Stochastic assessment of slope failure run-out triggered by earthquake ground motion

Abstract

Analysis of the run-out of landslides is essential and vital for disaster mitigation. However, accurate run-out analysis is difficult because of the uncertainty of earthquake ground motion and variability of soil properties. To solve this problem, a new run-out assessment framework that combines the methods of probability density evolution and smoothed particle hydrodynamics is proposed. This novel framework can consider multiple stochastic factors and different slope failure models of changing sliding surfaces. We used a homogeneous 2D slope as an example and generated stochastic seismic loading samples with an intensity-frequency non-stationary ground motion model. Soil property parameters (cohesion and internal friction angle) were assumed to obey logarithmic normal distribution, and run-out parameters were evolved. Moreover, based on an equivalent extreme event, the distributions of final run-out parameters were obtained. In an example with slope height of 100 m and angle of 45°, the probability that the run-out distance is < 150 m is 90%. The probability of flow depth more than 24.5 m was about 50%. The new framework has the potential to provide references for coordinating relief efforts after landslides by predicting the extent and scope of the earthquake-induced hazard; thereby minimizing casualties and property losses caused by geological disasters triggered by earthquakes.