Abstract:

This study evaluated forecast performance of the 24-h precipitation induced by Northeast Clod Vortex using the European Center for Mesoscale Weather Forecast (ECMWF) model, the Global/Regional Assimilation and Prediction System (GRAPES), and Global Forecast System of the National Centers for Environmental Prediction (NCEP_GFS), in order to improve the prediction capability of numerical models in Fuxin and provide scientific references for the choice and adjustment of physical parameter schemes in models.Using the precipitation products from May to August of 2019, the tests for the accuracy of sunny/rainy weather, general precipitation, and grading precipitation, as well as their mean absolute errors and mean errors were conducted at two national observation stations (Fumeng station and Zhangwu station) in Fuxin.The results showed that the precipitation forecast of the ECMWF model is better than that of the other two models.The TS score of ECMWF products is higher during the cold vortex development period and the mature period and is lower during the decay period.In terms of the mean absolute error, the ECMWF has the smallest error at Fumeng station from May to August, and the forecast performance of GRAPES and NCEP_GFS is unstable.At the Zhangwu station, the ECMWF has no obvious advantage, the GRAPES performs best from May to June, and the NCEP_GFS performs best in July and August.The mean error analysis shows that all three models underestimate precipitation at Fumeng and Zhangwu stations except for an overestimation of precipitation at Zhangwu station in June.All models underestimate the precipitation of heavy rain and above, with the highest TS score of 40.2% from the ECMWF model.Moreover, all three models have higher precipitation TS scores at Fumeng station than Zhangwu station.

Key words: Global model, Northeast cold vortex, Periodic precipitation, Test evaluation