集合经验模态分解在长江中下游梅雨变化多尺度分析中的应用

基于长江中下游流域5个梅雨监测站1961~2012年的日数据，利用集合经验模态分解（EEMD）方法，对研究期内梅雨时间序列进行多尺度的分析，探讨其在不同时间尺度上的振荡模态结构特征。结果表明：近50多年来，长江中下游梅雨变化呈现出显著的年际和年代际尺度振荡特征，在年际尺度上表现出准3 a和6 a的周期变化，而在年代际尺度上显示准13 a和24 a的周期变化；各分量方差〖JP2〗贡献率显示，年际振荡在梅雨长期变化中占据主导地位；自1961年以来，EEMD分解的梅雨长期变化趋势表现出先增加后减少的倒“U”型特征，其中1961~1985年呈上升趋势，1985~2012年呈下降趋势，尤其是在2000年之后的下降趋势最为明显。由此可以看出，EEMD能够有效地揭示梅雨长期序列在不同时间尺度上的变化规律，可用于诊断非线性、非平稳性信号变化的复杂性特征

APPLICATION OF ENSEMBLE EMPIRICAL MODE DECOMPOSITION METHOD IN MULTISCALE ANALYSIS OF MEIYU IN MIDDLE-LOWER REACHES OF YANGTZE RIVER

Ensemble empirical mode decomposition (EEMD) method has been developed suitable for non linear and non stationary signal analysis and it has been proved powerful tool in the long time series analysis. Compared with the wavelet analysis, though the scaling mode of the EEMD method is similar to wavelet transform, the signal resolutions in different frequency domains do not decrease by down sampling. In addition, compared with the EMD method widely used in climate change analysis, EEMD method also solves the problem of mode mixing and it is a good method of screening large scale circulation and non linear trend. With the EEMD method, the signal is decomposed into several intrinsic mode functions (IMFs) and the frequencies of IMFs are arranged in decrease order (high to low) after the EEMD processing. When the EEMD method applies to the time series of climate factors, the real climate change signal can be extracted. Specifically, the intrinsic time scale of climate change can be gotten with EEMD method, it is helpful to identify the trend of climate change. Furthermore, for non stationary time series, EEMD method can not only isolated inter annual and inter decade trend from several years of observation sequence, but also separated the general trend of climate change from the time series of climate observation. Therefore, it is helpful to explore the problem of global climate change. In this study, based on daily precipitation time series from 5 Meiyu monitoring stations in the middle lower reaches of the Yangtze River basin from 1961 to 2012, the multi scales characteristics of annual Meiyu were analyzed using EEMD method, and oscillation modal structure characteristics at different time scales were also investigated. We propose the EEMD method to decompose the Meiyu series in the middle lower reaches of the Yangtze River basin during 1961-2012 into several IMFs, then extract the information and get the characteristics of multi scales. Results indicated that in the last more than 50 years, Meiyu change in the middle and lower reaches of the Yangtze River have shown the obvious oscillation characteristics of inter annual and inter decadal scales. It exited 3 a and 6 a quasi periodic changes at inter annual scale, whereas 13 a and 24 a quasi periodic changed at decadal scale. The variance contribution rates of each IMF show that inter annual oscillation was dominant in longer term Meiyu change. The Meiyu in the middle and lower reaches of the Yangtze River overall presented inverted “U” shaped trend, that is to say rose first, and then decreased over time. The Meiyu series during 1961-1985 exhibited upward trends, while during 1985-2012 revealed a downward trend, the downward trend after 2000 was most obvious. Therefore, EEMD method can effectively reveal variation of long term Meiyu sequence at different time scales and can be used for the diagnosis of nonlinear and non stationary signal change of complexity