长江流域夏季降水及其与海温的滞后相关

摘要

基于月平均台站降水资料分析了中国东部降水的气候特征。由于其降水分布在空间和时间上的复杂性，为便于研究其与海温的关联，特别是海温对降水的影响，根据台站间降水变化在季节变化和年际变化以及与海温相关的一致性，将中国东部区域分成四个子区域。以长江流域的降水为代表，分析了海温对降水的超前相关。给出了主要影响海区和它们的超前时间。综合各影响海区海温和超前时间建立的回归方分析表明：回归降水和观测降水的相关系数大于0.85。回归的平方根误差是其标准差的60％和平均雨量的15％。另外还分析了海温与降水相关与环流变化的一致性。%Some features associated with Eastern China Precipitation (ECP), in terms of mean climatology, seasonal cycle, interannual variability are studied based on monthly rainfall data. The rainfall behavior over Eastern China has fine spatial structure in the seasonal variation and interannual variability. The revealed characteristics of ECP motivate us dividing Eastern China into four sub-regions to quantify significant lag-correlations of the rainfalls with global sea surface temperatures (SSTs) and to study the ocean's predominant role in forcing the eastern China summer monsoon rainfalls. Lagged correlations between the mid-eastern China summer monsoon rainfalls (MECSMRs) and the global SSTs, with SST leading to rainfall, are investigated. The most important key SST regions and leading times, in which SSTs are highly correlated with the MECSMRs, are selected. Part of the results confirms previous studies that show links between the MECSMRs and SSTs in the eastern equatorial Pacific associated with the El Nino - Southern Oscillation (ENSO) phenomenon. Other findings include the high lag correlations between the MECSMRs and the SSTs in the high and middle latitude Pacific Ocean and the Indian Ocean, even the SSTs over the Atlantic Ocean, with SST leading-time up to 4 years. Based on the selected SST regions, regression equations are developed by using the SSTs in these regions in respective leading time. The correlation coefficient between the observed rainfalls and regressed rainfalls is over 0.85. The root mean square error (RMSE) for regressed rainfall is around 65% of the standard deviation and about 15% of the mean rainfall. The regression equation has also been evaluated in a forecasting mode by using independent data. Discussion on the consistence of the SST-rainfall correlation with circulation field is also presented