Application of GIS-based logistic regression for landslide susceptibility mapping in Gajapati district, Odisha: A case study

Landslides are gravity-induced mass movements of rock and soil, posing persistent dangers in Odisha's Gajapati district, especially in densely inhabited and infrastructure-sensitive regions. The present study employs a data-driven approach to generate a Landslide Susceptibility Map (LSM) for Gajapati district using a logistic regression (LR) modeling framework. A suite of influential terrain and environmental predictors such as slope gradient, aspect, profile curvature, land cover classification, distance from hydrological and transportation networks, and the Topographic Wetness Index (TWI) were selected based on their geomorphological relevance and statistical correlation with landslide occurrence. The logistic regression model, integrated with bivariate statistical analysis, facilitates the quantification of landslide probability while maintaining model interpretability and minimizing multicollinearity effects. The logistic regression approach analyzes landslide likelihood combining both continuous and categorical variables to generate a probabilistic susceptibility map. The results indicate that approximately 2.31% of the district falls within the high susceptibility class, 1.67% within the moderate class, and 92.7% of the terrain is categorized as having low susceptibility to landslides.