摘要
由于影响我国盛夏(7—8月)降水异常的主要因子前期(或从冬季到夏季)不同的月际演变特征对其后降水异常的影响不同,所以针对影响因子的月际"持续异常"和"转折变化"对盛夏降水异常的预测方法也应该是不同的。本文以国家气候中心整理的78项月环流特征量指数和CPC/NOAA的30项指数为因子,设计了能够反映影响因子的月指数特征量"持续异常"和"转折变化"的两套影响因子库,用"滑动相关一逐步回归一集合分析"方法分别建立预测模型,改进预测方案,这样建立的统计预测模型重点考虑了前期影响因子不同的变化特征。用距平符号一致率(PSS)与空间距平相关系数(ACC)为指标,对比分析不同因子处理方案对我国160站盛夏(7—8月)降水预测效果的影响,结果显示:两种因子选择方案均具有一定的预测能力;相比之下,"转折变化"因子选取方案的预测效果更好;用单站交叉建模序列与实测序列的相关系数为指标,挑选其中相关系数更高的因子选取方案结果作为集合方案预测结果,集合方案的预测效果有进一步提高的潜力。
Because the main earlier circulation indices with different monthly variation features have differ- ent influences on midsummer （July--August） precipitation in China, different methods should be used to predict the midsummer prediction anomalies with respect to the monthly continuity and transition features of the impact factors. The 78 monthly circulation indices from National Climate Centre and the 30 indices from CPC/NOAA are employed to construct two factor schemes to describe the continuity and transition of circulation anomaly characteristics respectively. With the above different schemes, the ＂Running-Correla- tion, Stepwise-Regression, Ensemble-Analysis＂ method is used to set up two models to predict the mid- summer precipitation of 160 stations over China. Therefore, different variation features of earlier circula- tion indices are used in our statistic prediction models and different effects of the above two schemes are well shown. The result indicates that both of the facto^treating schemes have certain predictive abilities but the transi tion one shows better effect. Meanwhile, for single station, the ensemble of the indices with its cross-modeling se- ries closely correlated to the observation time series holds more potential prediction ability.
出处
《气象》
CSCD
北大核心
2013年第9期1176-1181,共6页
Meteorological Monthly
基金
国家自然科学基金项目(41065006)
国家体育总局体育科学研究所基本科研业务费(基本10-05)共同资助
关键词
持续
转折
预测效果
7-8月降水
前期环流特征量
continuity, transition, prediction, July-- August precipitation, earlier circulation characteris tics
