Regional scale satellite monitoring for hydrogeological risk reduction

In the last decade satellite remote sensing has become an effective tool for monitoring geo-hazard-induced ground motions, and has been increasingly used by scientific community. Geo-hazards direct and indirect costs are currently rising, causing serious socio-economics and casualty losses. Therefore, creating a priority list turns out to be essential to highlight the most relevant ground deformations and to better focus the risk management practices at regional scale. The Sentinel-1 constellation, thanks to the 6-days repeatability and the free availability of the data, allows to easily update the geo-hazard-induced ground motions, compared to other kind of satellite sensors. In this PhD Thesis, the potentialities and drawbacks of the interferometric technique have been presented and then exploited to define three different procedures, applied in different environment and at different scales, for the use of multi-band PSI products. This thesis work represents the main outcome of a three yearslong activity at the Department of Earth Sciences of the University of Florence, Centre technologic de telecomunicacions de Catalunya research center and TRE-Altamira (Barcelona). The main goal is to test and evaluate the potential and applicability of space-borne SAR data, processed by means of different PSI approaches, as operational tools for the characterization of geohazards in different geological and geomorphological environments. For this work, two Italian test areas at regional scale are been selected: Tuscany and Valle d’Aosta Regions. Furthermore a site at detailed scale was analyzed, the mining area of Saline di Volterra (Tuscany). The main goal of the thesis hinges on illustrating different methodologies that could be merged in one single workflow to detect active moving areas, characterize them in detail and cross-correlate the satellite data with ancillary information, implementing the obtained products and results in the Civil Protection chain and geohazard risk management. The proposed case studies were intended as examples, although referred to different environments and geohazards, for the working approaches to be used from regional to detailed scale. For the Tuscany Region, there were exploited Sentinel-1 images for active moving areas detection at regional scale. A hotspot-like methodology was used, exploiting the temporal repetitiveness of Sentinel-1 data analysed by means of the SqueeSAR algorithm to create deformation maps in three different periods with a 6 months update. Thanks to a filtering approach based on a velocity threshold, it was possible extract a total of 652 deformation clusters, divided in three different periods, to study their spatial and temporal evolution. The final output is a flexible geo-database that contains interferometric parameters, geographical, geomorphological and geological information, a brief evaluation of the possible triggering cause and information about the temporal evolution of the moving areas. For the Valle d’Aosta Region, it was used a clustering analysis applied to a large stack of Sentinel-1 satellite interferometric products that has been derived using the SqueeSAR algorithm. Valle d’Aosta Region is an alpine region characterized by a wide spectrum of mass wasting phenomena. The approach, based on simple GIS tools and indexes, allowed detecting 277 moving areas above the selected velocity threshold. Overall, landslides (complex, rotational, Deep Seated Gravitational Slope Deformation (DSGSD)), rock glacier evolution and detrital-related deformation are responsible for the detected motions. In mountainous areas, where the field data collection is sometimes limited or impossible, the presented approach is intended to create priority areas to be focused for further investigations. In this way, it is possible to increase, with reduced economic and personnel costs, the “landslide knowledge” of all the actors involved within the landslide risk management chain at regional scale. In this context, easily updatable clustering methodologies are very useful tools for MTInSAR data analysts; it is possible to obtain reliable results in a fast way and to compare them with previous results. Well knowing the limitations of the interferometric technique, especially in mountain regions, it is reasonable to rely on clustering approaches in order to derive multi-temporal synoptic views of ground motions over wide areas. For the Saline di Volterra case, it is presented a local scale application of multi-temporal satellite interferometry targeting a solution mining area in southern Tuscany. The surroundings of Saline di Volterra host several brine wells that pump water into a salt level at a depth ranging between 60 and 400 m below surface. The mining activity has a relevant environmental impact in terms of depletion of the water resources and in terms of ground motion, creating several sinkholes which were mapped through multi-temporal analysis of orthophotos. The deformation map, obtained through the analysis of Sentinel-1 images, revealed the presence of several subsidence bowls, sometimes corresponding to sinkholes formed in the recent past. The subsidence bowls have a common deformation pattern, with LOS velocities increasing forward the center of the bowl. The temporal evolution of the measurement points can vary a lot on case-basis. Finally, a correlation between LOS velocities and age of formation of sinkholes have been found. The Sentinel-1 images were processed with an ad hoc processing chain, elaborated in Centre Technologic Telecomunicacions Catalunya research center (Barcelona). With this new procedure, it was possible to detect fast deformation rates that are usually puzzling to solve in mining areas. This detailed scale and target-oriented approach demonstrated its capability to provide useful information in terms of density of measurement points and quality of the time series.