Time-frequency analysis: Beyond waveletsZhaohua Wu Absract As is well known, one of the major drawbacks of Wavelets Decomposition (WD) is the low-degree relevance of the components of data to its driven physical processes.The recently developed Empirical Mode Decomposition (EMD) for nonlinear non-stationary time series has demonstrated its great capability in isolating the physical events but it often suffers the scale (frequency) mixing problem. To overcome drawbacks of WD and EMD, the Ensemble Empirical Mode Decomposition (EEMD) method is developed. This new method consists of an ensemble of decompositions of data with added white noise, and then treats the resultant means of the corresponding components as the final true result. The effect of the added white noise is to present a uniform reference frame in the time-frequency (time-scale) space and to provide natural filter windows for the signals of comparable scale to collate in one component, essentially eliminating the scale (frequency) mixing problem in the EMD while keeping capability of isolating physical events. Climate data are used to demonstrate the advantages of the EEMD over both the EMD and Wavelets Decomposition |