Dynamics-constrained rainfall projection reveals substantial increase in population exposure to unprecedented floods in the North China Plain

Understanding the changing flood magnitude and frequency is crucial for managing flood risks. Current flood projections are hindered by uncertain rainfall changes and inadequate representation of spatial-temporal rainfall characteristics from global climate models. Here we develop a framework for flood projection integrating large ensemble climate model outputs and dynamic downscaling simulations. By unveiling the shifting importance of changes in atmospheric circulation and water vapor to rainfall extremes of varying magnitudes, a scaling ratio is derived to constrain rainfall projections from thermodynamics-perturbed experiments. Applying this framework to the North China Plain, we reveal substantial rainfall increases partially attributed to more favorable synoptic conditions. This leads to a nearly fivefold rise in population exposure to moderate and high flood depths (i.e., 0.5 ~ 2 m). The region with heightened flood risk is spatially offset from that with rainfall increases. These findings highlight an urgency for improved flood risk management in the densely populated region. The North China Plain is experiencing a nearly fivefold increase in population exposure to moderate and high flood depth due to increased rainfall and favorable synoptic conditions, according to a methodology that combines large ensemble climate model outputs with dynamic downscaling simulations.