Landslide susceptibility modelling using the weight of evidence and logistic regression approaches in the Balease watershed, Indonesia

Abstract

North Luwu Regency exhibits a high susceptibility to landslides; ten major landslides in South Sulawesi occurred in the Balease watershed of North Luwu Regency over a ten-year period.We aimed to inventory landslide occurrences, identify the controlling factors of landslides, generate landslide hazard maps, and determine the most reliable method for landslide susceptibility modeling in the Balease watershed.Spatial statistical analyses using the weight of evidence (WoE) and logistic regression (LR) models were performed.Landslide inventory mapping was conducted through visual interpretation of SPOT-6 satellite imagery (2020) and WorldView-3 imagery (2024).The parameters included topographic, hydrological, geological, climatic, soil, and anthropogenic factors.Model validation was performed using the area under the curve (AUC) approach.The results indicate that approximately 1.832 mapped landslide locations occurred during 2020-2024.Notable differences in area coverage between the LR and WoE methods across landslide susceptibility classes, particularly, the LR classified a larger proportion of the study area as low susceptibility (99,932.95ha) compared to WoE (79,055.12ha).In contrast, WoE allocated a substantially larger area to the high susceptibility class (34,073.73ha) than LR (7,780.50ha), indicating that WoE is more sensitive in identifying areas with a higher likelihood of landslide occurrence.The prominent factors controlling landslide occurrence in Balease watershed were rainfall with 0.91 AUC value, slope gradient (0.89), elevation (0.86), and stream power index (SPI) (0.85).The WoE model achieved an AUC value of 0.88, classified as good, while the LR model yielded an AUC value of 0.77, classified as fair.The study's finding, the WoE of the evidence model driven mainly by rainfall and topographic factors, provides the most accurate landslide susceptibility assessment for the Balease watershed, supporting effective disaster mitigation and spatial planning.The resulting landslide hazard maps are expected to support spatial planning, disaster mitigation strategies, and conservation planning in landslide-prone areas of North Luwu Regency.

Access to Document

Other files and links

• Link to publication in Scopus
• Open Access Version Available

Fingerprint