月平均环流的动力集合预报研究  

A EXPERIMENTAL RESEARCH ON THE ENSEMBLE DYNAMIC PREDICTION OF MEAN MONTHLY CIRCULATION

作　　者：王向东[1] 李维京[1] 罗艳 叶正青[1] 梁益国[1] 罗勇[1] 刘海波[1] 

机构地区：[1]国家气候中心

出　　处：《应用气象学报》1997年第A00期134-144,共11页Journal of Applied Meteorological Science

基　　金：国家"九五"重中之重项目"我国短期气候预测系统的研究";国家气候中心业务发展基金

摘　　要：文章对１９９６年１至９月共２３次月动力延伸集合预报的情况作了总体评估．结果表明，集合预报距平相关系数比单个成员预报距平相关系数的算术平均有明显提高，因而采用集合预报方法有利于提高预报的可靠性．在此基础上，对预报距平相关系数的季节变化及其原因作了初步的分析、解释．还从集合预报距平相关系数与集合成员离散程度之间的关系就集合预报技巧的预报问题进行了初步探讨．此外，文章还就Ｔ６３Ｌ１６全球谱模式对北半球欧亚地区月尺度环流异常的预报能力进行了个例分析．最后，考虑到月动力延伸预报距平相关系数随时间迅速衰减的事实，对逐日预报结果设计了两种不同的方法进行集成处理，制作月预报集合．试验发现两种方法对集合预报距平相关系数的改进有比较明显的作用．By use of a T63 L16 global spectral model, 23 ensemble dynamic predictions of the mean monthly circulation of 500 hPa and their anomaly during the quasi operation period from January to September of 1996, have been carried out and analysed. It is shown that the anomaly correlation coefficients (ACC) of ensemble predictions are much higher than those of the arithmetical mean of the corresponding individual predictions. As a result, the confidence level could be increased by using ensemble dynamic prediction. Besides, the seasonal variation properties of ACCs and the causes have also been primarily analyzed. The forecasting problem of the skill of ensemble prediction has been discussed by analyzing the relationship between the ACCs of ensemble predictions and the discrete degree of individual predictions. Meanwhile, a case study on the ability of T63 L16 model to predict the persistent abnormality of general circulation has been done. Finally, considering the fact that the ACC of monthly dynamic extended forecast usually decrease with integration time, two schemes have been designed to process the daily output of the predictive integration. It has been found that, both schemes can improve the current prediction to a certain degree.

关 键 词：环流动力 集合预报 大气环流 月平均环流 

分 类 号：P457[天文地球—大气科学及气象学] P434