Landslide susceptibility assessment using different slope units based on the evidential belief function model

Selecting an appropriate mapping unit is an important step for landslide susceptibility evaluation. This study compares the results of applying different slope units for landslide susceptibility maps. Landslide susceptibility maps based on the evidential belief function model were obtained using two different slope units: hydrological analysis-based and curvature watersheds-based slope units. An inventory map with 249 landslide locations was constructed from multiple sources. Landslide locations were randomly split into two datasets: 70% for building the models and 30% for validation. Nine landslide-influencing factors were prepared for landslide susceptibility mapping in the study area. Area under the curve (AUC) was used to analyse the landslide susceptibility maps by calculating the success rate and prediction rate. The verification results indicated that the AUC for the curvature watersheds and hydrological analysis models were 0.8038 and 0.7885 with prediction accuracy 0.8056 and 0.7798, respectively. The resultant maps will be useful for hazard mitigation purposes.