Landslide Detection in Long-term and Lowcoherence Scenario Using Faster Intermittent Stacking InSAR Method

Time-series Interferometric Synthetic Aperture Radar (InSAR) technology is a powerful tool for monitoring surface deformation and has extensive applications in landslide monitoring. River gorge areas are typically characterized by steep terrain and dense vegetation, resulting in low coherence for shortwavelength SAR signals, such as those from Sentinel-1. This often leads to the minimal capability of conventional time-series InSAR methods, which poses challenges for effective ground deformation detection, particularly in low-coherence areas over long-term scenarios. In this study, we present an optimized time-series InSAR method called intermittent Stacking (IStacking), which allows rapid calculation of high-coverage deformation and detection of landslides in long-term and low-coherence scenarios. Using the Three Gorges Reservoir Area (TGRA) as a case study for deformation monitoring and landslide identification, we demonstrated the principles and application of the IStacking method. Analyzing 152 ascending Sentinel-1 SAR images from January 2018 to June 2023, we determined that conventional timeseries InSAR methods detected deformation in only 4% of the study area, the intermittent small baseline subset (ISBAS) method achieved a 70% coverage ratio, and the proposed IStacking method reached a coverage ratio of 96%. The identified landslides with the IStacking method were validated and assessed using highresolution Google imagery and ground-based GPS data. Because of its ability to fully exploit the intermittent coherence properties of pixels and its simple and efficient implementation, the IStacking method can exploit the potential of short-wavelength SAR data in low-coherence regions and provide a powerful tool for wide-area InSAR deformation monitoring.