Hazard zoning of individual landslide-debris avalanches considering complex 3D spatial variability by random material point method

Landslide-debris avalanche poses a serious threat to adjacent infrastructure and human safety. Due to their complex geological structures and depositional characteristics, natural slope soils often exhibit spatial variability, rotated anisotropy, and nonstationarity, making the prediction of slope post-failure behavior particularly challenging. These soil properties significantly impact slope stability and landslide movement. However, the influences of three-dimensional (3D) stratification and nonstationarity on consequences and hazard zones of the individual landslide remain underappreciated and relatively unexplored. This study proposes graphics processing unit (GPU) accelerated random generalized interpolated material point method for hazard zone assessment of the individual landslide that incorporates the 3D autocorrelation structural rotation, discretization of cross-correlated non-Gaussian random fields, modeling of nonstationary random field, landslide movement modeling, and estimation of exceedance probability surface. This innovative method is designed to effectively evaluate post-failure behaviors of slopes and hazard zones of the individual landslide under complex geological conditions. The findings demonstrate that considering the 3D spatial variability of soil is crucial for accurate individual landslide hazard assessments, as neglecting this variability leads to underestimation of potential hazard zones. This study establishes an improved framework for the hazard assessment of individual landslide and enhances our understanding of post-failure behavior of heterogeneous slopes.