Zukhrufia Rahmi, Parwati Sofan, Ike Sari Astuti
Floods are the primary hydrometeorological disaster in Indonesia, with rainfall serving as a key controlling factor. Beyond environmental damage, floods cause significant human casualties across affected regions. Satellite data can effectively map flood inundation patterns. The SATGPT platform, based on Joint Research Center (JRC) Global Surface Water Mapping data, quantifies how frequently areas experience flood inundation over specified time periods, excluding permanent water bodies such as rivers and lakes. This frequency map serves as a flood hazard map indicating where floods commonly occur. This study validates the SATGPT flood inundation map against historical flood records from the National Disaster Management Agency (BNPB) and determines rainfall thresholds from Climate Hazards Group InfraRed Precipitation with Station (CHIRPS) data that trigger flood events. The validation methodology employs spatial buffering of flood geo-location data combined with descriptive and predictive statistical analyses, including mean, median, standard deviation, and percentile calculations. Results demonstrate that the SATGPT flood map achieves 83% accuracy when compared to BNPB historical data from 2021-2025, particularly for January when most floods occur. The SATGPT data successfully represents historical flood events across various Indonesian regions. Using percentile analysis, the threshold for extreme rainfall associated with flood conditions is determined based on CHIRPS data. However, this calculation covers the entire Indonesian territory without accounting for regional climate variability. These findings demonstrate the potential of remote sensing technology for flood mapping and early warning systems in flood-prone communities. © 2025 IEEE.
Malang State University, Geography Department, Malang, Indonesia; National Research and Innovation Agency, Research Center for Geoinformatics, Bandung, Indonesia
