​Evaluation of the Accuracy Level of Landslide Vulnerability Maps for Various Rainfall Models

Abstract

The Regional Disaster Management Agency (BPBD) and the Communication and Information Agency (Diskominfo) of Garut Regency recorded seven landslide events in 2020 and eight in 2023 in the Banjarwangi District. These events were triggered by steep to very steep slopes with landslide-prone soil and rock types and heavy rainfall. This study aimed to evaluate the accuracy of landslide hazard maps for various rainfall models, including daily, ten-day, monthly, and annual rainfall, validated with landslide occurrence points using the DVMBG 2004 parameter. The evaluation results showed that the maximum rainfall model had a hazard classification of 99.14%. The average rainfall model showed a less vulnerable classification, covering an area of 76.10%. The level of rainfall affects the classification of vulnerability, thereby impacting accuracy. The results of evaluating the accuracy of landslide hazard suitability for various rainfall models showed a low accuracy of 45.5%. Therefore, further analyses are required to improve the accuracy of landslide hazard maps.