Characterizing 4D post-failure slope kinematics of the 2020 Aniangzhai landslide combining different remote sensing measurements

With the avalanche of satellite remote sensing sensors, significant efforts have been made to develop methods to integrate optical and SAR remote sensing efficiently to quantify the kinematics and lifecycle of landslides. In this study, we design a framework that integrates multi-sensor satellite remote sensing data to investigate post-failure kinematics of the 17 June 2020 Aniangzhai landslide in the Danba County of Southwest China. This ancient landslide was partially reactivated due to rapid river incision and toe erosion during a complex cascading event, which led to an evacuation and relocation of more than 20,000 people.

First, time series of Planet images are exploited using the sub-pixel offset tracking method to generate horizontal deformation. Then advanced Multi-temporal InSAR (MTI) techniques are utilized to analyze the line-of-sight (LOS) displacements for 16 months after the failure. Eventually, the dynamics of the post-failure mechanism are modeled by integrating optical and radar data using an exponential decay model with independent component analysis (ICA) and least squares methods. Besides, the performance of a newly designed corner reflector (CR), consisting of two sets of semi-circular metal plates with a radius of 30-40 cm, is evaluated using both TerraSAR-X (TSX) and Sentinel-1 SAR data.

Optical results show that the landslide underwent large deformation up to around 14.3 meters within 1.5 months after the failure, then the rate of deformation decreased slowly with time. InSAR analysis suggests that the LOS velocity reached a maximum of approximately 300 mm/year, indicating the active status of the ancient landslide body after failure. Using ICA decomposition, we extracted different features with various spatiotemporal patterns from the landslide body, which was then applied in data integration and 4D modeling of landslide kinematics. Our experiment using newly designed CRs indicates improvement in the background intensity in TSX images by around 30 dB, with signal-to-clutter ratio (SCR) exceeding 25 dB. The radar cross-section (RCS) of CRs in both TSX and S1 images remains relatively stable, ranging from 15-23 dB, making them suitable for CR-InSAR analysis.