横断山区模式降水预报质量检验——以丽江市为例

Quality Test of Precipitation Forecasting of Numerical Models in Hengduan Mountain—A Case Study of Lijiang

摘要: 为充分了解全球数值预报模式和中国气象局数值业务区域模式(CMA区域模式)在丽江市的预报效果，本研究对欧洲中心全球模式(ECMWF)、中尺度天气数值预报系统(CMA-MESO)、上海数值预报模式系统(CMA-SH9)、广东快速更新同化数值预报系统(CMA-GD)、云南区域模式(科研所WRF)、中国气象局全球同化预报系统(CMA-GFS)共6种数值预报模式的降水要素进行了晴雨预报与分级降水预报检验。结果显示：1) 欧洲中心全球模式晴雨预报正确率低于国产模式，且国产模式中以科研所WRF和CMA-SH9准确率最高。2) 随着降水量级增大，各模式TS评分和预报偏差显著下降，空报率和漏报率显著上升，且各模式对于小雨量级具有较好的预报效果。随着预报时效增长，各模式检验指标无显著差异，但其变化趋势与降水量级间表现一致。3) 综合多个模式的预报检验结果，丽江市进行晴雨预报和降水预报时可着重参考CMA-SH9、科研所WRF和CMA-GD这3种国产数值模式。

Abstract: In order to fully understand the forecasting effect of the global numerical prediction model and the numerical operational regional model (CMA regional model) of the China Meteorological Administration in Hengduan Mountain, this study evaluated the prediction results of barometer and graded rainfall precipitation in Lijiang of six numerical prediction models, the European Central Global Model (ECMWF), CMA-MESO, CMA-SH9, CMA-GD, WRF of the Research Institute and CMA-GFS. The results showed that: 1) The accuracy of ECMWF is lower than that of the domestic models, and the accuracy of the domestic model WRF and CMA-SH9 is the highest. 2) With the increase of precipitation, the TS score and forecast deviation of all models decrease significantly, and the empty reporting rate and underreporting rate increase significantly, and all models have a good forecasting effect on light rain. With the increase of forecast duration, there are no significant differences between the test indicators of each model, the varied trend is consistent with that of precipitation levels. 3) Based on the forecast results of multiple models, three domestic numerical models can be used to refer to CMA-SH9, WRF and CMA-GD in Lijiang for weather and precipitation forecasting firstly.