Ahmad Hasan Aoishiro Basori, Parwati Sofan, Bagus Setiabudi Wiwoho, Ike Sari Astuti
Drought has become an increasingly serious threat in Indonesia over recent decades, expanding in both coverage and duration, and significantly impacting agriculture, water availability, and food security. Although Indonesia typically experiences year-round rainfall as a tropical country, elevated Tropical Pacific Ocean surface temperatures can trigger El Nino events, reducing rainfall and extending drought periods across Indonesian territories. This study aims to map the spatial distribution of drought potential on Java Island using satellite data and machine learning techniques. The research employs meteorological, agronomic, and hydrological drought indices: Keetch-Byram Drought Index (KBDI) for atmospheric conditions, Vegetation Health Index (VHI) for biological conditions, and Topographic Wetness Index (TWI) for hydrological conditions. Population density data was incorporated to represent socioeconomic aspects across Java Island. The random forest method was used to distinguish between dry and normal conditions. Results were validated using drought data from the National Disaster Management Agency (BNPB). Classification results for July 2019 showed (53.5%) of dry land, with a validation accuracy of 84.2%. This research can serve as a reference for developing early warning systems for drought disasters in Java. © 2025 IEEE.
State University Of Malang, Departement of Geography, Malang, Indonesia; National Research and Innovation Agency (BRIN), Research Center for Geoinformatic, Bandung, Indonesia
