Landslide mapping using K-Means cluster by NDVI data in Garut, West Java, Indonesia

Abstract

Landslide mitigation is a primary concern in areas with significant elevation gradients. In addition to elevation, vegetation and hydrological factors also influence landslide potential. Therefore, this study analyzes landslide risk using satellite-derived vegetation data (NDVI). The results will be mapped to show relative susceptibility to landslides. The clustering method used is K-Means. NDVI data was obtained from NASA's Terra and Aqua satellites, and the analysis is based on vegetation levels. The study identifies six areas in Garut where landslides frequently occur. The K-Means clustering process categorizes gridcodes 3 and 4 in the NDVI data as potential landslide zones due to sparse vegetation. The clusters are divided into three groups, with cluster 1 having a silhouette coefficient close to 1, indicating robust clustering. The mapping results highlight the North Garut area as having a high potential for landslides.