An enhanced neighborhood differential method for potential landslide identification from stacking-InSAR results

Abstract

Stacking-InSAR is widely used for identifying potential landslide. However, in Stacking-InSAR results, there are often challenges to clearly identify the potential landslide due to wide-scale system errors, especially the elevation-related atmospheric disturbances, leading to low precision of potential landslide identification. This paper introduces an enhanced neighborhood differential method that employs a concept of step size to identify potential landslide from Stacking-InSAR. Applied to the mountainous regions in the northwestern part of Sichuan Province, China, this method successfully identified 117 potential landslides, verified through field investigations. Compared to traditional Stacking-InSAR, this method additionally identified 34 potential landslides. Further analysis indicates that the optimal step size in the neighborhood differential method should ensure that one differential pixel is located inside the potential landslide area while the other is outside, thereby making the step size closely related to the spatial extent of the potential landslide. The results indicate that this method could effectively reduce wide-scale system error in Stacking-InSAR results, providing an important technical reference for the rapid and efficient identification of potential landslide over a wide mountainous area in the future.