A rainfall infiltration-redistribution model for heterogeneous slopes considering multi-parameter uncertainties

Intense rainfall infiltration is one of the primary triggers for landslides. Developing a robust model for rainfall infiltration analysis is crucial for mitigating landslide disasters. Although the numerical solution of Richard’s equation provides high computational accuracy, it often encounters convergence issues. In contrast, the Green-Ampt (GA) model, which is more computationally efficient, lacks accuracy in dealing with the non-uniform distribution of the initial volumetric water content (VWC) and the pore-water redistribution process. Therefore, this study proposes a novel model for analyzing the slope rainfall infiltration process based on the GA model. The proposed model discretizes both the geological layers of the slope and the rainfall event in spatial and temporal scales, respectively, improving accuracy by adjusting step sizes of discretization adaptively. The proposed model is applied to analyze the permeability, stability and reliability of heterogeneous infinite slopes considering uncertainties in multiple parameters. Comparative studies with the numerical solution of Richard’s equation and other models demonstrate that the proposed model can provide high computational accuracy and superior analysis convergence in rainfall infiltration modeling. It also indicates that neglecting the pore-water redistribution underestimates the probability of slope failure, overestimates the factor of safety (FOS) of the slope, and inaccurately estimates the depth of the critical slip surface. Moreover, the uncertainties in shear strength parameters may overshadow the influence of initial VWC uncertainties on the slope reliability. This study provides a theoretical basis for the analysis of rainfall infiltration on heterogeneous slopes and the formulation of strategies for landslide disaster prevention.