Convection‐Permitting Simulations of Extreme Hourly Rainfall and Land‐Surface Impacts in the Middle‐Lower Yangtze River Basin

Abstract Extreme precipitation events frequently cause severe socioeconomic losses, making accurate forecasting critical. However, prediction of extreme precipitation remains challenging due to multi‐scale nonlinear interactions. Previous studies have largely attributed extreme rainfall to large‐scale circulation or mesoscale convective systems, the role of complex underlying surfaces remains insufficiently understood. Using station observations from 2010 to 2020, we identify that hourly extreme precipitation (HEP) is statistically correlated with the distribution of lakes and water bodies. This pattern is exemplified by an extreme rainfall event, which was poorly predicted by operational models. To investigate the underlying mechanisms, we conduct convection‐permitting simulations that can resolve lakes and urban surfaces. Two configurations are compared, a nested and a non‐nested simulation. The non‐nested simulation better captures rainfall intensity and propagation, avoiding the nesting‐induced suppression of convective potential. Building on the non‐nested simulation that better represent the precipitation characteristics, we apply the water vapor tracer (WVT) to quantify the local land‐surface contributions. The precipitation recycling ratio, defined as the fraction of precipitation originating from locally recycled moisture, reaches 10% even during this extreme event under strong large‐scale forcing. Furthermore, sensitivity experiments with observed lake surface temperature adjustment reveal that lakes modulate convection through latent heat release, particularly north of 30°N. Our findings highlight that representing heterogeneous lake–urban surfaces at convection‐permitting scales, and avoiding nesting‐induced biases, could improve HEP simulation and understanding land‐atmosphere interactions during extreme rainfall event.