Comparing the Validity of Statistical and Knowledge-Based Methods for Landslide Susceptibility Mapping

In the Shahroud Watershed, there has been an increasing occurrence of landslides that have caused a lot of human and financial losses. Therefore, landslide susceptibility zonation is crucial for reducing landslide risk. The aim of this study was to compare the Landslide Susceptibility Maps (LSMs) of different methods. Therefore, thematic layers of the ten causal factors were prepared. Then, a landslide inventory map consisting of 104 landslides covering 1401 hectares was compiled and partitioned into two subsets including 70% for training and 30% for testing purposes. Three landslide susceptibility maps were prepared using the Frequency ratio (Fr), Statistical index (Si), and Analytic Hierarchy Process (AHP) methods. The validation process showed that the Si [Area Under the Curve (AUC)= 0.732] and Fr (AUC= 0.707) models presented a more valid LSM than AHP (AUC= 0.651) method. The Qs (Quality sum) index values also confirmed the results of the ROC (Receiver Operating Characteristic) curve such that the Qs index values of 1.71, 1.43, and 0.62 for, respectively, Fr, Si, and AHP models implied a more accurate LSMs of the Fr and Si models than the one from the AHP. The results of this study can be used as a basic step for landslide risk management in the study area.