ECMWF细网格10 m极大风速预报在海南岛的评估与订正  

Verification and Correction of the ECMWF Ten-meter Wind Forecast for Hainan Island

作　　者：钟有亮 李勋[1,2] 陈静 刘凑华 ZHONG Youliang;LI Xun;CHEN Jing;LIU Couhua(Meteorological Observatory of Hainan Province,Haikou,Hainan 570203,China;Hainan Key Laboratory of Meteorological Disaster Prevention and Mitigation in the South China Sea,Haikou,Hainan 570203,China;National Meteorological Center,Beijing,100081,China)

机构地区：[1]海南省气象台,海南海口570203 [2]海南省南海气象防灾减灾重点实验室,海南海口570203 [3]国家气象中心,北京100081

出　　处：《热带农业科学》2024年第10期140-147,共8页Chinese Journal of Tropical Agriculture

基　　金：国家自然科学基金联合基金项目(No.U21A6001);海南省气象局科研项目(No.hnqxSJ202110);海南省重点研发项目(No.ZDYF2023SHFZ125)。

摘　　要：利用欧洲中期天气预报中心(European Centre for Medium-Range Weather Forecasts,简称ECMWF)细网格10m风预报场和海南岛国家气象站地面风观测资料,基于天气学误差统计等方法对2019—2021年10m极大风速预报结果进行评估,以期为预报员更好地应用模式产品提供依据。结果表明:(1)海南岛四周地区10 m极大风速预报效果优于中部地区;预报误差与海拔高度密切相关,海拔较低站点与实况观测一致性更高;模式对海拔低且开阔地区的极大风速具有较好可预报性。(2)10 m极大风速预报误差随时效略有增长,昼夜误差呈现波峰特征,具有一定日变化。(3)海南岛6级风速预报效果最佳,5级及以下风速预报次之,7级及以上风速预报效果则最差;对于大风预报,ECMWF细网格预报量级具有偏小的特征。(4)基于机器学习方法,选取ECMWF细网格预报场,对海南岛极大风速预报进行订正,独立样本预报模型表明,该方法可以有效减小预报误差,改善效果显著,为海南岛大风预报的准确性提供可靠方法。Using the ECMWF(European Centre for Medium-Range Weather Forecasts)ten-meter wind forecast and observa-tion data from the Hainan Island National Meteorological Station,the data from 2019 to 2021 were verified and evaluated via synoptic error statistics.It is better by forecasters.The results reveal that the surrounding areas of Hainan Island are better than those in the central region.The forecast error is closely related to the altitude.The effect of the ten-meter extreme wind speed in the surrounding areas of Hainan Island is better than that in the central region.The forecast error is closely related to alti-tudes and stations with lower elevations.The model has good predictability for extreme wind speeds in low altitude and open areas.The forecast time error of the ten-meter extreme wind speed slightly increased at any time,and the diurnal error exhib-ited wave peak characteristics with certain diurnal variations.It is best for wind speeds of level 6,followed by wind speeds of level 5 and below,and speeds of level 7 and above are deficient.The ECMWF has a littler magnitude characteristic of the Gale forecast.Based on machine learning methods,the ECMWF model was selected to correct the extreme wind speed forecasts on Hainan Island.The independent sample forecast model showed that this method could effectively reduce forecast errors,sig-nificantly improve the effect,and provide a reliable method for ensuring the accuracy of gale forecasts on Hainan Island.

关 键 词：ECMWF 检验评估 产品释用 10 m风场 

分 类 号：S533[农业科学—作物学]