APPLICATION OF THE GUSTAFSON–KESSEL ALGORITHM FOR IDENTIFYING SPATIAL PATTERNS OF NATURAL DISASTERS IN EAST NUSA TENGGARA

Abstract

This study examines spatial patterns of disaster vulnerability across districts and cities in East Nusa Tenggara Province, one of Indonesia’s most disaster-prone regions. Although previous studies have highlighted the province’s exposure to multiple hazards, limited attention has been given to clustering methods capable of capturing non-homogeneous and elliptical data structures. This research aims to classify regional disaster vulnerability based on the characteristics of disaster occurrences and to provide empirical support for more targeted mitigation strategies. Secondary data on floods, forest fires, hurricanes, and landslides recorded in 2023 were analyzed using the adaptive Gustafson–Kessel clustering algorithm. The optimal number of clusters was determined using the Silhouette validity index. The results identify three distinct vulnerability groups: regions highly prone to multiple types of disasters, regions predominantly affected by a single hazard, and regions with relatively low disaster risk. The resulting spatial patterns reveal clear differences in disaster intensity and complexity among regions, emphasizing the need for location-specific disaster management policies. This study contributes to disaster risk analysis by demonstrating the applicability of the Gustafson–Kessel algorithm in capturing complex spatial vulnerability patterns that are often overlooked by conventional clustering approaches.