《海洋预报》｜2019年海面风网格化预报检验

　　摘要：使用FNL（Final Operational Global Analysis）再分析资料中的10 m风速和我国沿海23个浮标的风速数据，并采用Cressman插值的方法对FNL再分析数据进行订正，将浮标数据融合到FNL中作为观测值，与欧洲中期天气预报中心（ECMWF）的数值预报进行对比，以达到对全国渔业网格化预报产品更优的检验效果。根据全国智能网格气象预报业务规定（试行）中的风向、风速检验规则进行判定，结果表明：2019年24 h、48 h和72 h的风向预报评分均可达0.80分以上，24 h和48 h的风向预报准确率达60%以上，72 h准确率则不足60%；24 h和48 h风速预报评分可达0.80分以上，72 h风向预报稍逊色，随着预报时效延长，预报准确率略有下降，预报偏强、偏弱率仅有2%左右的衰减，误差在1 m/s左右。自2022年起ECMWF数值预报已被国家海洋环境预报中心自主研发的集合预报产品取代，本文为其权重分配起到重要参考作用。

　　关键词：FNL再分析资料 数值预报 网格化 预报检验

　　Abstract：We use 23 buoys data over China offshore areas to optimize the 10 meter wind in the Final Operational Global Analysis (FNL) data by using the Cressman interpolation method, and then use the optimized FNL data to validate the national fishery gridded forecast product by comparison with the European Centre of Medium-Range Weather Forecast (ECMWF) product. With respect to the wind direction and speed inspection rules from the national smart grid weather forecast service regulations (trial), the wind direction forecast scores of 24 h，48 h and 72 h show a decreasing trend over time but all exceeding 0.80. The accuracy of 24 h and 48 h wind direction forecasts are more than 60% while that of 72 h forecast is less than 60%. Wind speed forecast scores of 24 h and 48 h are higher than 0.80. Along with the increase of forecast leading time, the forecast accuracy slightly decreases, and the rate of forecast bias only has an attenuation of 2%, while the wind speed error is about 1 m/s. Since 2022, the National Marine Environmental Forecasting Center utilizes the self-developed operational ensemble wind forecast product instead of the ECMWF data. This study provides an important reference in determining weights of the multi-products.

　　Key words：FNL reanalysis data, numerical forecast, gridded, forecast test