基于最优训练期的风力预报小时级订正释用方法  

Revised interpretation method of wind forecasting on hourly level based on optimal training period

作　　者：刘菡 於敏佳 LIU Han;YU Minjia(Zhoushan Meteorological Bureau,Zhoushan 316000,China)

机构地区：[1]舟山市气象局,浙江舟山316000

出　　处：《海洋预报》2023年第5期10-22,共13页Marine Forecasts

基　　金：浙江省气象局一般项目(2021YB29)。

摘　　要：基于2012—2021年浙江省自动气象站逐日10 min测风资料和2020-2021年浙江省气象局Fruit细网格产品,对比了线性回归、加权误差、平均误差3种最优训练期模型,分析了指标站点最大风速与极大风速关系。结果表明:加权误差法和平均误差法的订正效果明显优于线性回归法,平均误差法的业务应用效果最好,2h、4h、10h分别为11~12h、13h、14~72h预报时效的大概率最优训练期;最大风速与极大风速以一元线性关系为主,且阵性系数与地形、下垫面、天气系统等有关;基于最优训练期的风力预报小时级订正释用方法能有效提高客观风力预报的准确度、精细度,与实际业务中惯用的人工经验方法相比,各站点的修正率为9%~34%,订正释用方法优于人工经验,因此实际业务应用价值更好。Based on the daily 10-minute wind measurement data of automatic weather stations in Zhejiang Province from 2012 to 2021 and Fruit fine grid product of Zhejiang Meteorological Bureau from 2020 to 2021,three optimal training period models including linear regression,weighted error and average error are designed and compared,and the relationship between maximum and extreme wind speeds at index stations is analyzed.The results show that:The correction effect of the weighted and average error methods is obviously better than that of the linear regression method.The average error method has the best operational application effect.2 h,4 h and 10 h are the large probability optimal training periods of 11~12 h,13 h and 14~72 h forecasts,respectively.The relationship between maximum and extreme wind speeds is mainly linear,and the fitting coefficient is related to terrain,underlying surface,weather system,etc.The hourly correction and interpretation method of wind forecasting based on the optimal training period can effectively improve the accuracy and precision of objective wind forecasting.Compared with the actual business customary manual experience,the correction rate of each station is 9%~34%,which is better than the manual experience.The practical application value is good.

关 键 词：小时级 风力预报 订正释用 最优训练期 业务应用 

分 类 号：P732.1[天文地球—海洋科学]