Spatiotemporal Analysis of Inundated Areas in San Miguel, Bulacan using Sentinel-1 and Sentinel-2 Imagery

Abstract. The municipality of San Miguel, Bulacan lacks time-series flood extent maps and corresponding spatiotemporal analyses of flood dynamics and trends despite an increase in flood frequency and severity the last few decades, which resulted in poor disaster preparedness and response. This study aims to extract the inundated areas and visualize the extent of flooding in San Miguel over time using Sentinel-1 images to determine the flood extents, and Normalized Difference Water Index (NDWI) and Sentinel Water Mask (SWM) derived from Sentinel-2 to extract permanent water bodies (PWB). The PWBs extracted from Sentinel-2 were combined with other PWBs based on the 2020 Land Cover Map of NAMRIA and stream network data from DENR. Results show that 40 of 49 barangays were classified as having a high frequency of inundation with pixels detected 21-28 times. A comparative analysis of rainfall data validates these results showing a generally positive correlation (0.0005 to 0.77). Moreover, Bagong Pag-asa, Pulong Duhat and Bantog were also identified to be the most frequently inundated barangays in the municipality. Using the flood hazard maps of Bulacan Provincial Disaster Risk Reduction and Management Office (PDRRMO) as reference data, the flood frequency map have accuracy metric values of 71.64%, 88.74%, and 79.28% for precision, recall, and F1 score, respectively, showing its potential in mapping flood extents and flood frequency. Future studies can explore other satellite data with better spatial and temporal resolution and utilize other surface water indices in identifying flood features in optical images.