CLOUDInSAR

The Sentinel-1 mission provides continuous all-weather, day-and-night radar imaging for various applications, including environmental monitoring and disaster management. One of the key capabilities of the Sentinel-1 mission is its use of Interferometric Synthetic Aperture Radar (InSAR) technology. InSAR involves the precise measurement of ground deformation by analyzing the phase difference between radar images acquired at different time.

The Sentinel-1 data is freely and openly available to the public, enabling researchers and decision-makers worldwide to access and utilize the data for various applications, including InSAR analysis. However, InSAR data analysis requires significant computational resources and expertise, creating challenges for many users. In this context, the main aim of the CLOUDInSAR project is to develop an open-source, cloud-based solution for processing and analyzing Sentinel-1 InSAR data. This system will be integrated in the Copernicus Data Space Ecosystem (CDSE), enabling the generation and analysis of various InSAR products such as coherence, interferograms, unwrapped interferograms, and displacement maps. By granting access to these products, CDSE will streamline the utilization of InSAR techniques within the Earth Observation (EO) community, thereby fostering the use of Sentinel-1 InSAR for applications in fields like natural hazard assessment.

Contact person: Mattia Callegari, mattia.callegari@eurac.edu

This project consists of five main implementation steps:

1. Sentinel-1 data indexing and access
2. Sentinel-1 SLC data pre-processing
3. InSAR algorithm implementation
4. Performance assessment
5. Use case evaluation

The first step is related to the implementation of the procedure to index the Sentinel-1 SLC data stored in CDSE object storage and provide a per-burst access. The per-burst access to the Sentinel-1 SLC data is one of the innovative elements that will be developed in the project and can ensure an optimization of the storage and processing resources consumption for the processing operations in the following steps. The documentation on data access procedure is the deliverable item connected to the first implementation step.

The second step is dedicated to the Sentinel-1 SLC data pre-processing: given a user defined area of interest, a time range and a track orbit number, the implementation which will be developed in this step will output a co-registered stack of all the SLC bursts which satisfy the criteria defined by the user. The algorithm for this will be provided as an OGC Application Package, so that it can be integrated into the openEO backend via a generic process for invoking such application packages (see openEO process graph in). This allows us to easily reuse existing software for the pre-processing step.

A parallelization strategy and an optimal integration into the CDSE cloud environment will be implemented, by determining optimal data access patterns and parameters.

The deliverable item of the second step will be a public GIT repository containing the code for Sentinel-1 SLC data preprocessing, the application package, and the related technical documentation.

In the third step all the InSAR algorithms will be implemented. These includes: interferogram formation, interferometric coherence, phase filters, phase unwrapping, phase to displacement inversion, geocoding. Output products will be compliant with open-source libraries for multi-temporal interferometry processing (e.g. MintPy). The deliverable item of the third step will be a public GIT repository containing the code of the InSAR algorithms, the application package, and the related technical documentation.

The fourth implementation step on performance assessment is divided in data provider and consumer performance evaluation. During the data provider performance assessment, we will build a standardized benchmark that relies on various metrics provided by openEO jobs to measure performance and cost. This allows us to further tune the integration if needed or determine performance bottlenecks. On the consumer side both qualitative and quantitative measures will be taken into account. A comparison with other state of the art processing will be performed and processing time with different computing resources allocation will be measured for all the implemented processing steps. The performances for different user inputs in terms of area of interest size and time period length will also be tested. The deliverable item of the fourth step will be a report on the performances.

During the fifth step two different use cases based on real application will be tested. The first case is related to the processing of a Sentinel-1 InSAR coherence time series to be used as input features for land cover classification. More in detail, the classification problem of debris cover glacier detection will be considered. This use case will also allow us to test the integration of Sentinel-1 InSAR products with optical images (e. g. Sentinel-2) within CDSE. The second use case is focused on the surface deformation monitoring through multi-temporal interferometry. This use case will be used to test the compatibility of the InSAR products generated with open-source libraries for multi-temporal interferometry processing and analysis (e.g. MintPy). The main deliverable of the fifth step will be a report on the scientific use evaluation.

The CLOUDInSAR project is divided into five Work Packages (WPs). The first WP is dedicated to project management and communication and Eurac Research is the responsible of this WP. WP 2000 and WP 3000 are related to the actual implementation of the proposed cloud-based system for InSAR processing. In detail VITO is responsible for the WP 2000 which deals with data indexing, access and pre-processing. Eurac research is responsible for WP 3000 dedicated to InSAR algorithms implementation. In WP 4000, with VITO as responsible, the performance of the proposed implementation will be tested. In WP 5000, with Eurac research as responsible, the proposed system will be tested on real use cases.

Activities to be performed in the year 2025:

WP1: Project management. Project coordination, reporting toward ESA, project outreach

WP3: InSAR algorithm development. DInSAR, InSAR time series and geocoding algorithms implementation

WP5: Use case evaluation. Land cover classification with InSAR coherence and InSAR time series processing and analysis for surface displacement monitoring