Investigation and dynamic analysis of the catastrophic rockslide avalanche at Xinmo, Maoxian, after the Wenchuan Ms 8.0 earthquake

Abstract

At 5:38:58 a.m. on June 24, 2017, a catastrophic rockslide avalanche occurred at the Xinmo village, Diexi town, Maoxian county, Sichuan Province, and the area could have been deeply affected by many strong earthquakes in history. The landslide killed 83 people and destroyed the whole village, with the runout horizontal distance of about 2800 m and the elevation difference of 1200 m, which was equivalent to a Fahrboschung of about 23°, and the accumulative volume reached nearly 18 million m3. The field investigation and preliminary analysis of landslide seismic signals indicated that, after sliding from the ridge crest, the rock mass continuously impacted and accumulated on the upper part of the slope, which caused the substrate, i.e., old rockslide accumulation, to slide and disintegrate. It was transformed rapidly into a pipeline-type debris avalanche that then turned into a fan-like debris flow due to the terrain, which also had the typical dynamic characteristics of high speed and long runout. However, in fact, as the cause and effect of the kinematic and dynamic processes of the Xinmo landslide is so complicated and difficult to understand, the Fast Lagrangian Analysis of Continua (FLAC) and the slope-stability analysis module of GeoStudio were firstly combined to analyze the failure process of the old rockslide accumulation body when suffering the added forces from the landslide. Then, the Engineering Discrete Element Method (EDEM) was used to simulate the runout behavior of the displaced landslide materials, the calculated velocity value of which was also compared with those using other dynamic modeling approaches (e.g., sled model, rheological model, and liquefied model). The results demonstrate that the dynamic characteristics and the kinetic type of the Xinmo landslide were successfully simulated. Besides, the velocity values simulated by EDEM and by the rheological model were close to the value of around 60 m/s based on the distance of debris avalanche and the corresponding duration recognized through landslide seismic signals. Also, their velocity variation curves with motion were rational. The above proposed method can be further applied for studies on early recognition and hazard zonation of similar large-scale rockslide avalanches in areas affected by strong earthquakes.