Conference Agenda

The ETAD product provides easy-to-use gridded timing corrections for Sentinel-1 level-1 data [1]. Such corrections are meant to enhance geolocation accuracy by compensating the effects of atmospheric path delay, Earth’s tidal deformation and other systematic effects not captured by the SAR image processor.

As a part of ETAD scientific evolution study, the capability of deriving accurate and consistent interferometric phase corrections from timing annotations is being assessed, both for conventional and multi-temporal InSAR applications (e.g. persistent scatterer interferometry). As already discussed in [1], this involves translating annotated time delays into phase offsets, and evaluating the corrections at the reference grid defined by the InSAR processing workflow.

For InSAR applications, only corrections resulting in a differential phase term between acquisitions are relevant. Some ETAD corrections might cancel out on interferogram formation, or be compensated during coregistration to the reference image geometry. There is, however, a set of correction layers available in ETAD that are considered relevant for the majority of scenarios [1][2]:

• Tropospheric range delay correction, which accounts for changes in the signal propagation velocity due to tropospheric conditions along the traversed path.
• Ionospheric range delay related to ionospheric activity, modelled as a function of the slant total electron content.
• Timing corrections related to solid Earth tidal deformations caused by the gravitational force of the Sun and Moon.
• Instrument timing calibration constant in range, which can lead to a differential phase term in case of changes in the ETAD configuration between the generation of two products, or in the instrument configuration between SLC acquisitions (also if S1-A and S1-B acquisitions are combined).

Ocean tidal loading is another well-known source of solid Earth deformation signal with a significant impact on InSAR time series in many coastal regions [3]. It is not a part of ETAD yet but we investigate the effect in our evolution study as a possible future ETAD product extension.

While most relevant corrections layers generally vary smoothly in space (e.g. solid Earth tides or ionospheric range delay), tropospheric corrections have a strong dependence on the topography. When applying ETAD corrections to high resolution InSAR data, with minimal or no multi-looking, accurate interpolation of the tropospheric corrections to the output grid is required, which involves accounting for the dependence on topography at the interpolation stage. Findings from the ETAD pilot study groups [1], and in particular from the IREA-CNR team, confirmed that neglecting this step results in artefacts in the (differential) ETAD tropospheric phase screens when applying the product to full-resolution Sentinel-1 interferograms. Our first experiments using a local estimate of the tropospheric range-delay-to-height-derivative to compensate for height effects during spatial interpolation have succeeded in providing meaningful differential tropospheric corrections for a common reference grid. Although the range delay to height derivative can be estimated (under certain conditions) directly from the available ETAD layers in its current version (planned to become an operational product by Spring 2023) it is foreseen that a more robust estimate is generated and delivered as an additional layer in a future release of the ETAD product.

In the final publication we plan to showcase the additional ETAD correction layers, including the tropospheric delay to height gradient as well as ocean tidal loading corrections. Study cases in the European Alps (strong topography) and French Brittany region (ocean loading) will be shown to assess the use of ETAD for InSAR corrections.

Acknowledgement

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.

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.

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., 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

[2] A. Parizzi, R. Brcic and F. De Zan: InSAR Performance for Large-Scale Deformation Measurement. IEEE Transactions on Geoscience and Remote Sensing, vol. 59, no. 10, pp. 8510-8520, Oct. 2021, doi:10.1109/TGRS.2020.3039006

[3] Yu, C., Penna, N. T., Li, Z., “Ocean tide loading effects on InSAR observations over wide regions,” in Geophysical Research Letters, 47, 2020. Doi: 10.1029/2020GL088184

[1] 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.

The Observational Products for End-users from Remote sensing Analysis (OPERA) project at Jet Propulsion Laboratory is supported by NASA to implement and produce multiple continental and near-global (all landmasses excluding Antarctica) products from remote sensing imagery. The OPERA products are defined to address the needs of the US federal agencies as identified by the Satellite Needs Working Group. Among the multiple products, OPERA is developing a ground surface displacement product from the Sentinel-1 data over North America. The OPERA project has decoupled the generation of the displacement products to two steps consisting of 1) coregistration of single look complex (SLC) images and 2) displacement time-series estimation. This decoupling has led to an additional OPERA product: a geodetically accurate and Coregistered SLC (CSLC) product from the Sentinel-1 data (CSLC-S1). The OPERA CSLC-S1 products will be produced with short processing latency and archived at NASA’s ASF DAAC where the products will be freely available to the user community.

In this presentation, we will introduce the OPERA CSLC-S1 algorithm. We present the baseline algorithm developed and implemented within InSAR Scientific Computing Environment Enhanced Edition (ISCE3). The algorithm accounts for the timing errors from environmental effects, SAR SLC processing approximations and solid earth displacements caused by the tidal effects and plate motions. Inspired by ESA’s Extended Timing Annotation Dataset (ETAD) algorithm, we investigate and demonstrate the impact of ETAD-like corrections on the geolocation and derived interferometric phase quality of the OPERA CSLC products.

We verify the algorithm by assessing the interferometric phase observations of pairs and triplets of interferograms, and by evaluating the estimated displacement time-series over permanent and distributed scatterers. We validate the CSLC products by evaluating the absolute geolocation accuracy using triangular trihedral corner reflectors and by assessing the relative geolocation accuracy using cross-correlation techniques. The algorithm verification results and the preliminary validation activities indicate that the baseline OPERA CSLC algorithm is capable of producing geodetically accurate stacks of aligned Sentinel-1 SLC products on pre-defined user-friendly geocoded grids through time satisfying the interferometric needs and ensuring high quality displacement time-series estimation.

In this paper we present an exhaustive experimental analysis aimed at testing the new available Sentinel-1 Extended Timing Annotation Dataset (ETAD) product for estimating and filtering out the atmospheric phase screen (APS) signal component from Differential Synthetic Aperture Radar (DInSAR) measurements.

The ETAD product consists of different correction layers which specify the azimuth and range timing shifts applicable to each burst of a Sentinel-1 TOPS data take to achieve precise geolocation for geodetic measurements in the centimeter accuracy range. The ETAD corrections can be applied in full or by selecting some layers to account only for specific effects. It is worth noting that, as assessed in [1], even if the ETAD product is not originally designed for interferometric phase corrections, it provides dedicated layers, based on numerical weather models, which may be converted into phase offsets to compute the APS corrections of the generated DInSAR products. In particular, these layers take into account:

i) Tropospheric range delay corrections associated with the refraction index variation due to changes of atmospheric properties like temperature, pressure and humidity along the path between the sensor and the point on the ground. These corrections strongly depend on the elevation of the considered area;

ii) Ionospheric range delay corrections evaluated based on the total electron content (TEC) of the ionosphere;

iii) Timing correction in range related to solid Earth tidal deformations due to the gravity of the Sun and the Moon;

iv) Instrument timing calibration in range which acts on the absolute phase difference compensating for possible changes in the instrument calibration between the SLC data acquisition or in the ETAD configuration between the generation of the two considered ETAD data [1].

In this work we focus on the exploitation of the ETAD correction layers accounting for the atmospheric path delays to retrieve and subsequently remove the APS from multi-temporal sequences of DInSAR interferograms generated at medium/high spatial resolution. More specifically, the interferometric products used for the analysis are generated through the P-SBAS [3] processing chain by exploiting a Sentinel-1 image dataset acquired over the Napoli bay area. It is worth noting that the P-SBAS interferograms are evaluated at the SLC full resolution and then multi-looked to obtain medium resolution products, i.e., with a 20x5 multi-look factor (range/azimuth, respectively), leading to a spatial resolution of about 80mx80m.

Methodology

The ETAD data are provided with a grid spacing on the ground of approximately 200 m for the entire data take; this means that they are sub-sampled of a factor 52 in range and 14 in azimuth with respect to the corresponding S-1 SLC full resolution burst images. Therefore, to generate the APS signal relevant to an interferometric pair of S-1 acquisitions, by using ETAD products, we follow the approach described in the documentation [1], which is here summarized:

i) Select the ETAD correction layers accounting for the atmospheric signal contributions relevant to the tropospheric delay (troposphericCorrectionRg), the ionospheric delay (ionosphericCorrectionRg), the Solid Earth Tidal displacements (geodeticCorrectionRg), and the instrument timing calibration [1];

ii) Resample the selected ETAD layers to the SLC burst resolution by applying, following [1], a bilinear interpolation step for the azimuth and range times;

iii) Apply the SAR SLC master to secondary image co-registration parameters available from the interferometric processing to the selected ETAD correction layers;

iv) Compute the differential range delay correction by summing the tropospheric, the geodetic and the instrument timing calibration correction layers and subtracting the ionospheric one, as explained in [1]:

= (troposphericCorrectionRg + geodeticCorrectionRg – ionosphericCorrectionRg + burst:instrumentTimingCalibrationRange )master – ( troposphericCorrectionRg + geodeticCorrectionRg – ionosphericCorrectionRg +

burst:instrumentTimingCalibrationRange )secondary, wherein the exploited symbols are self-explanatory;

v) Convert the differential range delay correction to interferometric phase

Note also that, that even if in [1] it is reported to subtract the computed ETAD APS from the interferograms, the proper step to correct the phase is achieved by adding the ETAD APS to it.

As further remark we underline that we performed the operations from i) to v) at the S-1 burst level and at the SLC full spatial resolution. After that, we mosaicked the burst interferograms and finally applied the multi-look operation, with 20x5 looks (range, azimuth) and obtaining, as already said above, a final resolution of 80mx80m. Accordingly, our approach is different from that presented in [2], where the S-1 burst interferograms are firstly multi-looked to approximately the resolution of the ETAD products (i.e., with 51x15 looks), then mosaicked to generate wide area interferograms and subsequently corrected by applying the ETAD APS corrections generated at the native ETAD data resolution.

Experimental Results

In agreement with [2], by considering the ETAD corrected interferograms generated at very coarse resolution (200m), the APS filtering procedure appears to work properly, as shown in Fig. 1, where we depict a cut of the 10012020-22012020 S-1 full slice interferogram over the Napoli bay area, generated through the P-SBAS processing chain, with a multi-look factor of 20x100 (azimuth, range) before and after the ETAD correction.

However, by performing a more detailed examination and analyzing the ETAD corrected interferograms at medium/high resolution, several artifacts, which are caused by the applied APS correction, become evident, as shown in Fig. 2. These artifacts are mostly present in areas characterized by a significant topography gradient and they often follow patterns similar to the foreshortening and layover effects. Therefore, they seem to be highly correlated to the DEM characteristics. Consequently, following an extensive analysis of the ETAD products and the interaction on the obtained results with the ESA and DLR colleagues involved in the ETAD test pilot activities, we came to the conclusion that the identified artifacts are caused by the DEM height variations due to the different projection within the specific range-azimuth grid of each S-1 burst image. Indeed, the ETAD layers, in particular the tropospheric ones, are computed on a data-take by data-take basis, which involves geolocation of ETAD's grid onto that of the DEM one. Such artifacts are clearly visible if we analyze the difference between the co-registered DEM layers corresponding to bursts acquired at different times and they show the same features of the artifacts retrieved in the corresponding ETAD atmospheric corrections (see Fig. 3).

The analysis presented in this work clearly highlights the limitation of the current version of the ETAD products, consisting in the presence of artifacts in the correction layers to be exploited for the APS filtering of DInSAR measurements generated at medium/high resolution. The obtained results have been achieved within the framework of the S-1 ETAD test pilot activities and have proved to be very useful to identify such problem and its cause in the testing phase, so that in the future the current ETAD products can be extended and improved for their use in advanced DInSAR scenarios.

Nevertheless, in order to overcome the presented limitation by exploiting the ETAD correction layers currently available, we developed a simple methodology for generating stacks of coregistered tropospheric correction layers, starting from the ETAD original ones, but referring to a unique DEM product, achieved in the range-azimuth grid by averaging the ETAD DEM layers relevant to the acquisition time series, so overcoming the problem of the afore-mentioned height variations.

The approach we implemented can be summarized in the following steps:

Considering a stack of coregistered ETAD tropospheric phase correction layers:

• Based on the assumption that the tropospheric phase signal is mostly linearly correlated with the topography [4], we divided each ETAD tropospheric phase correction layer into small patches partially superimposed, for which we calculated the parameters of the phase/elevation linear regression, by using for each ETAD layer its own DEM layer. Note that the patch size is chosen small enough to retrieve very small values of standard deviation for the linear regression;
• We used the so calculated linear regression parameters, properly interpolated, to estimate new tropospheric phase layers which are all referred to the computed average DEM, which is assumed as the reference one;

REFERENCES

[1] T. Fritz, L. Krieger, C. Gisinger, and M. Lachaise, “S1-ETAD Project Product Definition Document,” ESA Technical Document, Doc. ETAD-DLR-PS-0002, Iss. 2.1, Date16.06.2021, 2021.

[2] C. Gisinger et al., "The Extended Timing Annotation Dataset for Sentinel-1—Product Description and First Evaluation Results," in IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1-22, 2022, Art no. 5232622, doi: 10.1109/TGRS.2022.3194216.

[3] 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.

[4] Romain Jolivet, Raphael Grandin, Cécile Lasserre, Marie-Pierre Doin, G. Peltzer. Systematic InSAR tropospheric phase delay corrections from global meteorological reanalysis data. Geophysical Research Letters, 2011, 38, pp.L17311. 10.1029/2011GL048757 . hal-00657439

Using Interferometric Synthetic Aperture Radar (InSAR) data to observe the coseismic deformation on the Earth’s surface is now an established method in earthquake geodetic studies. However, the majority of earthquakes measured with InSAR are shallow events (depth < 30 km) whose surface displacement signals are relatively easy to capture, even for smaller magnitudes (M_w ~5.0) when these are very shallow. Conversely large, intermediate-depth (M_w > 6.5, 70-300 km depth) earthquakes, which are usually located in subduction zones, are rarely the focus of geodetic work, due to the efforts required to establish if a ground deformation signal can be robustly observed. Here we present a case study of an M_w 6.8 earthquake with a 112 km centroid depth which occurred on 3 June 2020 in Chile. We perform ~4 years of Sentinel-1 InSAR time series analysis (spanning Jan 2018 to Nov 2021) over the potential deformation area to better resolve the coseismic deformation that may otherwise be masked by atmospheric noise in single interferograms.

Due to the high Total Electron Content (TEC) in Northern Chile, especially for ascending data acquired in the morning, we also apply the split spectrum method to correct the ionospheric delay in addition to the tropospheric correction. We assess the performance of the split spectrum algorithm and find that it greatly improves the quality of data on ascending (33.7% standard deviation reduction), while making it worse on descending (5.0% standard deviation increase). We later will compare the ionospheric component derived from the split spectrum method to that from the Sentinel-1 Extended Timing Annotation Dataset (ETAD), as well as from the model-based approaches, to explore the impact of the ionospheric correction on Sentinel-1 time series at low latitude region.

After doing both ionospheric and tropospheric atmospheric noise correction, and masking the pixels which contain unwrapping errors or show a high fading signal bias (> 3mm/year), we successfully observe this deep earthquake (with peak displacements < 10 mm) on time series data and retrieve the coseismic deformation field using Independent Component Analysis (ICA) approach. Combining with the independent observations from Global Positioning System (GPS), we obtain the earthquake source parameters using a numerical model and compare them to those from seismology. We later also do joint inversion of geodesy and seismology to achieve better constrain of the fault geometry. Our work demonstrates that the significant surface displacements caused by large intermediate-depth earthquakes in subduction zone are observable, and shows the capability of InSAR for tracking these small magnitude deformation signals with sufficient archives of data.

A key indicator of potential and ongoing volcanic activity is deformation of a volcano's surface due to magma migrating beneath it. The European Sentinel-1 radar archive contains a large number of examples of volcano deformation, and provides an opportunity to build a database that can be used to train deformation-based volcano monitoring algorithms. We therefore aim to systematically extract all deformation signals at volcanoes globally, including smaller scale signals associated with processes such as landslides and local changes in hydrothermal systems.

We have developed an approach to automatically derive high-resolution displacement time series at all subaerial volcanoes. To avoid the loss of decorrelated signal in areas of winter snow and seasonal heavy vegetation, we build a highly redundant small baseline network of interferograms, tailored to each volcano using coherence tests. We implement an improved phase unwrapping algorithm, which separately unwraps signals at different spatial scales, to achieve better results in decorrelating areas. To mitigate the effect of phase propagation through the atmosphere, we provide multiple atmospheric correction methods, including a spatially-varying scaling method, which uses interferometric phase to refine the interpolation of a weather model in time and space. Moreover, we remove points with phase loop closure errors from each interferogram and exclude non-redundant interferograms during the small baseline subset inversion, resulting in more precise measurements.

Our processor was designed for Sentinel-1 Synthetic Aperture Radar (SAR) data, but we have adapted it to automatically process non-Sentinel-1 SAR data acquired over volcanoes, including images come from ESA’s ERS1, ERS2 and Envisat, as well as from other satellite missions such as TerraSAR-X/TanDEM-X, COSMO-SkyMed, Radarsat-1/2 and ALOS/ALOS-2. To deal with more variable perpendicular baselines in older data, we incorporate the coherence test algorithm to select interferometric pairs with good coherence. Furthermore, we update our atmospheric correction module to make it compatible with low-resolution weather model data, and so allow us to operate with data that is several decades old from legacy satellites.

The resulting products, stored in a database with annotated metadata (VolcNet), are available for further interpretation. We show how volcanic unrest at a large number of volcanoes worldwide can be identified in this database using the LiCSAlert algorithm. We demonstrate that spatial patterns of volcanic deformation can be detected and localised from the processed products. Based on the derived high-resolution displacement time series, we also show a statistical analysis for the assessment of volcanic risk.