Abstract:

Aiming at a significant heavy rainfall event that occurred in Inner Mongolia in July 2021, the paper conducfed a set of convection-allowing ensemble prediction (CAEP) experiments to evaluate its forecasting capability for intense precipitation processes, and compared the results with the global ensemble forecasts from the European Center for Medium-Range Weather Forecasts (ECMWF), the National Centers for Environmental Prediction Global Ensemble Forecast System (NCEP-GEFS), and the China Meteorological Administration Regional Ensemble Prediction System (CMA-REPS). The results show that the ensemble mean of global ensemble forecasts tended to underestimate the intensity of heavy precipitation centers, although ECMWF provided relatively accurate predictions of their locations. Both CMA-REPS and CAEP precipitation intensities forecasts close to observations but with some positional deviations, whereas NCEP-GEFS performed poorly in forecasting both the location and intensity of heavy rainfall. The Probability Matching Ensemble Mean (PM) effectively improved the simulated precipitation intensity compared with the traditional ensemble mean, leading to a notable increase in the threat score (TS), particularly for ECMWF and CAEP. The CAEP outperformed both global and regional ensemble forecasts in predicting the magnitude and temporal evolution of single-station precipitation. Objective verification indicated that ECMWF, CMA-REPS, and CAEP ensemble members exhibited certain forecasting capability for 25 mm·(6 h)^-1 precipitation, while NCEP-GEFS performed poorly. For 60 mm·(6 h)^-1 precipitation, CAEP achieved the highest TS, the lowest Brier score, and the highest AROC score among the ensemble systems, demonstrating its superior capability in forecasting heavy rainfall over the Inner Mongolia region.

Key words: ensemble forecast, Inner Mongolia, rainstorm, CMA-REPS, convective-allowing ensemble forecast

摘要：

针对2021年7月发生在内蒙古地区的一次显著性大暴雨过程，开展了一组对流尺度集合预报（Convection-Allowing Ensemble Prediction，CAEP）试验，以评估其对强降水过程的预报能力，并与欧洲中期天气预报中心（European Center for Medium range Weather Forecasts，ECMWF）全球集合预报、美国国家环境预报中心全球集合预报（National Centers for Environmental Prediction Global Ensemble Prediction System，NCEP-GEFS）以及中国气象局区域集合预报（China Meteorological Administration Regional Ensemble Prediction System，CMA-REPS）进行对比分析。结果表明，全球集合预报的集合平均对强降水中心的强度预报偏弱，其中ECMWF对强降水中心位置预报较准确；CMA-REPS和CAEP预报的降水强度与实况接近，但存在一定位置偏差；NCEP-GEFS在降水中心及强度的预报中均表现较差。概率匹配集合平均可有效改善集合平均的降水强度预报，TS评分较传统集合平均明显提高，其中ECMWF和CAEP的提升最为显著。CAEP在单站降水量级及其发展趋势的预报中优于全球及区域集合预报。客观检验结果显示，ECMWF、CMA-REPS和CAEP集合成员对25 mm·（6 h）^-1降水具有一定预报能力，而NCEP-GEFS表现较差；对于60 mm·（6 h）^-1降水，CAEP集合成员的TS评分最高，概率预报的Brier评分最低且可辨识度最高，表明CAEP在内蒙古地区强降水过程的预报中具有显著优势。

关键词: 集合预报, 内蒙古, 暴雨, CMA-REPS, 对流尺度集合预报