A physically-based distributed cell model for predicting regional rainfall-induced shallow slope failures: Fid-Bau Portal

A physically-based distributed cell model for predicting regional rainfall-induced shallow slope failures

Chen, H.X. / Zhang, L.M.

Abstract Rainfall-induced slope failures are one of the most frequent hazards on hilly terrains. This paper proposes a physically-based distributed cell model to predict regional rainfall-induced shallow slope failures in two-layer soils under realistic rainfall conditions. The model consists of four components; namely, a digital terrain model, a spatial rainfall distribution model, an infiltration analysis model, and a slope stability and reliability evaluation model. The digital terrain is discretized into a grid of numerous cells first, with the properties of the soils in each cell assigned. Universal kriging is then adopted to interpolate the spatial rainfall distribution. Afterwards, the infiltration analysis model is used to analyze the infiltration process in two-layer soils under realistic rainfall conditions. The slope stability and reliability evaluation model is finally adopted to assess the regional slope stability and reliability. The distributed cell model is applied to evaluate the spatial and temporal response of a 164.5km² area to rainfall near the epicenter of the 2008 Wenchuan earthquake zone. Comparison between the predicted and observed slope failures triggered by the 13 August 2010 storm shows that this model is capable of predicting the locations of rainfall-induced slope failures reasonably well. The model is intended for use as a module in a real-time warning system for rainfall-induced slope failures.

Highlights • We developed a cell model to predict regional rainfall-induced shallow slope failures. • The spatial rainfall and soil information are obtained through kriging analysis. • Water infiltration in two-layer soils under stepwise rainfall condition is considered. • The model is capable of evaluating the spatial and temporal response of a large area. • The model is intended for use in a real-time landslide warning system.