On Assessing the Probability of Rainfall-Induced Slope Failure During a Given Exposure Time
Meng Lu,Jianguo Zheng,Jie Zhang,Hongwei Huang
DOI: https://doi.org/10.1007/s11440-022-01655-w
2023-01-01
Acta Geotechnica
Abstract:Assessing the probability of rainfall-induced slope failure during a given exposure time is essential for the quantitative risk assessment of landslide hazards. In current mechanics-based methods, the focus is often on the annual failure probability, and the failure events of a slope under different rainfalls are assumed to be fully correlated. In statistical methods, it is often assumed that the failure events of a slope under rainfall in different years are statistically independent (i.e., the Poisson process assumption). In this paper, three mechanics-based methods are devised to assess the failure probability of a slope during an arbitrary exposure time, i.e., a method in which prior assumptions about the correlations among slope failure events under different rainfalls are not needed (Method 1), a method based on the Poisson process assumption (Method 2), and a method based on the assumption that the failure events of a slope during different rainfalls are fully correlated (Method 3). It is found that slope failure events in different years are not entirely statistically independent. As such, the method based on the Poisson assumption tends to overestimate the failure probability of the slope during an exposure time. Nevertheless, when the annual failure probability is small and the exposure time is short, the method based on the Poisson assumption is also quite accurate. For comparison, the failure probabilities estimated based on Method 3 are smaller than but close to those estimated based on Method 1. Among the three methods, Method 1 is theoretically most stringent but computationally most expensive. When the failure probability is large, Method 2 is computationally most efficient. When the failure probability is small, Method 3 is computationally most efficient.