Modeling Dynamic Flood Population Exposure in Coupled Human‐Water Systems: The Role of Reservoir Regulation and Population Movement

Flooding is one of the most devastating natural hazards affecting human society. Accurate exposure assessment is critical for effective risk management. Traditional approaches based on static overlays of population and flood hazard data overlook both cross‐city mobility dynamics in urban agglomerations and reservoir‐induced hazard shifts. This work calculates dynamic flood population exposure in coupled human‐water systems by integrating hydrological modeling with human mobility data. The ERA5‐Land and Today's Earth data sets were used as forcing data to drive the Global River Hydrodynamics Model (CaMa‐Flood) for fluvial flood simulation. Human spatiotemporal distribution maps were derived from mobile positioning data through spatiotemporal reorganization and Kernel Density Estimation (KDE) interpolation. It is found that incorporating reservoir operations reduces the hazard level for 8%–11.8% of the exposed population under flood scenarios with return periods of 5–100 years. Diurnal population fluctuations alter flood exposure, with nighttime exposure surpassing daytime levels by 681–777 thousand people for 5–100 years return period events. The proposed method provides a novel framework to assess dynamic flood population exposure by coupling water conservation regulation with cross‐city population movement, supporting collaborative flood management in urban agglomerations.

Plain Language Summary: Floods are one of the most severe natural disasters. Accurate assessment of population exposure to flooding provides a critical scientific basis for urban disaster prevention and mitigation planning. Traditional approaches, based on static overlays of population and flood hazard data, critically overlook cross‐city mobility dynamics in urban agglomerations and reservoir‐induced hazard shifts. To address this, we simulated fluvial flood inundation under reservoir‐operated and natural conditions across multiple return periods. Overlaying these results with high spatiotemporal‐resolution population distributions yielded dynamic flood population exposure results. Our analysis revealed that reservoir regulation effectively reduced flood depths. Furthermore, nighttime population exposure was substantially higher than daytime levels. The key contribution of this work is a new framework for assessing dynamic flood population exposure in coupled human‐water systems. It places particular emphasis on the understudied effects of diurnal commuting patterns. While our study area focuses on the Pearl River Delta urban agglomeration, the framework holds significant potential to aid adaptation efforts globally.