Abstract: Establishing a reasonable rainfall infiltration model is crucial for understanding the mechanisms of rainfall-induced slope failures and for preventing and controlling landslide disasters. Currently, numerical solutions to the Richards equation and the Green-Ampt model are commonly employed to analyze rainfall infiltration in slopes. However, numerical solutions to the Richards equation often encounter convergence issues, while the Green-Ampt model suffers from relatively low accuracy and an inability to account for non-uniform initial soil moisture distribution and pore water redistribution. In this study, we propose a spatiotemporal approximation method based on the Green-Ampt model for analyzing rainfall infiltration in slopes. This method involves discretizing the slope stratum into multiple spatial elements along the depth and dividing the rainfall event into multiple time intervals. Using this method, we investigate the changes in seepage, stability, and reliability of heterogeneous infinite slopes under rainfall infiltration. The results show that, compared to numerical solutions of the Richards equation, the proposed method is a robust and accurate approach for analyzing rainfall infiltration, with no convergence issues. Additionally, ignoring the pore water redistribution process can lead to underestimating the probability of slope failure and misidentifying the location of the critical sliding surface. The research outcome can provide theoretical support for rainfall infiltration analysis of complex slopes and prevention and control of rainfall-induced shallow landslide disasters.

Key words: slope reliability analysis, spatial variability, rainfall infiltration analysis, initial soil water content, pore water redistribution