Advanced Flood Risk Prediction Leveraging Geospatial Satellite Insights

Accurate flood risk prediction is critical for disaster management, particularly given the increased unpredictability of weather patterns caused by climate change. This work applies to a 1-day-ahead flood forecast model that combines geospatial data, including Landsat and Sentinel-1 SAR observations, with machine learning methods. The results show great prediction accuracy, with a correlation coefficient. Surface water conditions and soil surface moisture are both significant predictors, accounting, respectively. The model accurately anticipated inundation patterns during the peak and recession stages of the Idai flood, with results closely matching satellite-derived data. This method emphasizes the need of merging machine learning and remote sensing to provide actionable flood risk assessments, resulting in improved disaster planning and resilience methods.