GRAPES-TYM模式对我国东部近海海雾预报性能评估

Assessment of Global/Regional Assimilation and Prediction System-Typhoon Model(GRAPES-TYM) for Sea Fog Prediction over Eastern China Seas

基于2020年上半年我国东部近海站点观测资料和葵花-8卫星反演海雾产品对我国自主研发的GRAPES-TYM模式进行了海雾预报性能评估。点、面检验结果表明:模式48 h和72 h TS分别为0.40和0.36,黄海海雾预报性能最优,34°~37°N海域内大部分站点TS高于0.50。黄海海雾落区预报检验显示均压场形势下预报最准确,平均临界成功指数为0.35;气旋后部海雾多空报。2 m相对湿度预报偏差具有局地性特征,相对湿度低估的站海雾预报击中率、TS相对低,反之亦然。另外,模式对成雾相关气象要素预报误差相对小且对成雾有利时海雾预报基本正确;模式预报风向较实际风向偏东南,易出现冷偏差和湿偏差,虚假的有利温湿条件导致海雾空报。

Based on stational observations and retrieved sea fog products by the Kuihua-8 satellite from January to June 2020,this paper have evaluated the performance of the independent researched and developed GRAPES-TYM model to forecast sea fogs over the Eastern China Seas.Results indicate that sea fog forecasting for the Eastern China Seas by GRAPES-TYM is relatively promising,for TS of 48 h and 72 h is 0.40 and 0.36,respectively.Especially the model shows the best ability to predict the Yellow Sea sea fogs with TS of most stations in the sea area between 34°N and 37°N higher than 0.50.The further assessment indicates that the fog area forecast under the uniformed pressure field is more accurate,with an average CSI of 0.35.The model often gives false alarms when under the back of the cyclone.It appears that deviations between observed and forecasted surface air temperature and relative humidity are relative to locality.The hit rate and TS are lower with underestimating relative humidity,and vice versa.Our results also suggest that for successful cases,the model accurately captures the near-surface meteorological conditions,which are favourable for fog formation.In contrast,the model shows a large error in near-surface relative humidity and temperature for false alarm cases compared to successful cases.The predicted near-surface wind direction is often in the southeast of the actual wind direction,which is prone to introduce cold bias and overprediction of relative humidity.False favourable temperature and humidity conditions lead to false sea fog alarms.