Slow-moving landslide risk assessment combining Machine Learning and InSAR techniques

This paper describes a novel methodology where Machine Learning Algorithms (MLAs) have been integrated to assess the landslide risk for slow moving mass movements, processes whose intermittent activity makes challenging any analysis worldwide. MLAs has trained on datasets including Interferometric Synthetic Aperture Radar (InSAR) and additional remote sensing such as aerial stereo photographs LiDAR tested in Termini-Nerano landslides system (southern Apennines, Italy). The availability of wealth materials allows also an unprecedented spatio-temporal reconstruction volume kinematic through which we could generate validate hazard map. Our identifies fifteen slow-moving phenomena, traceable since 1955, total area amounts 4.1 × 105 m2 ~1.4 106 m3. InSAR results prove that seven out are currently active characterized by seasonal velocity patterns. These new insights dynamic selected main independent variables train three (Artificial Neural Network, Generalized Boosting Model Maximum Entropy) derive area. Finally, official population buildings census data used highest values located crown area, south Termini village, nearby Nerano. provides different scenario compared existing documents study overall how develop management strategies worldwide based better understanding slope thanks latest satellite technologies available.