Back-analysis of a rainfall-induced landslide case history using deterministic and random limit equilibrium methods

Abstract

Because recurrent rainfall-induced landslide episodes mimic prior failures, this study builds a back-analysis approach to identify those prior failure events and assess the prospective slope performance going forward. Engineers use the results of this approach to project the slope performance based on the quantity of rainfall accumulation during the anticipated rainfall event. A landslide case that occurred in 2004, as a result of a typhoon event, was selected for the back-analysis procedure. Four typhon events were chosen to simulate the pore water pressure conditions, which may have caused damage of varying severity (including landslide failure) to the slope. The numerical model was initially analyzed via the deterministic limit equilibrium slope stability approach to evaluate the most critical failure surface. By adopting the Monte Carlo simulation approach, the following step assessed the random limit equilibrium slope stability on the most critical failure surfaces. The probabilistic distributions of the safety factor were then generated for each typhoon event. Subsequently, the simulation outcomes provide safety indicator values incorporating the mean safety factor, failure probability, and reliability index. From the failure probability and reliability index, a zonation approach with various site-specific performance levels were established by using the corresponding accumulated rainfall to define the exceedance threshold values for each zone. In order to confirm the site-specific performance, we compared the normalized error values acquired from the analysis technique to those measured under two recent typhoon occurrences. An acceptable consistency was attained, validating the procedure's appropriateness for future rainfall events.