GIS-Based landslide susceptibility mapping using weight of evidence (WoE) and random forest (RF)

Abstract

In this study, the Landslide Susceptibility Map of Tokat province was produced. Slope classes, elevation classes, land use classes, geology classes, aspect classes and proximity to fault lines classes were used during the study. The Weight of Evidence method was applied to determine the relationship between the classes of the parameters and the landslide events. Random Forest method was used to determine the weights between parameters. Weighted Overlay operation was applied to the classified and weighted map data using ArcGIS program. As a result of the process, the data were divided into 5 classes and the Landslide Susceptibility Map was produced. When susceptibility classes are examined, it was seen that 92,42% of the old landslide events occurred in high and very high classes.