1921～2010年天津气温和降水量序列的多尺度分析

Multiscale Examination on the Temperature and Precipitation Series in Tianjin during 1921-2010

基于集成经验模态分解（EEMD）方法，对1921-2010年天津年平均气温和降水量序列进行了多尺度分析。并结合功率谱对年平均气温和年降水量及其本征模态函数（IMF）分量进行周期分析。结果表明：天津年平均气温的变化主要是由第1、第2本征模态分量和趋势项构成，即准5a和2～3a的振荡与“先降后升”的长期趋势变化起主要作用。而第4、第5本征模态分量则反映出天津近90年来气温年代际尺度的冷暖变化，它们对1920年代至1940年代的暖期以及1950年代至1970年代的冷期有重要贡献。降水量的变化主要由第1、第2本征模态分量构成，即4．5a、准9a和2～3a的振荡起主要作用。与气温序列相比，降水序列中年代际尺度的变化和长期趋势的贡献明显要小得多，但也反映了1980年代以后降水减少的趋势。

A multiscale examination of temperature and precipitation in Tianjin during 1921-2010 was performed using the EMD （Empirical Mode Decomposition） method. Moreover, power spectrum analysis was performed on the IMFs （Intrinsic Mode Functions） derived from EMD. This analysis indicates that the variations of the annual mean temperature are mainly characterized by high-frequency oscillations, with 2-3-year periods and quasi-5-year periods detected in the first two IMFs, and a long-term V-shaped variation in the trend component. In addition, interdecadal variations were detected in the fourth and fifth IMFs. The precipitation series are also mainly characterized by high-frequency oscillations, with periods of 2-3, 4.5, and 9 years detected in the first two IMFs. It is concluded that the interdecadal variations and the long-term trend are more apparent in the temperature series, and therefore have a much smaller contribution to precipitation variation.