Earthquake-driven acceleration of slow-moving landslides in the Colca valley, Peru, detected from Pléiades images

Abstract

Major earthquakes in mountainous areas often trigger rapid landslides. Some observations also suggest that earthquakes can damage landslide prone areas or cause slow-moving landslides to accelerate, with a risk of evolution to rapid landslides in the following months after the earthquake. Here, we use optical images from the Pléiades satellites to detect slow-moving landslides and quantify the effect of earthquakes on the landslide motion. We process multi-temporal Pléiades images acquired in March, April, and July 2013 over an area of 210km2 in the Colca valley (South Peru), to obtain two Digital Elevation Models (DEM) and three displacement fields of the area. The processed DEMs have an uncertainty of 0.6m (1σ), an order of magnitude better than two global and freely available DEMs (GDEM-v2 and SRTM), whereas the displacement fields have an uncertainty of between 0.11 and 0.18m (1σ) in both horizontal directions. Using these data, we detect 9 slow-moving landslides and compare their velocities during the March–April and April–July periods. We find that landslide velocities are highest during the wet season, which suggests a strong groundwater control, and we also highlight a landslide acceleration caused by a regional Mw 6.0 earthquake. The major parameters controlling the acceleration of the slow-moving landslides are the rock type and the distance to the source, suggesting that friction at the basal interface in the weeks after the earthquake is dependent on the shaking intensity.