Failure Probability of Deposit Landslide Under Gradient Rainfall Based on Conditional Probability

Abstract

Predicting the deposit landslide’s failure probability is a critical parameter in the geotechnical process. The cohesion and the internal friction are usually selected as random parameters for the numerical constitutive model. The existing method can predict the failure probability function (FPF) using the Monte Carlo method (MCM). However, it struggles with the limitation of the current probability density function (PDF). In this study, a small interval of parameters is introduced to estimate the parameters. The FPF can be converted into the augmented failure probability (AFP) approximation and the conditional probability related to the parameter interval using the Bayes model. Once the criterion of the small parameter intervals required for estimating the measure is selected, a single MCM combined with the adaptive Kriging nested method and MCM (AK-MCM) is established to approximate AFP. The Dahua deposit landslide is chosen as a case study to evaluate the accuracy and efficiency of the proposed solution. The copula model is used to correlate the rainfall intensity and the duration. The results reveal that the proposed solution is more accurate than the direct MCM and show that the return period could influence the failure probability.