Semi-automatic analysis of InSAR large datasets for landslide mapping and monitoring: the Great Britain case study

In recent years, satellite-SAR interferometry has established itself as a widely used global monitoring technique, enabling timely detection, monitoring and mitigation of both natural disasters and human-induced ground movements. When dealing with large multi-temporal InSAR data, the ground deformation detection becomes a fundamental and complex task with the consequent pressing need to establish new approaches and tools for effectively analysing large interferometric datasets. The advanced capabilities of the satellite systems and the continuously updated processing techniques provide unprecedented amounts of data to analyse the ground deformation processes for large territories in reduced time frames. Within this context, the fast detection and characterization of the ground deformation processes constitute a milestone for what concerns the correct management and mitigation of their impact on vulnerable populations and infrastructures. As a result, the starting point for all ground deformation detection and monitoring techniques is to work with updated inventories, a fundamental yet often overlooked issue in most countries such as Great Britain. Despite the availability of a national landslide database, less than half of the landslides reported are mapped as polygons, and their state of activity is unknown. In this regard, in this work we updated the national landslide inventory by mapping new events or simply identifying their current condition of motions through the use of the data freely provided by the European Ground Motion Service (EGMS), which represents an unprecedented baseline for ground deformation applications at continental, national and local level with millimetre accuracy. The approach relies on a semi-automatic tool, recently developed at the Centre Tecnològic de Telecomunicacions de Catalunya to identify the Active Deformation Areas (ADAs). Following an initial analysis of the InSAR data, the tool allows the identification of unstable areas characterized by a minimum number of persistent scatterers with velocity values over a specific threshold. The results consist of two ADA maps corresponding to the two Sentinel-1 velocity components and, subsequently, the output can be combined with landcover and topographic maps. The study has been carried out by exploiting the horizontal and vertical velocity maps provided by the EGMS which has enabled a national-scale analysis. Subsequent steps involve the classification and temporal analysis of the identified ADAs, followed by the analysis of more relevant local case studies.