Abstract: The risk status of a slope under rainfall conditions is closely related to the spatial variability of soil parameters. In order to efficiently evaluate the risk level of rainfall-induced landslides, this paper uses the modified Green-Ampt infiltration model combined with the infinite slope model to establish the limit state function, and combines the first-order reliability method (FORM) with the random field theory to study the probabilistic stability subjected to uncertain infiltration parameters with spatial variability. These calculation procedures are systematically realized in the Python platform. Finally, we make a probabilistic analysis of rainfall-induced landslides and establish a fragility curve of the slope in the case area. The research results show that: compared with traditional stability analysis methods, probabilistic analysis methods can more comprehensively reflect the risk status of slopes under different rainfall intensities and rainfall durations. When the rainfall intensity exceeds a certain level, the rainfall has little effect on the slope failure probability under the same rainfall duration. In a specific study area, when the rainfall intensity is known, the fragility curves can be used to clarify the failure time of the slope under allowable risk level and can provide an important reference for the disaster prevention and mitigation work for relevant departments.

Key words: shallow landslide, rainfall infiltration, random field, FORM, fragility curve