The Multi-Pairwise Image Correlation (MPIC) processing chain, an end-to-end online service for Earth surface motion monitoring using optical imagery

Abstract:

Documenting ground deformation is important for a range of areas in Earth Sciences (such as earthquake, volcanoes, landslide and ice glacier monitoring and hazard assessment). The availability of optical satellite constellations with a frequent revisit time at medium to high spatial resolution (e.g. Sentinel 2, Landsat 7/8) provides the potential to contribute to ground deformation monitoring on a global basis. This observational capability also represents a challenge in term of storage capacity and computing resources which together with the complexity of the tuning of the different parameters, may prevent users from exploiting the data.

We proposed here a new version of the Multi-Pairwise Image Correlation for OPTical images (MPIC-OPT) algorithm. The new version of the algorithm proposes a complete chain to process optical images including data download, image pairs creation and advanced analysis of the displacement field. It offers the choice to compute the ground displacement associated to image pairs with two correlation techniques (MicMac, developed by IGN; GéFolki developed by ONERA). A new de-striping correction is integrated to correct jitter undulation.

The processing chain is accessible through the Geohazards Exploitation Platform (GEP) in the framework of the Thematic Exploitation Platform initiative of the European Space Agency and the runs are performed using the High Performance Computing facility at the A2S/Mesocentre of University of Strasbourg. Two services based on MPIC have been integrated on-line: MPIC-Mass-Movement (MPIC-MM) designed for the monitoring of glacier or landslides using long time series, and MPIC-EarThQuake (MPIC-ETQ) for the monitoring of co-seismic deformation.

The processing service was developed for Sentinel-2 and tested on two slow-moving landslides: Slumgullion (USA) and La Valette (France), on the European Alps glaciers and on the 2019 Ridgecrest earthquake sequence (USA). The new jitter correction significantly improves the displacement maps of the Ridgecrest earthquakes. The jitter undulations are successfully removed without decreasing the spatial resolution of the displacement maps and without modifying the measured coseismic displacement. The results on the landslides and the glaciers are in agreement with previous studies and field measurements.