Validating Modelled Post-Earthquake Landslide Runout Using Timeseries Landslide Inventories: Nepal 2015

The Nepal 2015 earthquakes triggered in excess of 20,000 landslides in 14 districts within Central and Western Nepal. As of mid-2019, a similar number of landslides remain active in this region, illustrating the legacy of the earthquake on slope stability. The affected districts are also the location of more than 1.1 million residential buildings, the majority of which are in rural mountainous locations. As a result, the hazard posed by the continued remobilisation of failed material and the ongoing failure of damaged hillslopes during the monsoons since 2015, is a significant risk. Whilst recent research has begun to explore the general evolution of post-earthquake landsliding, how this maps onto generating risk and how this risk changes through time is less well understood. Here we present the analysis of a timeseries landslide inventory collated prior to and since 2015 for the area impacted by the earthquake. We map landslides bi-annually from satellite imagery to monitor how landslides change, and to identify where new landslides occur. We then use these inventories as an input into a distributed empirical model (Flow-R, Horton et al, 2013) that routes debris down slope simulating the 'worst case' condition - the scenario in which landslides transition to debris flows. In the absence of detailed field data from across the region, uniquely here we use the time series landslide inventories, which document how each landslide has changed through time, to refine and then validate the results of the runout prediction from the model. This analysis identifies key patterns in the nature of those landslides and areas which continue to pose a hazard after the earthquake, and constrains the post-earthquake remobilisation of landslide debris.