LANDSLIDE SUSCEPTIBILITY MAPPING OF SHIMLA DISTRICT USING INFORMATION VALUE AND WEIGHT OF EVIDENCE MODELS

Abstract

Himalayan region is a major natural hazard prone area where landslides cause severe damage to infrastructure, property and lives. The steep topography, complex geological environment, high anthropogenic activities and fragile mountain ecosystem make Shimla district very prone to landslides. In the present study, the landslide susceptibility over Shimla district has been assessed by employing two statistical and machine-learning models namely Weight of Evidence (WoE) model and Information Value (IV) algorithm in a Geographic Information System (GIS) environment. A thorough inventory of landslides was compiled using historical records and interpretation of satellite images. To minimize sampling bias, an equal number of non landslide points were also generated from areas of stability. To account for the processes of landslides in a mountainous area, 12 landslide conditioning factors were selected: slope, aspect, curvature, elevation, hillshade, roughness, lithology, topographic wetness index, drainage density, distance to stream, normalized difference vegetation index, and contour derived information. The Weight of Evidence model was used to assess the statistical association of the occurrence of landslides with each of the conditioning factors, and the Information Value model was used to determine the complex non-linear relationships among the factors. Two kinds of landslide susceptibility maps were produced with both methods and then divided into low, moderate, and high susceptibility zones. The results reveal that the most influential factors controlling the occurrence of landslides in Shimla are slope, roughness, lithology and elevation. The results of this study are useful for landslide risk reduction, land use planning and disaster management in Shimla district and also showcases the effectiveness of combining the statistical and machine learning methods for landslide susceptibility mapping in high relief Himalayan areas.