王在文,陈敏. 2019. 相似集合预报方法在北京区域地面气温和风速预报中的应用[J]. 气象学报, (0):-, doi:10.11676/qxxb2019.044
相似集合预报方法在北京区域地面气温和风速预报中的应用
Application of Analog Ensemble method to surface Temperature and wind speed prediction in Beijing area
投稿时间:2018-06-19  修订日期:2018-12-27
DOI:10.11676/qxxb2019.044
中文关键词:  相似集合,SVM,要素释用,集合预报
英文关键词:ANEN, SVM, Elements prediction, Ensemble
基金项目:北京市科技计划项目(Z161100001116098)
作者单位E-mail
王在文 中国气象局北京城市气象研究所 zwwang@ium.cn 
陈敏 中国气象局北京城市气象研究所 mchen@ium.cn 
摘要点击次数: 534
全文下载次数: 334
中文摘要:
      相似集合(Analog Ensemble: AnEn)是近年来提出的一种基于相似理论、大数据挖掘和集合预报思路的统计释用方法。首先介绍了相似集合的基本原理,并应用[资料和方法]该方法对BJ-RUCv3.0系统预报[目的]地面要素开展了订正释用试验。结果表明,相似集合订正后,[结果]10-m风速和2-m温度在0-36h预报时段内,10-m风速的均方根误差降低44%,2-m温度的均方根误差降低22%,均方根误差均显著减小。对比测站预报误差的水平分布,相似集合方法的应用对于提升非城区站点的10-m风速预报、复杂地形区域的2-m温度预报具有更为明显的效果。相同预报因子的相似集合和支持向量机(Support Vector Machines: SVM)方法对模式10-m风速和2-m温度预报均具有显著且相似的订正效果,但相似集合方法具有计算资源需求较少、不需要大量人工干预的优势。相似集合方法形成的集合较好地模拟了模式平均误差的增长情况,集合离散度与集合平均均方根误差表现出理想的统计一致性,即相似集合方法在形成确定性预报的同时,还能够提供预报要素的不确定性或概率信息。因此[结论]相似集合方法在模式预报订正及释用方面将具有广阔的应用前景。
英文摘要:
      Analog Ensemble (AnEn) is a statistical interpretation method that based on similarity theory, big data mining and ensemble forecasting. At first introduces the basic principle of this method, and applied the method to revise the ground elements prediction of BJ-RUCv3.0. The results showed that the root-mean-square-error (RMSE) of 10-m wind speed and 2-m temperature was significantly decreased during the lead times of 0-36h after used AnEn, the RMSE of 10-m wind speed decreased by 44%,and the RMSE of 2-m temperature decreased by 22%,the RMSE decreased significantly. Compared with the prediction error of stations, the application of AnEn method has a more obvious effect on non-urban area stations for 10-m wind speed prediction, and on complex terrain area stations for 2-m temperature prediction. There are significant and similar effects on 10-m wind speed and 2-m temperature prediction of NWP model after used AnEn and Support Vector machines (SVM) with the same predictors. However, AnEn method has the advantage of need less computing resource and less manual intervention work. AnEn method simulated well for the pattern of average error growth, and showed the ideal statistical consistency between RMSE of AnEn mean and average ensemble spread. It not only give deterministic predictions, but also provide uncertainty, or probability information for prediction factors. Therefore, AnEn method will have a broad application prospect in NWP model interpretation prediction.
查看全文   查看/发表评论  下载PDF阅读器
分享按钮