Landslide susceptibility, social vulnerability, and risk assessment in Kumaun Himalaya, Uttarakhand, India

Landslide susceptibility and vulnerability assessment are essential for landslide risk reduction in hilly landscapes like Indian Himalayas. The objective of this research is two-fold. The first is to develop a landslide susceptibility zonation (LSZ) map of the Pithoragarh region, a seismically active part of Kumaun Himalaya in Uttarakhand, India. Geographic Information System (GIS) tools were used to integrate ten landslide conditioning parameters: slope angle, slope aspect, slope curvature, distance to drainage, geology, normalized difference vegetation index (NDVI), distance to tectonic, distance to road, rainfall, and seismic parameter. A total of 174 landslide events are reported through the landslide inventory map. To analyze landslide susceptibility, an integrated approach has been adopted. The analytical hierarchy process (AHP) and relative frequency ratio (RFR) models were applied to calculate factor weights (Wfi) and factor class weights (FCWi,j), respectively. The efficiency and performance of the susceptibility map were evaluated through training (70% of landslide inventory) and testing (30% of landslide inventory) dataset, respectively. The second objective is to evaluate social vulnerability and risk for the localized selected area, a part of the region considered for LSZ. The social vulnerability was assessed by combining the household, population, literacy rate, and working-class population (economy level). An attempt has been made to integrate the landslide susceptibility and social vulnerability to investigate the area covered by various risk classes. The relevance of this advanced approach is demonstrated in a case study of the Pithoragarh region along a National Highway (NH) number 9. As a result, the LSZ map was categorized into five classes, i.e., “Very Low (12%),” “Low (24%),” “Moderate (30%),” “High (21%),” and “Very High (13%).” Further, the validation of these results by the area under the curve (AUC) method and root mean square error (RMSE) has been carried out. The success rate and prediction rate of LSZ map are 0.77 and 0.76, respectively, whereas RMSE is 0.14, indicating the good accuracy of the model in the identification of landslide susceptibility zones. Engineers, planners, and decision-makers can use the final maps for the occupants’ safety and update the current environmental plans for sustainable development.