The size, distribution, and mobility of landslides caused by the 2015 Mw7.8 Gorkha earthquake, Nepal

Abstract

Coseismic landslides pose immediate and prolonged hazards to mountainous communities, and provide a rare opportunity to study the effect of large earthquakes on erosion and sediment budgets. By mapping landslides using high-resolution satellite imagery, we find that the 25 April 2015 Mw7.8 Gorkha earthquake and aftershock sequence produced at least 25,000 landslides throughout the steep Himalayan Mountains in central Nepal. Despite early reports claiming lower than expected landslide activity, our results show that the total number, area, and volume of landslides associated with the Gorkha event are consistent with expectations, when compared to prior landslide-triggering earthquakes around the world. The extent of landsliding mimics the extent of fault rupture along the east-west trace of the Main Himalayan Thrust and increases eastward following the progression of rupture. In this event, maximum modeled Peak Ground Acceleration (PGA) and the steepest topographic slopes of the High Himalaya are not spatially coincident, so it is not surprising that landslide density correlates neither with PGA nor steepest slopes on their own. Instead, we find that the highest landslide density is located at the confluence of steep slopes, high mean annual precipitation, and proximity to the deepest part of the fault rupture from which 0.5–2Hz seismic energy originated. We suggest that landslide density was determined by a combination of earthquake source characteristics, slope distributions, and the influence of precipitation on rock strength via weathering and changes in vegetation cover. Determining the relative contribution of each factor will require further modeling and better constrained seismic parameters, both of which are likely to be developed in the coming few years as post-event studies evolve. Landslide mobility, in terms of the ratio of runout distance to fall height, is comparable to small volume landslides in other settings, and landslide volume-runout scaling is consistent with compilations of data on larger slope failures. In general, the size ratios of landslide source area to full landslide area are smaller than global averages, and hillslope length seems to largely control runout distance, which we propose reflects a topographic control on landslide mobility in this setting. We find that landslide size dictates runout distance and that more than half of the landslide debris was deposited in direct connection with stream channels. Connectivity, which is defined as the spatial proximity of landslides to fluvial channels, is greatest for larger landslides in the high-relief part of the High Himalaya. Although these failures are less abundant than those at lower elevations, they may have a disproportionate impact on sediment dynamics and cascading hazards, such as landslide reactivation by monsoon rainfall and landslide dams that lead to outburst floods. The overall high fluvial connectivity of coseismic landsliding in the Gorkha event suggests coupling between the earthquake cycle and sediment/geochemical budgets of fluvial systems in the Himalaya.