Assessment of Km‐Scale Rainfall Forecasts Over Eastern China by GRIST: Intercomparison of Explicit and Parameterized Convection

Abstract This study conducts numerical weather prediction experiments utilizing the Global‐to‐Regional Integrated forecast SysTem (GRIST) model during the 2021 Meiyu period. A global 60‐3 km variable‐resolution mesh with a refined region centered over eastern China is applied in this study. The GRIST with the conventional new Tiedtke (GRIST‐NTDK) convection parameterization scheme (CPS), the scale‐aware NTDK CPS (GRIST‐SAT), and the explicit convection (GRIST‐noCPS) setups are intercompared. GRIST‐noCPS forecasts are also compared with those from the Weather Research and Forecasting (WRF) model, using equivalent configurations (WRF‐noCPS) over the refined region. Compared to the WRF‐noCPS experiment, both spatial distributions and statistical metrics reveal that the GRIST‐NTDK, GRIST‐SAT, and GRIST‐noCPS experiments can capture the observed rainfall for a few days, although with some overestimation. Most of the rainfall overestimation from GRIST experiments is due to nocturnal rainfall bias, which is associated with excessive water vapor transport in the lower troposphere. Relative to GRIST‐NTDK and GRIST‐SAT, GRIST‐noCPS could somewhat alleviate the pressure gradients and southerly winds bias, favoring less severe overestimation. Although scale‐aware configurations of CPS can reduce light rainfall and increase heavy rainfall over the refined region to some degree, there are still some gaps between the GRIST‐SAT and observations in terms of rainfall frequencies. In contrast, GRIST‐noCPS can mitigate the widespread light rain, as exhibited in GRIST‐NTDK and GRIST‐SAT, resulting in improved quantitative precipitation forecast skills. Notably, the performances of GRIST over the refined region demonstrate its potential to overcome the time limit of regional models, such as WRF, at extended forecasting periods.