An Integrated Framework for Coastal Flood Inundation Forecasting: Advancing Early Warning Systems in Vulnerable Deltaic Regions

Coastal regions, particularly deltaic systems, are highly susceptible to flood risks arising from the complex interactions of storm surges, riverine flooding, upstream reservoir discharges, and heavy inland rainfall. Traditional flood forecasting models often struggle to integrate these diverse factors effectively, leading to significant uncertainties in predicting flood extents. To address this critical gap, this study presents a novel and comprehensive coastal flood inundation forecasting framework designed for regions frequently impacted by tropical cyclones and extreme hydrological events. The framework integrates multiple components, including real-time reservoir inflow forecasting by SWAT-Pothole+WBiLSTM model, HEC-ResSim-based reservoir outflow predictions governed by operational rule curves, storm-surge and tide predictions utilizing ADCIRC+SWAN hydrodynamic and WBiLSTM deep learning approaches, and flood inundation modeling by HEC-RAS two-dimensional hydrodynamic simulation. The methodology was applied to the twin Brahmani-Baitarani river systems in eastern India, a region prone to recurrent cyclonic storms and severe flooding. Validation of simulated flood extents was conducted using Sentinel-1 satellite imagery from several tropical cyclone events, demonstrating the model's robust predictive capabilities. The results showed that the framework achieved accuracy levels ranging from 86.72% to 38.12% for lead times between one and eight days. Additionally, the model underscores the importance of incorporating all contributing factors, including the dynamic interaction of coastal and inland flooding processes, to achieve realistic flood forecasts. This research not only advances the understanding of coastal flooding but also offers a practical and scalable tool for enhancing early warning systems through informed flood risk management strategies in vulnerable coastal regions worldwide.