Landslide Susceptibility Mapping in Nepal Using Spatial Feature Vectors

Section 1: Introduction According to the USGS (U.S. Geological Survey), the earthquake of April 2014 in Nepal and its aftershocks claimed the life of almost 9,000 people. Many of the deaths were caused by landslides which also blocked important routes for emergency response and evacuations. These kinds of disasters are not uncommon in other parts of the world and vastly underestimated as recent research shows. Landslide prediction can be a valuable tool for planning first response (Where to send first response helpers if there is no communication after an earthquake?) and emergency evacuations. Unlike the US, countries like Nepal do not have extensive infrastructure to measure and monitor seismic occurrences that could help predict and locate disasters. For this project we try to identify high risk zones prone to landslides based on features that have been studied by research institutions (vanWesten et al., 2008). We use expert labeled GPS locations, which occurred in Nepal as a consequence of the earthquake, as our training set. Figure 1 shows a zoomed-in image of some of Nepal’s landslides taken from Google Earth.