Landslide Hazard in Sukabumi Regency based on Weight of Evidence (WoE), Logistic Regresion (LR) and WoE-LR Combination Methods

Bahaya Longsor di Kabupaten Sukabumi berbasis Metode Weight of Evidence (WoE), Logistic Regression (LR) dan Kombinasi WoE-LR

Abstract

The high frequency of landslides in Sukabumi Regency caused the need for data and information of potential landslides areas. The most widely used method to identify potential landslides is stastical method. Therefore, this study aims to predict landslide hazard in Sukabumi Regency. This research used Weight of Evidence (WoE), Logistic Regression (LR), and WoE-LR combination methods. Results showed that suitable parameters for running the models are distance from road, distance from river, distance from fault, SPI, TWI, elevation and slope. WoE method’s results showed elevation <300 m, distance from road >200 m and distance from friver >100 m are bad parameter classes to predict landslides in this study area. Whereas slope 8-15%, distance from road 31-70 m and elevation 700-800 m are good parameters to predict landslide potential. As for LR method, elevation and distance from road have significant effect on landslides. WoE-LR combination method’s results showed distance from road and SPI are bad parameters for predicting landslide potential. Conversely, slope and TWI are the best parameters to predict landslide hazards, including elevation, distance from fault and distance from river. Therefore it can be concluded that WoE-LR combination method is the best for predicting landslide hazard in the study area.