长江中下游汛期降水优化多因子组合客观定量化预测研究

A Study of the Objective and Quantifiable Forecasting Based on Optimal Factors Combinations in Precipitation in the Middle and Lower Reaches of the Yangtze River in Summer

针对长江中下游区域汛期降水模式预报误差,利用历史资料的有用信息订正模式预报误差。根据国家气候中心预测室提供的74项环流特征量和美国国家海洋和大气管理局发布的40项气候指数,确定降水预报误差场的相似年份,利用相似年的降水误差场对模式结果进行相似误差订正,有别于传统的模式系统误差订正。研究发现,相似误差订正效果明显优于系统误差订正。在此基础上,对前期因子进行筛选,确定相似误差场的多因子优化组合,并结合加权集合平均的方法,提出了基于优化多因子组合的客观定量化预测（OQF）技术。2003~2009年7年的独立样本回报结果表明,基于优化多因子组合的OQF方案具有较高的预报技巧,平均距平相关系数为0.43。最后对历史资料中误差场间的距平相关系数进行分析,发现确定误差场相似年的关键技术还有待于进一步完善,并指出了改进方法的可能途径。

A new approach to improve precipitation in the Yangtze River in summer by using the useful historical data is suggested.The outcomes of the model can be improved by determining the years in which there are similar prediction error fields of precipitation based on 74 circulation characteristic indexes of Weather Diagnostic Forecasting Room of National Climate Center and 40 climate indexes of NOAA,and then the fields are integrated to revise the output errors of the model.The results show that the revised errors using the above method are better than system ones.On the basis of the above conclusions,the objective and quantifiable forecasting（OQF） technique can be established.The steps are selecting pre-factors,determining optimal factor combinations（OFC） of the analogy forecasting error fields,and using the method of weighted average integration.It shows that the technique of OQF based on OFC has a better prediction skill to the regression results of independent samples between 2003 and 2009.Finally,the anomaly correlation coefficient（ACC） of historical data in the error field is analyzed.It is found that the determining of the analogy error field in different years can be improved more,and the possible way to improve is to be discussed.