Synergetic Use of Random Forest and H3 Grid Cell for the Flood Monitoring Using Sentinel-1 Data

Floods pose significant natural hazards worldwide, resulting in damage to lives, properties, and economies. Accurate flood mapping is critical for assessing and mitigating risks during such events. Microwave remote sensing data plays a pivotal role in monitoring flood hazards due to its all-weather capabilities. In this study, Sentinel-1 SAR data has been leveraged to estimate flood occurrences over time. The methodology integrates advanced techniques of H3 grid system for spatial aggregation and Random Forest models from SAR images. These approaches illustrate the extent of flooding and the associated damages during both pre- and postflood periods. It also conducts vulnerability assessments of agriculture area to gain insights into the severity of damage in areas adjacent to river channels. The analysis of polarization patterns consistently correlates with flood severity. It contributes significantly to understanding the factors influencing flood extents, thereby facilitating improved planning for risk reduction and management during future flooding events.