Medium–Extended Range Forecasting of Meiyu in 2023: Performance of CMA and ECMWF GEPSs

Meiyu delivers prolonged periods of rainfall to the Yangtze–Huai River basin (YHRB). While it replenishes water resources and maintains ecological balance, it also has the potential to cause disasters. Early and accurate forecasting of Meiyu is crucial for effectively deploying prevention strategies against potential flooding. To help further refining numerical models and providing useful guidance for operational forecasters, this study explores the capabilities of two global ensemble prediction systems (GEPSs) of CMA and ECMWF in forecasting the Meiyu characteristics in 2023 over the YHRB at the medium–extended range. Results show that ECMWF GEPS reasonably forecasts the Meiyu rainfall in 2023, while CMA GEPS presents notable underestimation. ECMWF and CMA GEPSs successfully capture the Meiyu onset date 8 days and 6 days in advance, respectively. For the regional rainstorm processes, the predictable lead times of the two GEPSs for reasonably forecasting the pattern of the heavy rainfall area generally reach 48–168 hours. Sources of the merits and shortcoming of the two GEPSs in forecasting Meiyu are further traced back to their capability in predicting associated key circulation systems. By verifying against their own analyses of the two GEPSs, it is demonstrated that ECMWF GEPS reasonably well forecasts the spatial coverages of the northwestern Pacific subtropical high (NWPSH) and the South Asian high (SAH), while CMA GEPS presents substantial underestimation. Both GEPSs exhibit northward deviations for the western ridge line position of the NWPSH during early forecast lead timess. Compared to ECMWF GEPS, the less Meiyu rainfall yielded by CMA GEPS is attributable to the weaker low-level convergence belt and the weaker upper-level divergence area. Further digging into the origins of the forecast discrepancies between the two GEPSs concerning upper-level divergence and lower-level convergence suggests that these differences likely stem from their respective initial analysis fields.