气候场主成分回归预测模型在海南岛汛期降水预测中的研究应用  

PC Stepwise Regression Forecasting Model of Climate Field of Hainan Island Flood Period Rainfall

作　　者：易灵伟 吴胜安[1] 张亚杰[1] Yi Lingwei;Wu Shengan;Zhang Yajie(Hainan Climate Center,Hainan Meteorological Bureau,Haikou 570203,China)

机构地区：[1]海南省气象局气候中心

出　　处：《海南大学学报（自然科学版）》2019年第2期165-171,共7页Natural Science Journal of Hainan University

基　　金：海南省青年基金(HNQXQN201704)

摘　　要：计算与分析了海南岛汛期降水量场的主成分分布及分型特征,选取北半球500hPa高度场、北半球海平面气压场、太平洋海温场3个影响海南岛汛期降水的高影响因子,提取其前25个主成分作为预测因子,通过相关性筛选及逐步回归检验方法,确立降水场与筛选预测因子关系,建立相关高影响因子对主成分的预测方程,通过对高相关因子所构成气候场的主成分预测,最终实现海南岛汛期降水量场的预测.将2016年海南岛汛期降水实况、现有多模式集合预报(MODES)汛期降水预测值及回归预测模型降水预测值进行比对,发现该模型对海南岛汛期降水及极值分布有较好的预测模拟能力,在实际业务中有一定的应用价值.In the report,the distribution and types of the principal components of the flood period rainfall in Hainan Island were calculated and analyzed,the 500 hPa height,the sea surface temperature of Pacific Ocean, and the sea-level pressure in the Northern Hemisphere were selected as predictors,25 principal components were used as the predictors,the correlation screening and double test stepwise regression were used to construct the relation between the precipitation field and screening predictors,the predictive equation of correlation high impact factor to principal component was established. By the principal component prediction of high impact factors, the prediction of the flood period precipitation in Hainan Island in 2016 was performed. The results of comparing the predict result of MODES system and the observed precipitation of 2016 indicated that the model is capable to predict the precipitation and its distribution in the flood period in Hainan Island.

关 键 词：海南岛汛期降水 主成分分析 双重检验逐步回归 气候统计预测 

分 类 号：P468.1[天文地球—大气科学及气象学]