Automated classification of A-DInSAR-based ground deformation by using random forest

Wide-area ground motion monitoring is nowadays achievable via advanced Differential Interferometry SAR (A-DInSAR) techniques which benefit from the availability of large sets Copernicus Sentinel-1 images. However, it primary importance to implement automated solutions aimed at performing integrated analysis amounts interferometric data. To effectively detect high-displacement areas and classify sources, here we explore feasibility a machine learning-based approach. This achieved by applying random forest (RF) technique large-scale deformation maps spanning 2015–2018. Focusing on northern part Italy, train model identify landslide, subsidence, mining-related with construct balanced training dataset. The presence noisy signals other sources also tackled within construction. proposed approach relies use explanatory variables extracted A-DInSAR datasets freely accessible informative layers such as Digital Elevation Model (DEM), land cover maps, geohazard inventories. In general, performance very promising an overall accuracy 0.97, true positive rate 0.94 F1-Score 0.93. obtained outcomes demonstrate that transferable may constitute asset for stakeholders in framework geohazards risk management.