SEISMIC TIME-FREQUENCY ANALYSIS USING AN IMPROVED EMPIRICAL MODE DECOMPOSITION ALGORITHM

Among the time-frequency analysis approaches, the EMD-based approaches have been proven to show higher spectral-spatial resolution than the traditional approaches. However, the mode mixing problem always exists in these approaches which will affect the subsequent interpretation performance. In this paper, we apply a novel improved complete ensemble empirical mode decomposition (ICEEMD) technique to time-frequency analysis of seismic data. The ICEEMD approach can help decompose a 1D non-stationary signal into intrinsic mode functions with less noise and more physical meaning, and result in a higher frequency resolution in the time-frequency maps. The application of the algorithm to 1D seismic signal can help obtain a more meaningful analysis regarding the non-stationary components. Its application to 2D and 3D seismic data has the potential to enable a better geological and geophysical interpretation. We use a 1D real seismic trace, a 2D seismic section and a 3D seismic cube to show the superior performance of the proposed approach.