Prediction analysis of landslide displacement trajectory based on the gradient descent method with multisource remote sensing observations

AbstractLandslides can result in extensive casualties and huge economic losses. The accurate discrimination of the main slip direction and deformation trajectory is an important prerequisite for studying landslide formation mechanisms and designing landslide control schemes. In the process of landslide evolution over time, the main slip direction also changes dynamically and provides a comprehensive reflection of the landslide displacement state. However, few studies on this topic have been published. In this paper, a new methodology for analyzing slope stability is proposed based on three techniques: interferometric synthetic aperture radar (InSAR), unmanned aerial vehicle (UAV), and ground-based interferometric synthetic aperture radar (GB-InSAR). The Small Baseline Subset Interferometric SAR (SBAS-InSAR) technique is combined with an overall analysis of the study area to identify the regions of interest (ROIs) with large deformation and the starting target points, and the fusion results of radar deformation data (RDD) and digital surface model (DSM) data are used to fit the deformation surface field of the ROIs. The gradient descent approach is executed to obtain the running trajectory points of the target masses so that the main slip direction and displacement trajectory in the study area can be predicted at small scales. The measured data for the Hongshiyan landslide in Yunnan Province are used to verify the effectiveness of the method, and the predicted results are consistent with the actual landslide direction. The experimental results show that the method can exactly identify the deformation area, especially in the case of a fast-changing deformation trend. This approach can provide more accurate monitoring area results to support the rapid control and prevention of landslide hazards by analyzing the minimum pixel grid (i.e. points), as the smallest spatial unit at a time interval of minutes. The study shows that the method can efficiently combine space–Earth multisource monitoring data to clarify the main slide direction and improve the postslide trajectory prediction of the slope, which is beneficial for assessing disaster risks and improving landslide prevention and control effects, reflecting the engineering application value of the approach.