摘要
针对内蒙古2020年11月17日08时至20日08时的一次暴雪过程,采用站点检验方法对降雪量最大的24 h进行多模式和产品预报检验。两个全球模式GRAPESGFS和ECMWF预报准确率高于中尺度模式GRAPESMESO和RMAPS,EMWFC稳定性更强,但GRAPESGFS预报落区的空间形态更准确,提前24h暴雪预报效果也最好。所有模式都存在严重漏报,尤其是中雪和大雪两个量级;中尺度模式漏报更明显,未表现出其通常在暴雨预报中存在的强命中率伴随高空报率的特征。中央指导预报量级偏大,空报偏多,暴雪位置不准确,但改善了模式普遍存在的暴雪漏报情况。内蒙网格预报量级也偏大,订正了中央指导中雪以上的位置偏差,大雪和暴雪分别提升了1.5%和16.9%;另外订正了大部分模式48h的整体漏报。实现了从模式到中央再到内蒙预报水平的逐级提升。所有模式和产品都未能预报极端降雪。
Aiming at a heavy snowstorm in Inner Mongolia on November 17 to November 20, 2020, the multi-model and product forecast were evaluated by the traditional station test method. The threat score(TS) of the two global models, GRAPES-GFS and ECMWF, is higher than that of mesoscale models, GRAPES-MESO and RMAPS. In addition, the EMWFC is more stable, while the GRAPES-GFS is more accurate in predicting the shape of snow area and the 24 h snowstorm area. All the models have serious missing ratio(MR), especially of the moderate and the heavy snow, as well as the mesoscale model, which does not show the characteristics of the higher TS with the false alarm ratio(FAR) that usually exists in the rainstorm forecast. The National Meteorological Centre(NMC) forecast has more heavy rainfall forecast with the FAR, and its location of the snowstorm is incorrect. The forecast of Inner Mongolia(IM) Meteorological Observatory is also has more heavy rainfall forecast, but the position deviation of moderate snow has been revised. Moreover its accuracy of heavy snow and snowstorm have been increased by 1.5% and 16.9% from NMC’s forecast. And its product corrected the higher MR of most models on the 48 h. This article realized the forecast level step by step from the models to the NMC and then to IM. All models and products can not predict extreme snow.
作者
杭月荷
孙永刚
李一平
Hang Yuehe;Sun YongGang;Li Yiping(Inner Mongolia Meteorological Observatory,Inner Mongolia Hohhot 010051)
出处
《内蒙古气象》
2020年第6期3-5,22,F0003,共5页
Meteorology Journal of Inner Mongolia
关键词
暴雪
数值预报
检验评估
Snowstorm
Numerical forecast
Model evaluation
