Analysis of landslide dams induced by the 2008 Wenchuan earthquake

Landslide dams caused by earthquakes are extremely hazardous disruptions of the flow of water and sediment in mountain rivers, capable of delivering large outburst floods that may devastate downstream areas. We analyzed a unique inventory of 828 landslide dams triggered by the . M _w 7.9 2008 Wenchuan tectonic earthquake, China, constituting ~1.4% of the >60,000 coseismic slope failures mapped and attributed to this event. While 501 landslides blocked the rivers completely, the remainder caused only partial damming or channel diversion. The spatial distribution of landslide dams follows the same trend of that of the total landslide distribution, with landslide dams being most abundant in the steep watersheds of the hanging wall of the Yingxiu-Beichuan Thrust Fault, and in the northeastern part of the strike-slip fault near Qingchuan. Besides the co-seismic landslide density, the river width also played a key role in determining the landslide dam formation. Narrow rivers are more prone to be dammed than the wide rivers. The correlation between river width and landslide dam volume follows a linear relation, which can be used to roughly estimate the dam formation possibility. However, the applicability of this correlation needs to be validated in other regions. The decay (failure) rate of dams, defined here as the percentage of the number and area of landslide dams that have failed over time, shows that ~25% of dams accounting for ~30% of total landslide dam area failed one week after the earthquake. These percentages increased to ~60% within 1. month, and to >90% within 1. year. The geomorphometric parameters were analyzed, revealing power-law relations between landslide area and dam width, landslide source area and dam area, as well as lake area and lake volume. The inventory presented in this study will enrich the worldwide earthquake-induced landslide dam database and will also contribute to a better understanding of the post-earthquake dam decay.