Probabilistic evaluation of landslide influence zones considering stratigraphic dips and nonstationarity of soil properties

Accurate prediction of landslide movement is essential for effective disaster prevention and control. However, current studies on probabilistic large deformation analysis of landslides assume transverse anisotropy of soil, overlooking the impact of the soil fabric and depositional orientation on the post-failure behavior. While the specific effects of stratigraphic dips and nonstationary soil orientations on slope stability are frequently analyzed, these effects on the post-failure behavior of slopes have not been thoroughly explored. This study proposes a new probabilistic framework for simulating landslides and quantifying hazard zones, incorporating complex stratigraphic dips and two typical nonstationary soil orientations. The new method integrates nonstationary random field (RF) theory with the rotation of spatial autocorrelation structure. It derives formulas for calculating the thickness and depth of the soil layer at various locations, considering different stratigraphic dips and nonstationary orientations. This approach enables the simulation of parameter distributions for bedding and inverse soils with both vertical and stratigraphic nonstationarity. The generalized interpolation material point method (GIMP) is then used to simulate the post-failure behavior of slopes. The findings indicate that neglecting the spatial variability of soil parameters leads to an underestimation of the influence zone of landslide. Additionally, the nonstationary characteristics of soil parameters and stratigraphic dips can affect the failure mechanisms of slopes and the exceedance probabilities of runout and influence distances. The proposed method enhances the accuracy of predicting runout and influence distances, serving as a novel valuable tool for disaster management and mitigation.