Evaluation of High-resolution Regional Numerical Models on the Diurnal Variation of Precipitation During the 2024 Meiyu Season in Zhejiang Province

To investigate the characteristics and underlying causes of forecast biases in the diurnal variation of precipitation in Zhejiang Province, this study evaluated two systems: the Zhejiang High-Resolution Rapidly Updated Assimilation Forecast System (ZJ1KM, 0.01°) and the China Meteorological Administration Mesoscale Model (CMA-MESO, 0.03°). Model performance was assessed across two key regions, Zhejiang–Shanghai and the Zhejiang–Anhui mountainous junction, using a multi-source suite including CMPAS_1 km merged precipitation data, ERA5 reanalysis, and station-based microwave radiometer observations, and the results indicated that both ZJ1KM and CMA-MESO captured the fundamental spatial distribution of precipitation across the study areas. Although ZJ1KM accurately represented the spatial distribution of the observed peak precipitation times, it significantly overestimated the precipitation intensity over Zhejiang–Shanghai and exhibited notable biases in the diurnal variation. In contrast, CMA-MESO provided a more accurate representation of both the diurnal precipitation variation and the associated low-altitude wind evolution in both key regions. In the Zhejiang–Shanghai region, the overestimation of seasonal rainfall was primarily driven by errors in convective precipitation. Although both models overestimated the morning precipitation peak owing to an inherent wet bias, the diurnal cycle simulated by CMA-MESO remained closer to the observations. Notably, the ZJ1KM model exhibited excessive convective intensity and vertical development. This led to a pattern of significant overestimation, followed by subsequent underestimation.