Refined and dynamic susceptibility assessment of landslides using InSAR and machine learning models

Abstract

Landslide susceptibility assessment is crucial in predicting landslide occurrence and potential risks. However, traditional methods usually emphasize on larger regions of landsliding and rely on relatively static environmental conditions, which exposes the hysteresis of landslide susceptibility assessment in refined-scale and temporal dynamic changes. This study presents an improved landslide susceptibility assessment approach by integrating machine learning models based on random forest (RF), logical regression (LR), and gradient boosting decision tree (GBDT) with interferometric synthetic aperture radar (InSAR) technology and comparing them to their respective original models. The results demonstrated that the combined approach improves prediction accuracy and reduces the false negative and false positive errors. The LR-InSAR model showed the best performance in dynamic landslide susceptibility assessment at both regional and smaller scale, particularly when identifying areas of high and very high susceptibility. Modeling results were verified using data from field investigations including unmanned aerial vehicle (UAV) flights. This study is of great significance to accurately assess dynamic landslide susceptibility and to help reduce and prevent landslide risk.