp. 5881–5884.
[6] R. Touzi and A. Lopes, "Statistics of the Stokes parameters and of the complex coherence parameters in one-look and multilook speckle fields," IEEE Transactions on Geoscience and Remote Sensing, vol. 34, no. 2, pp. 519–531, Mar. 1996.
[7] A. Refice et al., "Remotely Sensed Detection of Badland Erosion Using Multitemporal InSAR," in IGARSS 2022 - 2022 IEEE International Geoscience and Remote Sensing Symposium, Jul. 2022, pp. 5989–5992.
[8] C. E. Rasmussen and C. K. I. Williams, Gaussian Processes for Machine Learning. the MIT Press, 2006.

Snow cover is the main component of the cryosphere and the knowledge of its properties such as thickness, water equivalent, and freeze / thaw conditions, is relevant for the study of global cycle water and the climate system. The snow water equivalent (SWE) is the water content obtained from melting a sample of snow and can be defined according to the snowpack depth and density. Compared to optical sensors and radiometers, SAR is potentially able to provide SWE estimations at high resolution, independently from daylight and in any weather conditions. The estimation of SWE can be performed by exploring both the backscattering coefficient and the interferometric phase of SAR acquisitions.

The SWE estimation through differential SAR interferometry (DInSAR) [1] is based on the change of interferometric phase induced by changes on both geometrical path and propagation velocity of the SAR signal due to different SWE conditions between the two interferometric acquisitions. By assuming that dielectric inhomogeneities are much smaller than wavelength, we can neglect the volume scattering. By further assuming that snowpack is made by dry snow, the absorption of the microwave signal is negligible. Under these hypotheses, the backscattered SAR signal comes from the ground surface under the snowpack and the signal time delay related to the snowpack depends just on the snowpack depth and density. So, the DInSAR phase can be approximated as a linear function of the SWE changes [2] (due to a change in snow depth and / or density) occurred between the two interferometric acquisitions. This linear relation between DInSAR phase and SWE changes, involves also the incident angle and the wavelength, and holds for a snowpack consisting of dry snow and an arbitrary number of layers each of uniform density. Of course, due to the differential nature of the DInSAR measurements both in space and time, only SWE changes can be measured. Absolute SWE values can be inferred either by assuming that one of the two interferometric acquisitions is snow free, or by using a reference SWE value coming from independent measurements. Moreover, the SWE estimation from DInSAR phase presents some critical aspects typical of the interferometric measurements: i) phase aliasing, which limits the maximum measurable SWE variation; ii) undesirable phase components related to residual topography, atmospheric signal, and orbital errors; iii) interferometric coherence, which depends on the scattering properties of the resolution cell. Recently, this last issue has been investigated by using a multiband interferometric SAR sensor under controlled test site, observing critical DInSAR phase decorrelation conditions occurring even after few hours at shorter wavelengths. [3]. Therefore, by all above considerations, the retrieval of SWE through DInSAR is feasible only under conditions of dry snow and spatial homogeneity of snowpack properties and is hindered by phase decorrelation, aliasing, and presence of spurious signals. In particular, temporal decorrelation is due to several concurrent causes such as rain, wind, and temperature changes, and it represents a very critical issue to be faced with most of wavelengths and revisit times of nowadays spaceborne SAR sensors. That’s why, this approach, despite proposed more than two decades ago, does not yet allow reliable and operational SWE monitoring at large scale.

This work revises some of the issues related to the SWE estimation, and experiments the use of multifrequency SAR data for deriving SWE maps over Alpine mountains trough both DInSAR-based and SAR backscattering-based algorithms. Case studies in Val Senales and Val d’Aosta (Italy) were investigated, characterised by critical settings such as steep topography, limited size, and potential spatial inhomogeneous snowpack.

Preliminarily, we performed a theoretical analysis aimed at assessing the performance of DInSAR-based SWE estimation at X, C and L bands. By neglecting phase contributions coming from ground displacements, atmosphere and processing errors, the SWE variation can be related to DInSAR phase estimations, incident angle, and wavelength. This relation was used for assessing the precision of the DInSAR based SWE, showing that it decreases as incident angle and coherence increase and wavelength decreases. Moreover, it allowed to evaluate the impact of residual signals related the atmosphere, as well as orbital and topographic inaccuracies. Finally, by using the constraint needed to avoid interferometric phase aliasing, we derived for different values of wavelength and incident angle, the maximum SWE variation measurable unambiguously. This analysis is very useful for assessing the reliability of both radiometric and geometric characteristics of a SAR dataset to perform SWE estimation. The work illustrates example of this performance analysis carried out by exploring L, C and X bands and by set the parameters according to the datasets available for the processing in Val Senales. As expected, the L-band is the more robust with respect to the phase aliasing, leading to maximum measurable SWE variation of about 6 cm at incident angle of 35° Thanks to this, it is potentially able to catch all the SWE variations measured by a permanent ground station, while for both C and X bands some variations would lead to aliased DInSAR phase values and so unreliable estimation. Of course, the SWE variation depends also on the time interval between SAR acquisitions, so that short revisit time improves the performance. About this, the Sentinel-1B failure occurred on 23.12.2021 by doubling will certainly negatively impact on the SWE estimation.

According to the indications coming from the performance analysis as well as from a literature review, C and L band are the more promising to overcome some of the factors limiting the SWE estimation. For the present work a large dataset of Sentinel-1 data (345 Sentinel-1 SAR images acquired between 2015 and 2022 in Val Senales) were selected with the aim to explore the interferometric coherence over time and to exploit the short revisit time of the Sentinel-1 constellation for SWE estimation. SAOCOM data were also used, for taking advantage of the long L-band wavelength, which should guarantee SAR penetration into the snowpack, snow homogeneity, suitable values of interferometric coherence, and low probability of phase aliasing. Both Sentinel-1 and SAOCOM datasets were processed by adopting a “cascaded” interferogram formation approach, in which each image is paired to the one acquired in the next following date. This allows minimizing temporal decorrelation and estimating SWE changes from one date to the next. The time sequence of absolute SWE values was then reconstructed by integration and using a reference SWE value set by external data. Interferometric phase measurements are sensitive to atmosphere changes, in particular in mountainous sites due to the tropospheric stratified delay. This is due to the varying thickness of the atmosphere from pixel to pixel and is thus greater for sites with strong topographic variations, may vary significantly between acquisitions, and thus give rise to phase contributions, which may corrupt the SWE estimation. In order to identify and remove such atmosphere artifacts, we used the zenith total delay maps derived by the Generic Atmospheric Correction Online Service for SAR Inteferometry (GACOS) generated through processing of HRES-ECMWF model data. A stack of consecutive DInSAR phase fields, unwrapped and corrected by the atmospheric and orbital artifacts were generated and used to derive a stack of SWE change maps. In order to select pixels suitable for performing a valuable SWE estimation, a sensibility map was generated for each interferometric pair. First, the map combines geometrical information coming from orbits and topography in order to mask out pixels affected by layover and shadow. Then, by exploiting the model developed for the performance analysis, the minimum value of expected precision of SWE estimations is derived for each pixel. Finally, according to a coherence threshold, pixels for which the expected precision of SWE estimation is unreliable, are masked out in the sensitivity map. Both C-band Sentinel-1 and L-band SAOCOM datasets selected over the test cases were processes according to described processing strategy. The SWE estimations resulting from C- and L-band data were combined and analysed looking at their behavior in space and time.

Moreover, the demonstrated sensitivity of X-band backscattering to SWE of dry snow [4] was also exploited to derive SWE estimations in the test areas, by processing Cosmo Sky-Med (CSK) data. Following the strategy outlined in [5], a retrieval algorithm based on Artificial Neural Networks (ANN) was implemented, having as input the CSK data at the available polarizations (HH and VV) along with the local incidence angle, on which the backscattering is greatly dependent in areas characterized by complex orography. The forest cover fraction is also considered as ancillary input of the algorithm, with the twofold scope to provide a threshold for masking out the dense forests in which the SWE retrieval is not feasible and to be used as ancillary input in the retrieval for compensating the effect of sparse forests on the CSK measurements. ANN output is the SWE parameter. The algorithm has been trained by using in-situ SWE measurements from ground stations, which have been integrated by distributed SWE values simulated by a nivological model, to make the training more representative of the observed conditions and to extend the generalization capabilities of the algorithm.

The SWE estimations derived through this backscattering-based approach, may be fruitfully combined with those coming from the DInSAR approach with aim of: i) setting the reference SWE value needed to calibrate the DInSAR-based SWE measurements; ii) aiding the integration of SWE change values derived from the DInSAR approach; iii) supporting the analysis and validation of the DInSAR-based SWE measurements. Finally, where available, measurements from ground stations were also used the result analysis. The work describes some of the results obtained in the selected Alpine test sites, critically discusses advantages and limitations of the proposed approaches, and suggests possible future developments.

References

[1] T. Guneriussen, K. A. Hogda, H. Johnson, and I. Lauknes, “InSAR for estimating changes in snow water equivalent of dry snow,” IEEE Trans. Geosci. Rem. Sens., vol. 39(10), pp. 2101-2108, 2001.

[2] S. Leinss, A. Wiesmann, J., Lemmetyinen, and I. Hajnsek, “Snow water equivalent of dry snow measured by differential interferometry,” IEEE J. Sel. Top. Appl. Earth Observ. Remote Sens., vol. 8(8), pp. 3773–3790, 2015.

[3] J. J. Ruiz, J. Lemmetyinen, A. Kontu, R. Tarvainen, R. Vehmas, J. Pulliainen, and J. Praks, “Investigation of environmental effects on coherence loss in SAR interferometry for Snow Water Equivalent retrieval.” IEEE Trans. Geosci. Rem. Sens., vol. 60(4306715), 2022. https://doi.org/10.1109/TGRS.2022.3223760.

[4] S. Pettinato, E. Santi, M. Brogioni, S. Paloscia, E. Palchetti, and Chuan Xiong, 2013, The Potential of COSMO-SkyMed SAR Images in Monitoring Snow Cover Characteristics, IEEE Geosci. Rem. Sens. Letters, vol. 10(1) pp.9-13, 2012. https://doi.org/10.1109/LGRS.2012.2189752.

[5] E. Santi, L. De Gregorio, S. Pettinato, G. Cuozzo, A. Jacob, C. Notarnicola, D. Gunther, U. Strasser, F. Cigna, D. Tapete, and S. Paloscia, “On the Use of COSMO-SkyMed X-Band SAR for Estimating Snow Water Equivalent in Alpine Areas: A Retrieval Approach Based on Machine Learning and Snow Models.” IEEE Trans. Geosci. Rem. Sens., 60(4305419), 2022. https://doi.org/10.1109/TGRS.2022.3191409.

Acknowledgments

This work was carried out in the framework of the project “CRIOSAR: Applicazioni SAR multifrequenza alla criosfera”, funded by ASI under grant agreement n. ASI N. 2021-12-U.0.

Ice losses from the Greenland Ice Sheet (GrIS) have expanded rapidly in recent decades. Ryder Glacier (RG) is one of the major outlet glaciers that terminate in the Lincoln Sea on the northwestern GrIS, accounting for approximately 2% of the total GrIS drainage. Paying attention to its dynamic changes is crucial for understanding the mass balance of the entire GrIS. Contemporary studies indicate that, compared to other marine-terminating glaciers in the Northern GrIS, such as the Petermann Glacier, the RG has remained relatively stable in terms of calving events. This work aims to investigate the stability of RG over the past few decades by analysing its grounding line (GL) position. Knowledge of the GL position can contribute to estimating mass flux and mass budget, analysing ice-shelf melting, and evaluating ice-shelf stability.

We employ the Double Differential Synthetic Aperture Radar Interferometry (DDInSAR), currently considered to be the most precise and dependable remote sensing approach, to European Remote-Sensing Satellite-1 (ERS-1) and Sentinel-1 SAR images, to detect the change of GL position of RG from 1992 to 2021. Our analysis indicates a significant retreat of the GL (1-8 km) during this period, with a nearly eight-fold difference in the rate of retreat on the eastern and western flanks. This suggests that RG has been in an unstable stage in the past decades., which could result in substantial ice loss and a rise in sea level. To investigate the causes of the retreat, we combine the data on ice-shelf thickness variation, surface and bed topography, and potential subglacial drainage-pathway to reveal that basal melt is the primary driver of the significant migration of the RG. Uneven melting dominates the asymmetric retreat on the eastern and western flanks, which is due to the disparity in ocean heat at different depths, and the bed topography slope. Greater ocean heat and steeper slopes result in more intense basal melt, further contributing to GL retreat, and posing a threat to the stability of the ice shelf. The experimental findings also demonstrate that RG is likely to continue retreating with a more drastic change expected in the west, in the coming decades.

Merapi volcano, Indonesia, exhibits activity typical of andesitic volcanoes: effusive lava flows and dome emplacement alternate with explosive, sometimes very destructive events. Assessing the location, shape, thickness and volume of viscous domes is crucial to evaluate the risks associated with sudden pyroclastic density currents (PDCs). Here we take advantage of bistatic mode radar acquisitions, TanDEM-X data, to produce twenty-six Digital Elevation Models (DEMs) over the summit area of Merapi volcano, between July 2018 and September 2021. We calculate the difference in elevation between each DEM and a reference DEM derived from Pléiades images acquired in 2013, in order to track the evolution of the dome in the crater. Uncertainties are quantified for each dataset by a statistical analysis of areas with no change in elevation. The DEMs derived from the TanDEM-X data show very good agreement with the DEMs calculated from Pleiades optical images and local drone measurements made by the BPPTKG in charge of monitoring the volcano. In addition, we use the amplitude and coherence images to detect changes in the dome morphology. The dataset allows for quantitative tracking of magma emplacement and estimation of the effusion rate during the last two episodes of dome growth, in 2018-2019 and 2021 respectively. In particular, we show that the dome growth was sustained by a relatively small effusion rate of about 2900 ± 580 m3/day from August 2018 to February 2019, when it reached a height of 40 m (± 5 m) and a volume of 0.64 Mm3 (± 0.03 Mm3). From February 2019 onwards, the dome elevation remained constant, but lava was continuously emitted (at a rate around 810 ± 90 m3/day). Lava supply was balanced by destabilization southwards downhill. From September 2019, several explosions led to the destruction of the summit dome. Subsequently, several flank destabilizations occurred, with a loss of 40 m (± 5m) over 300 m on the south-west flank and an accumulation of material further down the slope. The DEMs of 2021 clearly show two new domes, with the central summit dome reaching about 80 m (± 5m) and the flank dome reaching about 50 m (± 5m) high. The new dome on the southwest flank appears to have developed at the point of maximum loss of topography induced by flank destabilization. This study highlights the strong potential of using TanDEM-X data to quantitatively monitor the domes of andesitic stratovolcanoes.

Volcanic eruptions threaten neighbouring populations. To mitigate the volcanic risk and give timely advice to authorities in charge of the evacuation, scientists try to forecast the occurrence of eruptions by monitoring volcanoes using both ground-based instruments and satellite remote sensing data in order to decipher signs of unrest. Once an eruption has started, the purpose of monitoring is to anticipate its evolution with time. Characterising the nature and the size of the structures forming at the surface of an erupting volcano and estimating lave fluxes are key to anticipate an eruptive transition from an effusive to an explosive regime. Such information is difficult to obtain during an eruptive crisis since some of the ground-based instruments might be out of service or destroyed and because the hazardous nature of the phenomenon may prevent scientists from going on the field. In this situation, Synthetic Aperture Radar (SAR) amplitude imagery could complement ground-based monitoring, providing an alternative means for tracking in near-real time the topographic changes at the surface of an erupting volcano. However, the requirements for this to be possible are, to our knowledge, not satisfied by any of the existing methods that use SAR imagery to detect and characterise topographic structures on volcanoes, at least quantitatively. Therefore, one must design a new method capable of reconstructing the morphology of syn-eruptive volcanic structures, assuming that information is limited to (a) the pre-eruptive topography and (b) syn-eruptive images coming from different SAR sensors with different viewing geometries. By incorporating a synthetic volcanic cone in a 2009 Digital Elevation Model (DEM) of the Piton de la Fournaise volcano, La Réunion island, and generating a synthetic SAR image from this modified DEM, we are able to reconstruct the shape and estimate the volume of the 2015-formed Kala Pelé volcanic cone from one single SAR image acquired in 2022 by the satellite Sentinel-1A. Our preferred synthetic cone is centred on a latitude of 21.25575°S and a longitude of 55.70475°E. It has a crater radius of ∼50 m, an external radius of ∼100 m, a height of ∼40 m and a volume of ∼0.6 × 10^6 m^3.

These values are in agreement with the actual location and geometry of Piton Kala Pelé. These results are promising and demonstrate the possibility to use SAR amplitude data in the monitoring of volcanoes, even though this ultimate goal has not been reached yet and many efforts still have to be made to automate the method and improve the temporal resolution of SAR data over volcanoes without degrading the spatial resolution.

MDA is developing CHORUS, a two-spacecraft SAR constellation consisting of both a C-band satellite (CHORUS-C) and a trailing X-band satellite (CHORUS-X). Together these provide a novel capability of wide area coverage combined with selective high resolution imaging through cross-cueing.

The two satellites may be independently tasked with CHORUS-C and CHORUS-X respectively providing 20 minutes and 3 minutes of imaging time per orbit. In addition, they may be operated with cross cueing where CHORUS-C imagery is acquired, downlinked, processed and analyzed in near-real-time and then the 1-hour trailing CHORUS-X is tasked based on the result.

Both satellites will follow the same mid-inclination (53.5°) orbit, which provides increased coverage over mid-to-low latitudes compared to near-polar orbiting systems. The orbit altitude will be 600 km and will provide full access to ± 62.5° latitude (89% global area) when combined with both left and right looking. The CHORUS orbit follows an approximately 10-day repeat cycle and is non sun-synchronous with the nadir local time increasing about 20 minutes per day.

CHORUS will provide both dedicated vessel detection modes and general purpose ScanSAR modes (20 m to 100 m resolution over 290 km to 700 km swath), multiple Stripmap Modes (8 m, 5 m and 3 m), and metre to sub-metre very high resolution Spotlight modes. For CHORUS-C, single, dual and compact-polarization will be available for all modes except high-incidence vessel detection modes, which are only available at single polarization. CHORUS-X acquires all modes at VV polarization.

The CHORUS-C design extends technology developed for RADARSAT-2 and the RADARSAT Constellation Mission (RCM) and makes a number of significant improvements to yield better revisit, broader swath coverage, lower noise, less data compression, faster data rates, and higher resolution. CHORUS-C will use dual receive apertures on all modes to significantly improve swath width and, as with RCM, will use stepped receive to improve the SNR and reduce range ambiguities.

Both satellites will provide repeat-pass InSAR capability with their Stripmap and Spotlight modes. Given the wide swath extents (120 km to 180 km) provided by the CHORUS-C 8 m and 5 m Wide Stripmap modes, we do not plan to support InSAR for the ScanSAR modes. The mid-inclination orbit significantly improves InSAR line-of-sight sensitivity to north-south axis surface movement compared to existing near-polar orbiting systems. This provides the opportunity to resolve surface movement in up to three dimensions when multiple complimentary image stacks are combined. Orbit tube maintenance and spacecraft attitude control will ensure sufficient repeat-pass two-dimensional spectral overlap to enable InSAR applications in both C- and X-band.

CHORUS is being designed for fast tasking through the Canadian Headquarters System and an extensive network of Global Ground Stations. Downlink will also use this same network as well as dedicated client network stations. CHORUS allows simultaneous imaging and downlink with guaranteed priority collections and will make frequent use of left/right slews to better respond to customer orders.

This paper will provide an overview of the CHORUS mission with a focus on parameters affecting repeat‑pass InSAR capabilities. Material will be updated from previous publications [1] to reflect the current program status.

References

[1] Sharma, Jayanti, and Ron Caves. “CHORUS – Changing How and When We Observe Our Planet.” In European Conference on Synthetic Aperture Radar, pp. 273–276. 2022.