A Physically Based Model for the Sequential Evolution Analysis of Rainfall‐Induced Shallow Landslides in a Catchment

Abstract

AbstractPredicting the occurrence time, volume, distribution and run‐out length of shallow landslides is critical for assessing the volume of debris flow in a catchment. Recent studies suggest that the landslides hydro‐mechanical triggering (LHT) model based on threshold‐based mechanical interactions can adequately predict rainfall‐induced landslides. However, this model assumes that the soil‐rock interface is the only sliding surface and cannot dynamically determine a sliding surface above the soil‐rock interface. Therefore, by calculating the wetting front depth in the shallow soil layer and using the limit analysis method, the most likely sliding surface can be dynamically calculated to improve the model. Through linking the Soil and Water Assessment Tool (SWAT, USDA) model and the depth‐resolved LHT model (D‐LHT), a framework for predicting shallow landslides in a catchment was proposed in this work. This framework considers the effects of antecedent rainfall, the mechanical reinforcement of roots, and the spatial distribution of soil properties on slopes. The D‐LHT model was applied to the Jiangjia gully in Yunnan, China, to adequately predicted the occurrence time, scale and spatial distribution of shallow landslides. In addition, the present and antecedent rainfall effects on the shallow landslide volume were analyzed. The results showed that the average soil moisture content threshold was determined by antecedent and same‐day rainfall, which affect the timing and volume of the shallow landslides. This study provides a new method for predicting shallow landslides. The stability and simplicity of the model make it suitable for early warning systems.