Conference Agenda

The TanDEM-X DEM Change Maps product aims to provide global terrain change information that is particularly useful for various fields, including mining, glaciology, and forest monitoring. The product shows changes between data collected in 2017 and 2020-2022 and the original edited TanDEM-X DEM. It consists of multiple layers, including two DEM Change Maps with date layers, two Change Indication masks, the Edited TanDEM-X DEM, the DEM Edited Mask, and the HEM. The product is planned to be available at the 30m and 90m levels in 2023.

The product consists of two change maps to provide a unique timestamp for each pixel being compared. It's important to note that elevation changes in the DEM Change Maps correspond to topographic changes with respect to the global TanDEM-X DEM and not to a physical height changes of the same magnitude. The Change Indication Masks provide information on possible terrain changes and their reliability, but they are not a substitute for a thorough temporal elevation change analysis. Jumps between adjacent acquisitions may occur if the two acquisition dates are separated by several months, and no calibration is performed between them to preserve possible large-scale terrain changes. The provided edited TanDEM-X DEM is an edited version of the first global TanDEM-X DEM, and it was edited automatically by filling gaps and flattening water surfaces.

The DEM Change Maps will be extended locally to provide detailed information on changes to the Earth's surface over time in form of a stack. It is particularly useful for monitoring changes in topography due to natural disasters, land subsidence, glacier melting, or deforestation. The product enable monitoring changes in glaciers, ice sheets, coastlines, and forests on a global scale which is important for understanding the impacts of climate change.

The “Sentinel-1 Interferometric Coherence for Vegetation and Mapping”, SInCohMap, project (sincohmap.org) is an ESA-founded project with the objective of developing, analyzing and validating novel methodologies for land cover and vegetation mapping using time series of Sentinel-1 (S1) Interferometric (InSAR) Coherence.

The experiments and analysis carried out from 2017 to 2020 demonstrated the contribution of the time series of interferometric coherence derived from S1 data in the generation of accurate land cover and vegetation-type maps. This analysis was done over three different test sites (South Tyrol in Italy, Doñana in Spain, and West Wielkopolska in Poland) which are characterised by different classes and geographical features. The results obtained in the SInCohMap project showed that time series of interferometric coherence from both polarimetric channels are complementary sources of information for land cover. They can be exploited along the intensity to improve mapping classification (Mestre-Quereda et al. 2020). Experiments included many different classification algorithms and strategies, as detailed by Jacob et al. (2020), which demonstrate the robustness of the project outcomes.

Along with the development of the SInCohMap project, several topics were identified for further analysis. Thus, the SInCohMap project has been extended to explore three new aspects:

A) Improvement of the land cover classification when combining both ascending and descending acquisitions. They offer different observation geometry of the same scene as well as different acquisition times. It has been found very relevant over mountainous terrain to increase the spatial coverage of the maps by avoiding shadow and layover areas.

B) Improvement of the land cover classification when combining S1 coherence and Sentinel-2 optical imagery. The complementarity of information provided at these two wavelengths (optical and microwave) is relevant for some classes which are better identified at one or another.

C) Exploratory application of 6-day S1 interferometric coherence for forest monitoring and classification. The dependence on repeat-pass coherence upon forest characteristics has been studied.

The main results of these three new aspects will be presented at the conference.

References

A. Jacob, et al. “Sentinel-1 InSAR Coherence for Land Cover Mapping: A Comparison of Multiple Feature-Based Classifiers,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Vol. 13, pp. 535-552, January 2020.

A. Mestre-Quereda, et al. “Time Series of Sentinel-1 Interferometric Coherence and Backscatter for Crop-Type Mapping,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, Vol. 13, pp. 4070-4084, July 2020.

After a major disaster there is a need for responders to quickly identify locations that have sustained damage, such as building collapse or landslides. SAR satellite remote sensing is a useful tool for this because it has all-weather and day-and-night imaging capability and wide spatial coverage. Probably the most successful category of algorithms so far for making damage proxy maps from SAR data is based on detecting atypical decreases in interferometric coherence. Decreased coherence in interferograms spanning the event relative to pre-event coherence indicates that there has been a change in the ground surface or structures at that location. Coherence decrease is therefore used as a proxy for damage. These algorithms perform well in areas that usually have consistent and high coherence, such as urban areas, but they have low sensitivity for land cover types that already have low or variable coherence pre-event, such as forests, agricultural fields, or small villages surrounded by trees. Damage in these locations may not cause a noticeable decrease in the already-low coherence so is likely to be undetected by existing damage proxy mapping methods.

Here we present a new style of damage proxy mapping algorithm that jointly uses coherence and SAR intensity correlation. Intensity correlation involves calculating the correlation of pixels’ SAR backscatter intensity between a pair of scenes on different dates. Intensity correlation is typically less sensitive to subtle movement of SAR scatterers on the ground than interferometric coherence but more robust for low coherence land cover. By combining the two we aimed to produce an algorithm that was more versatile across diverse land cover types than either method individually.

We will show the results of applying our algorithm to case studies of landslides, liquefaction, and earthquake damage using data from the ALOS-2 and Sentinel-1 satellites. We will show performance analysis of our new algorithm compared to a benchmark coherence-only algorithm and assess how the performance varies depending on the pre-event coherence and the local terrain. By-products of our algorithm include rasters of the average pre-event coherence, spread in pre-event coherence, and shadow and layover masks. These are all related to the anticipated performance of the algorithm for a given pixel, and we present options to incorporate such information into the damage proxy map products for end-users.

To support the national scientific and industrial community in consolidating their expertise in algorithms development based on the integration of SAR data collected by different space-borne sensors, and help them grow in the field of downstream applications, ASI launched the “Multi-mission and multi-frequency SAR” program (2021-2023) [1]. Ten projects – two led by companies (project acronyms: MUSAR and CLEXIDRA) and eight by academia and research bodies (SARAGRI, MultiBigSARData, CRIOSAR, SMIVIA, COAST, APPLICAVEMARS, MEFISTO and DInSAR-3M) – were funded and kicked-off in 2021. In particular, five different R&D areas of specific interest were covered: agriculture, urban areas, cryosphere, sea and coast, natural hazards. The present paper reports the main achievements after two years and outlines future perspectives.

One of the key objectives of the program was to stimulate the national community to develop processing algorithms that enabled the integration of SAR data collected in different wavelengths. To this purpose, ASI facilitated the access to a wide spectrum of SAR data, in particular X-band COSMO-SkyMed First and Second Generation (CSK and CSG), and L-band SAOCOM. These data resources added onto the freely available collections from the Copernicus C-band Sentinel-1 constellation, and other SAR data that the research consortia themselves had already access to.

With regard to CSK and CSG, as of January 2023 a total amount of 4915 products was provided. The majority of CSK/CSG data were collected in StripMap mode, single polarization, and were exploited for interferometric and change detection analyses. However, CSG satellites were also tasked to collect series of quad-pol StripMap images to test the enhanced polarimetric capabilities compared to what previously allowed by CSK, for instance for maritime applications. Furthermore, by adding CSG satellites into the MapItaly Project since nearly the beginning of the program, the temporal revisit at X-band was much improved up to 1 day (tandem pairs). Benefits were therefore achieved for those applications that were time-sensitive and required higher frequency of observations. For example, the project SARAGRI [2] was provided with a total of 41 CSK/CSG images, collected by November 2021 with temporal revisit between 1 and 3 days. This proved essential to test and validate integrated C- and X-band tillage maps and assess the contribution brought by CSK/CSG with regard to regular 6-day revisit time Sentinel-1 observations.

With regard to SAOCOM data, the program was an opportunity for ASI to boost the exploitation phase of these L-band data within the geographic zone in which ASI has full utilization rights (i.e. the so-called “Zone of Exclusivity – ZoE”). Data were disseminated through the dedicated ASI SAOCOM portal [1,3] that was populated by ASI according to the research consortia’s requests. All the existing archive images were imported into the portal. Moreover, new acquisitions were specifically tasked. In the majority of the cases the new images were collected in StripMap mode, dual polarization, in order to concatenate with the available archive images to create regular time series and support both interferometric and change detection analyses. The tasking activity started in January 2022. After one year of implementation, most of the test sites in Italy are covered by at least 20 images per single geometry (compliant with the common threshold for an interferometric analysis to be reliably undertaken). Best revisit times were equal to 8 days in each geometry, i.e. the nominal value of the full SAOCOM constellation (i.e. SAOCOM-1A and 1B). Furthermore, ad hoc acquisitions of quad-pol StripMap images were successfully achieved. These data, for example, allowed the whole crop season of selected farmlands to be covered, and enabled the consortium of the project CLEXIDRA to investigate the benefit for model inversion and improvement of soil moisture retrieval, compared to dual polarization data.

The program enabled the consortia to achieve at least two objectives: (1) develop algorithms to process novel SAR data such as SAOCOM; (2) consolidate existing routines in order to integrate multi-frequency SAR data and generate new valued-added products to support civilian applications in the different R&D areas of ASI’s specific interest. To this purpose, the focus was on demonstrating the perspective of such algorithms for being engineered and brought to a pre-operational development stage.

With regard to objective (1), interferometric routines were developed to process SAOCOM time series, and were successfully tested on sites of known surface deformation and/or where ground-truth collection allowed validation of satellite measurements. For example, CNR-IREA leading the project DInSAR-3M developed the whole P-SBAS StripMap workflow for the SAOCOM-1 data exploitation [4]. The University of Naples (UNINA) leading the COAST project [5] proved the effectiveness of a ship detection pipeline on SAOCOM-L1A data.

With regard to objective (2), several consortia worked on algorithms and workflows to make the best out of the multitude of multi-frequency SAR data made available. For example, NHAZCA S.r.l. leading the project MUSAR [6] developed a data fusion algorithm allowing the retrieval of the three displacement components, better delineation of land subsidence patterns and a reliable estimation of the north-south component of the motion.

Other examples are provided by UNINA and the University of Naples Parthenope (UNP), leading the projects COAST and APPLICAVEMARS [7], respectively, who successfully extended the application of ship-wake and surface wind speed approaches to process CSK/CSG and SAOCOM data.

Finally, the paper will outline future perspective in relation to not only the national roadmap of scientific downstream applications [8], but also the international context. In particular we will show how the developed suite of algorithms pave the way for a more systematic exploitation of L-band data, given the expected data flow from new missions (e.g. ROSE-L) and more abundant multi-frequency SAR acquisitions with short temporal time span, if not even co-located, thanks to greater coordination and cooperation between space missions.

References

[1] D. Tapete et al., "ASI's “multi-mission and Multi-Frequency SAR” Program for Algorithms Development and SAR Data Integration Towards Scientific Downstream Applications," 2022 IEEE International Geoscience and Remote Sensing Symposium, pp. 4498-4501, 2022. doi: 10.1109/IGARSS46834.2022.9884937.

[2] F. Mattia et al., "Multi-Frequency SAR Data for Agriculture," 2022 IEEE International Geoscience and Remote Sensing Symposium, pp. 5176-5179, 2022. doi: 10.1109/IGARSS46834.2022.9884627.

[3] E. Lopinto et al., "Access to the SAOCOM mission over the ASI Zone of Exclusivity: features, approaches, results," ESA Living Planet Symposium 2022, Bonn, Germany, 26 May 2022.

[4] C. De Luca et al., "On the First Results of the DInSAR-3M Project: A Focus on the Interferometric Exploitation of SAOCOM SAR Images," 2022 IEEE International Geoscience and Remote Sensing Symposium, pp. 4502-4505, 2022. doi: 10.1109/IGARSS46834.2022.9884715.

[5] R. Del Prete et al., "Maritime Monitoring by Multi-Frequency SAR Data," 2022 IEEE International Geoscience and Remote Sensing Symposium, pp. 5188-5191, doi: 10.1109/IGARSS46834.2022.9884613.

[6] A. Brunetti, M. Gaeta and P. Mazzanti, "Multi-frequency and multi-resolution EO images for Smart Asset Management," 2022 IEEE International Geoscience and Remote Sensing Symposium, pp. 5192-5195, 2022. doi: 10.1109/IGARSS46834.2022.9883325.

[7] F. Nunziata et al., "Ocean Wind Field Estimation Using Multi-Frequency SAR Imagery," 2022 IEEE International Geoscience and Remote Sensing Symposium, 2022, pp. 5184-5187, doi: 10.1109/IGARSS46834.2022.9884018.

[8] D. Tapete and A. Coletta, "ASI’s roadmap towards scientific downstream applications of satellite data," EGU General Assembly 2022, 2022, EGU22-5643. doi: 10.5194/egusphere-egu22-5643

Performance monitoring of highway infrastructure is vital for the integrity of the transport network and the safety of the user. To date, the most time-efficient pavement monitoring is performed using Non-Destructive Testing (NDT) (e.g., ARAN, laser profilometer, Ground Penetrating Radar (GPR)) methods (Tosti et al., 2021). These methods allow a wider coverage and a more reduced temporal monitoring frequency compared to visual inspection and conventional destructive methods. In this context, the satellite remote sensing technology can task itself to further enhance the NDT standards and monitoring capabilities in terms of allowing full network-level coverage and a regular revisit of the infrastructure. Recent research has proven the viability of satellite radar observation for transport infrastructure monitoring (Gagliardi et al., 2023). However, these applications are not yet well established.

In this study, a new methodology is presented that uses medium to high-resolution SAR data to analyse the backscattered coefficient and evaluate the surface regularity, thus, the pavement surface quality. Interferometric SAR can provide information about the backscattering intensity in the form of coherence, which can reveal information about changes on the pavement. For this purpose, the medium-resolution Sentinel-1 and high-resolution X-Band SAR data are utilised. A standard InSAR data processing method is followed to calibrate the raw satellite data. The focus here is to implement a post-processing methodology to analyse the interferometric coherence and amplitude intensity to detect any change on the surface of the pavement. The main challenge in this study might be the noise from temporary scatterers from road traffic. To reduce the interference from transport users, scenarios with low traffic on the road are selected. The scattering property of the pavement is also analysed with different polarisation (e.g., Bashar et al., 2022), and the seasonal variability will be assessed. Outcomes are validated by way of comparison with the in-situ measurements of other non-destructive testing methods and observations. The innovative approach proposed here will give first-hand input towards large-scale pavement monitoring and will open an opportunity to explore further the capability of Radar Satellites for pavement and other civil infrastructure monitoring.

Keywords

InSAR coherence analysis; pavement surface monitoring; non-destructive testing; structural health monitoring

Acknowledgements

The SAR products utilised in this study are provided by ESA (European Space Agency) under the license to use.

References

F. Tosti, V. Gagliardi, L. B. Ciampoli, A. Benedetto, S. Threader and A. M. Alani, "Integration of Remote Sensing and Ground-Based Non-Destructive Methods in Transport Infrastructure Monitoring: Advances, Challenges and Perspectives," 2021 IEEE Asia-Pacific Conference on Geoscience, Electronics and Remote Sensing Technology (AGERS), Jakarta Pusat, Indonesia, 2021, pp. 1-7, doi: 10.1109/AGERS53903.2021.9617280.

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.

Mohammad Z. Bashar and Cristina Torres-Machi, “Deep learning for estimating pavement roughness using synthetic aperture radar data,” Automation in Construction, Volume142, 2022, https://doi.org/10.1016/j.autcon.2022.104504