Spatiotemporal landslide susceptibility mapping using machine learning models: A case study from district Hattian Bala, NW Himalaya, Pakistan

The Himalayan region, a rugged mountain zone is among the most susceptible zones to landslide hazard due its terrain, geography, and active tectonics. Machine learning (ML) techniques are advanced precise methods develop susceptibility model (LSM). current study was designed analyze assess using ML approaches for District Hattian Bala, NW Himalayas, Pakistan. historical satellite imageries used generate spatiotemporal inventories of year 2005, 2007 2012. A spatial database created pertaining topographic, environmental, geologic, anthropogenic factors including slope, aspect, elevation, curvature, plane profile topographic wetness index (TWI), lithology, distance faults, streams, roads, normalized difference vegetation (NDVI) land use/ cover (LULC). These LCFs were selected periodic in region. experimental design utilized 349, 393, 735 inventory 2007, 2012 respectively. Two models, i.e., Random Forest (RF) Logistic Regression (LR) applied determine by thirteen causative (LCFs). partitioned into training (70%) testing (30%) landslides respective years check prediction accuracies models. Comparative analysis different LSMs performed Receiver Operator Curves – Area Under (ROC-AUC). resultant accuracy, MAE, RMSE, Kappa, Precision, Recall, F1 indicated that RF outperformed LR model. aims minimize losses lives potential economic damage linked with recurrent slope instabilities It anticipated use algorithms would support concerned authorities organizations effectively plan manage