ABSTRACT
The East Khasi Hills district of Meghalaya is widely renowned for its extremely high rainfall and complex terrain. This study applies a GIS-based Logistic Regression (LR) model to map landslide susceptibility using 13 geo-environmental and anthropogenic factors, including slope, land use, lithology, drainage proximity, and rainfall. Satellite imagery and historical data were used to create the landslide inventory. The relevant conditioning factors were integrated into the Logistic Regression model within a GIS environment. The ROC curve technique was utilized to assess model performance, resulting in an AUC of 0.9. The susceptibility map classified 86.28%, 5.27%, 3.1% and 5.35% of the district as very low, low, moderate and highly susceptible to landslides. The high-risk zones were mostly located on steep slopes nearby roads. The study underscores the effectiveness of GIS-based logistic regression for assessing landslide risk in complex terrains thereby aiding disaster mitigation.
