秦正坤,林朝晖,陈红,孙照渤. 2011. 基于EOF/SVD的短期气候预测误差订正方法及其应用[J]. 气象学报, 69(2):289-296, doi:10.11676/qxxb2011.024
基于EOF/SVD的短期气候预测误差订正方法及其应用
The bias correction methods based on the EOF/SVD for short-term climate prediction and their applications
投稿时间:2009-02-02  修订日期:2009-07-14
DOI:10.11676/qxxb2011.024
中文关键词:  短期气候预测,误差订正方法,奇异值分解,经验正交函数分解
英文关键词:Short-term climate prediction, Bias correction method, SVD analysis, EOF analysis
基金项目:国家科技支撑计划项目(2007BAC29B03,2008BAC40B02)、国家自然科学基金项目(40821092)及南京信息工程大学科研基金资助项目(20070098)
作者单位
秦正坤 南京信息工程大学气象灾害省部共建教育部重点实验室南京210044中国科学院大气物理研究所国际气候与环境科学中心北京100029 
林朝晖 中国科学院大气物理研究所国际气候与环境科学中心北京100029 
陈红 中国科学院大气物理研究所国际气候与环境科学中心北京100029 
孙照渤 南京信息工程大学气象灾害省部共建教育部重点实验室南京210044 
摘要点击次数: 3676
全文下载次数: 2952
中文摘要:
      利用中国科学院大气物理研究所第2代短期气候数值预测系统(IAP-DCP II)1980—1999年共20年的集合回报结果,提出了基于经验正交函数(EOF)和奇异值分解(SVD)的模式误差订正方法,并考察了上述订正方法对中国科学院大气物理研究所气候预测系统预测性能提高的季节差异及其稳定性,分析了订正方法对不同预报场的适用范围及其可能原因。结果表明:基于EOF/SVD的订正方法均可显著提高IAP-DCP II东亚地区各个季节降水预测的水平,在这20年的降水订正预报试验中,各季节80%左右的年份订正预报效果要优于模式原始结果。其中EOF订正方法在夏、冬季略优,回报与实测降水之间的距平相关系数(ACC)平均值分别从订正前的-0.07和0.16提高到0.25和0.44。而SVD方法则在春、秋季明显较好,ACC平均值分别从订正前的-0.09和-0.07提高到0.36和0.30。两种订正方法多年的ACC提高效果比较还表明SVD订正结果在各个季节ACC为负值的年份明显少于EOF方法。因此,对于实际预测而言,SVD订正方法效果更为稳定。ACC的方差比较也证明,SVD订正方法具有相对较好的稳定性。同时针对订正方法实际应用的需要,以夏季降水、地表气温和500 hPa高度场这3个变量为代表,文中还进一步分析了两种订正方法对夏季不同预报场的订正效果。结果表明,模态订正方法对不同预报场订正效果存在一定差异,对于中国科学院大气物理研究所的气候预测系统,上述方法并不适用于对夏季地表气温和500 hPa高度场预报结果的订正。
英文摘要:
      Using the 20-year hindcast results by the second generation of the Institute of Atmospheric Physics, Chinese Academy of Sciences(IAP) Dynamical Climate Prediction system (IAP-DCP II), the bias correction methods based on the Empirical Orthogonal Function (EOF) and the Singular Value Decomposition (SVD) have been proposed and validated for the prediction of precipitation anomalies, furthermore, the efficiency of these two correction methods for the different seasons has been investigated, together with their applicability to the different predictive variables. The verification results show that, both of the methods can significantly improve the prediction skill of precipitation anomalies over China by the IAP-DCPII with about 80% predictions improved obviously and the SVD-based method showing a higher stability compared with the another method. Generally, the EOF-based method is better in summer and winter, with the anomaly correction coefficients (ACC) between the hindcast and the observation improved from -0.07 to 0.25 in summer, and from 0.16 to 0.44 in winter, respectively, while the SVD-based method performs better in spring and autumn season, with the ACCs improved from -0.09 to 0.36 in spring, and from -0.07 to 0.30 in autumn, respectively. In addition, for the sake of real-time application, the summer surface air temperature and the 500 hPa geopotential height are also used to evaluate the efficiency of the two methods. However, for the IAP-DCP II, it is proved that both of the bias correction methods are not applicable for the prediction of the surface air temperature and the 500 hPa geopotential height. The some prime reasons are shown based on the results of two ideal experiments with and without cross-validation respectively.
HTML   查看全文   查看/发表评论  下载PDF阅读器
分享按钮