Assessing Landslide Risk Probability in the Garhwal Himalayas, India Using a GIS-Based Bivariate Statistical Approach

Landslides is deadliest disasters which occur frequently without warning causing damages and human causalities in the vulnerable areas. The topography of the region affects the frequency of landslides occurrences, as well as the impact of outside factors including intense rain, seismic activity, changes in groundwater levels, snowmelt, stream erosion, flooding, or any combination of these natural events. The research study investigates the risk probability of Garhwal Himalaya with the help of several causative factors, including slope, aspect, curvature, elevation, proximity to river, proximity to road, rainfall, lineament density, NDVI, NDBI and census data of 2011. Landslide inventory was prepared and classified into training data (70%) and testing data (30%). The landslide risk probability and susceptibility for the area of interest have been obtained using the frequency ratio (FR) approach. The resultant susceptibility and risk probability maps were classified into five i.e very low, low, medium, high, very high. The study reveals that 15.8% of the areas fall under the very high susceptibility zone, while 17.3% area in the very high risk zone. Further, the receiver operating characteristic-area under the curve (ROC-AUC) was used to calculate the landslide risk probability map's overall model accuracy, that turned up to 75.25%. The findings can be used further by planners and relevant authorities for landslip mitigation and control.