Landslide Susceptibility Mapping using Machine Learning Algorithm

Landslides are natural disasters that have resulted in the loss of economies and lives over years. The landslides caused by 2005 Muzaffarabad earthquake heavily impacted area, slopes region become unstable. This research was carried out to find which areas, as district, sensitive define relationship between geo-environmental factors using three tree-based classifiers, namely, Extreme Gradient Boosting (XGBoost), Random Forest (RF), k-Nearest Neighbors (KNN). These machine learning models innovative can assess environmental problems hazards for any given area on a regional scale. consists steps: Firstly, training validation, 94 historical were randomly split into proportion 7/3. Secondly, topographical geological data well satellite imagery gathered, analyzed, built spatial database GIS Environment. Nine layers landslide-conditioning developed, including Aspect, Elevation, Slope, NDVI, Curvature, SPI, TWI, Lithology, Landcover. Finally, receiver operating characteristic (ROC) curve under ROC (AUC) value used estimate model's efficiency. values RF, XGBoost, KNN 0.895 (89.5%), 0.893 (89.3%), 0.790 (79.0%), respectively. Based techniques, outputs show performance model has maximum AUC 0.895, it is more efficient than other classifiers. Elevation Slope determined most important affecting this area. landslide susceptibility maps classified four classes: low, moderate, high, very high susceptibility. result useful future generalized construction operations, such selecting conserving new urban infrastructural areas. Doi: 10.28991/CEJ-2022-08-02-02 Full Text: PDF