Evaluation of factors controlling large earthquake-induced landslides by the Wenchuan earthquake
During the 12 May 2008, Wenchuan earthquake in China, more than 15 000 landslides were triggered by the earthquake. Among these landslides, there were 112 large landslides generated with a plane area greater than 50 000 m 2. These large landslides were markedly distributed closely along the surface rupture zone in a narrow belt and were mainly located on the hanging wall side. More than 85% of the large landslides are presented within the range of 10 km from the rupture. Statistical analysis shows that more than 50% of large landslides occurred in the hard rock and second-hard rock, like migmatized metamorphic rock and carbonate rock, which crop out in the south part of the damaged area with higher elevation and steeper landform in comparison with the northeast part of the damaged area. All large landslides occurred in the region with seismic intensity ≥ X except a few of landslides in the Qingchuan region with seismic intensity IX. Spatially, the large landslides can be centred into four segments, namely the Yingxiu, the Gaochuan, the Beichuan and the Qingchuan segments, from southwest to northeast along the surface rupture. This is in good accordance with coseismic displacements. With the change of fault type from reverse-dominated slip to dextral slip from southwest to northeast, the largest distance between the triggered large landslides and the rupture decreases from 15 km to 5 km. The critical acceleration ac for four typical large landslides in these four different segments were estimated by the Newmark model in this paper. Our results demonstrate that, given the same strength values and slope angles, the characteristics of slope mass are important for slope stability and deeper landslides are less stable than shallower landslides. Comprehensive analysis reveals that the large catastrophic landslides could be specifically tied to a particular geological setting where fault type and geometry change abruptly. This feature may dominate the occurrence of large landslides. The results will be useful for improving reliable assessments of earthquake-induced landslide susceptibility, especially for large landslides which may result in serious damages.