Numerical modeling of the Xinmo landslide from progressive movement to sudden failure

In the early morning of 24 June 2017, a slow-moving landslide suddenly accelerated and impacted Xinmo village after a sustained period of low-intensity rainfall. The landslide caused more than 80 casualties and damaged approximately 100 houses. Although many studies have been performed to understand the failure mechanism, the entire failure process of the Xinmo landslide is still not clear. In this study, a depth-integrated particle method coupled with a soil–water mixing model is used to back-analyse both slowly progressive movement and sudden failure processes of the Xinmo event. A representative volume element of the sliding zone is modeled, with consideration of erosion that progressively changes the solid concentration of the sliding mud. Numerical results show that the continuous erosion of sliding mud by rainwater accelerates the displacement rate of the landslide body. During the slow-moving stage, the simulated displacement rate is approximately 0.6 mm/year. On the other hand, the rapid failure process lasts 105 s from sudden failure to final deposition, with a maximum velocity of 58.6 m/s. As evidenced by the analysis of the seismic signals, the depth-integrated particle method has a good performance in simulating the rapid failure process of a landslide. The results demonstrate that the erosion of sliding mud resulted from the rainwater may play a critical role in the Xinmo landslide. This study contributes to understanding failure processes of rainfall-induced slow-moving landslides and provides guidelines on hazard prevention and mitigation.