Flood exposure in rapidly urbanizing regions is increasingly dynamic, influenced not only by the spatial extent of flood hazards but also by temporally varying human mobility patterns and urban morphological structures. This study assesses dynamic flood exposure risk across the Yangtze River Delta Urban Agglomeration (YRDUA) by integrating the CaMa-Flood hydrodynamic model with hourly population distribution data from Baidu heatmaps. A multi-dimensional urban morphology indicator system is constructed, and random forest modeling, complemented by SHapley Additive exPlanations (SHAP) analysis and Spearman rank correlation, is applied to quantify the drivers of spatiotemporal exposure variability. Results reveal significant diurnal variation in flood exposure, with intraday differences reaching up to 77.0% due to commuting-driven population shifts into flood-prone urban cores. Cities with monocentric-dominated spatial structures exhibit 91.8% higher exposure than polycentric ones, and maximum exposure can exceed minimum levels by 351.7%. Built environment and socioeconomic-functional dimensions collectively play a more influential role than natural-physical morphology, as impervious surface ratio (21.7%), economic radiation intensity (13.1%), and spatial clustering index (12.6%) rank as the three most important predictors. Natural-physical indicators such as river network density and channel sinuosity contribute less prominently (<8% each). This framework provides a scalable, time-sensitive approach for assessing flood risk in complex urban systems and highlights the critical role of urban form and population mobility in shaping more flood-resilient cities.
Liu et al. (Wed,) studied this question.