Probabilistic simulation of entire process of rainfall-induced landslides using random finite element and material point methods with hydro-mechanical coupling

Rainfall-induced landslides are a major geohazard worldwide, which involve a complicated process, starting from rainfall infiltration, to landslide triggering and initiation, and post-failure large deformation of soils. To effectively mitigate landslide hazards, quantitative risk assessment (QRA) of landslides may be performed, which includes assessment of landslide probability and consequences. Because rainfall-induced landslide probability is controlled by rainfall infiltration and landslide triggering, while landslide consequences are often dictated by post-failure large deformation of soils, it is necessary to model the entire process of a landslide in QRA. However, most previous QRA studies did not model the entire landslide process (e.g., without modeling of rainfall infiltration or post-failure large deformation of soils) due to the difficulty of slope stability analysis methods (e.g., limit equilibrium methods and finite element method (FEM)) in modeling large deformation of soils. This study proposes a probabilistic method for simulating the entire process of rainfall-induced landslides considering spatial variability of soil properties. A two-stage method containing FEM and material point method (MPM), both with hydro-mechanical coupling, is proposed to simulate a landslide from triggering by rainfall to post-failure large deformation of soils. It quantifies landslide probability and consequences simultaneously and provides new insights for QRA of rainfall-induced landslides.