Application of bivariate statistical techniques for landslide susceptibility mapping: A case study in Kaghan Valley, NW Pakistan

Landslides are dangerous events that threaten both human life and property. This research presents a case study of Kaghan valley catchment, which is an area frequent landslide activity. We aim to present comparison between the bivariate Landslide Numerical Risk Factor (LNRF), Statistical Index (SI) Information Value (InfV) statistical models evaluate susceptibility. A total 1556 landslides were identified using earlier reports, field surveys, GOOGLE Earth imagery. The abundance primarily controlled by acute deformation caused major thrust fault system proximity Hazara Kashmir Syntaxis (HKS). inventory was randomly partitioned into two datasets. 70% (1106) used as training phase models, whereas 30% (450) validation three models. spatial database 11 conditioning factors produced consisting slope, aspect, elevation, lithology, land use, Topographic Wetness (TWI), rainfall, Stream Power (SPI), distance faults, rivers streams. All susceptibility assessment parameters obtained from different sources maps prepared on GIS software. Performance validated through Receiver operator Characteristics (ROC) success prediction rate curves. Results show under ROC curve (AUC) for InfV, LNRF SI 70.95, 83.99 67.56 curves 70.75, 67.85 curves, respectively. having highest AUC value proved be superior generating regional scale maps.