月平均气温集成预测方法研究  

The Ensemble Forecasting Study of Temperature Prediction

作　　者：雷向杰[1] 

机构地区：[1]陕西省气候中心,西安710015

出　　处：《气象》2011年第12期1560-1565,共6页Meteorological Monthly

基　　金：陕西省科学技术研究发展计划项目(2011K17-02-03);中国气象局2011年现代气候业务建设试点项目(气预函[2010]174号);陕西省气象局2010年气象科技创新基金项目(2010M-19)共同资助

摘　　要：为了有效综合各种预测方法的预测结果,以陕西3个气候区域1999年7月至2010年6月月平均气温的集成预测为例,利用3种预测方法和国家气候中心业务产品月平均气温预测结果历史评分建立二类6种动态客观集成预测方法进行对比研究,结果表明:(1)各种集成预测方法年评分平均值均高于参加集成的任何一个成员,6种集成预测方法评分平均值比4个成员评分平均值高4.6,比同期陕西发布的业务产品高5.1;各种集成预测方法预测与实况距平符号一致率至少高于其中3个成员,6种集成预测方法预测与实况距平符号一致率平均值比4个成员平均值高4.7%,比陕西业务产品高5.1%。(2)3个气候区域中,榆林集成预测效果最好。6种集成预测方法评分平均值比4个成员评分平均值高5.7,比陕西业务产品高5.7;预测与实况距平符号一致率比4个成员高6.9%,比陕西业务产品高7.3%。(3)第二类集成预测方法预测技巧高于第一类,全年评分平均值高0.5,预测与实况距平符号一致率高0.7%。其中Z_(23)全年评分平均值比4个成员评分平均值高5.1,比陕西业务产品高5.7,预测与实况距平符号一致率比4个成员高6.0%,比陕西业务产品高7.6%,推荐首先选择使用。In order to effectively integrate the prediction results of various forecasting methods,the monthly mean temperatures of the three climatic zones in Shaanxi from July 1999 to June 2010 are employed in the experiments.Based on the historical precision derived from products of the National Climate Centre and three kinds of prediction methods,the two schemes involving six objective-integrated dynamic forecasting models are established.The results show that;（1） The yearly average score is higher than each member in the range of the whole ensemble forecasting models.Average score of six ensemble forecasting models is 4.6 higher than other four members,and is 5.1 higher than operational publishing products provided by Shaanxi Climate Center.The yearly anomaly sign consistency rate of each ensemble forecasting model is higher than every value resulting from at least three prediction methods among all.The average value of six ensemble forecasting models with the anomaly sign consistency rate is 4.7%higher than other four members,and is 5.1%higher than operational publishing products predicted by Shaanxi Climate Center. （2）Yulin received the best forecasting result among three climatic zones.Average score of six ensemble forecasting models is 5.7 higher than other four members,and is 5.7 higher than operational publishing products provided by Shaanxi Climate Center.The average value of ensemble forecasting models with the anomaly sign consistency rate is 6.9%higher than four members,and is 7.3%higher than operational publishing products predicted by Shaanxi Climate Center.（3）The second scheme of ensemble forecasting models is better than the first one.Especially,the average score of Z_（23） is 5.1 higher than other four mem- bers,and is 5.7 higher than operational publishing products.The average value of ensemble forecasting models with the anomaly sign consistency rate is 6.0%higher than four members,and is 7.6%higher than operational publishing products score.Thus,it is recommended to be used in climate p

关 键 词：月平均气温 气候预测 客观集成 预测准确率 

分 类 号：P456[天文地球—大气科学及气象学]