Earthquake-triggered landslides susceptibility assessment by critical acceleration: grid cells vs. slope units vs. ground-motion

Earthquake ground motion is the primary driver of earthquake-triggered landslides; however, their predictability is poor. We use Newmark's critical acceleration method to map landslides triggered by the 2015 Gorkha earthquake in central Nepal. In this method, we compare the critical acceleration of a block (landslide pixel or slope unit) with the peak ground acceleration (PGA) to assess landslide susceptibility. We analyse earthquake-triggered landslide occurrence on each grid cell and each slope unit using two different sources of peak ground acceleration (from USGS ShakeMap and KTP seismic station). A total of four susceptibility maps were developed, which were compared with the existing landslide inventory of the 2015 Gorkha earthquake and validated using the Matthews correlation coefficient. Results from this study showed that the slope unit-based mapping using ShakeMap's PGA yields the best results. The landslide susceptibility map generated using this technique resembles well with the landslide inventory map and can be used for prioritizing rapid earthquake response, detailed investigations and planning.