A novel fluid identification method based on a high-precision spectral decomposition method

Abstract

Time-frequency decomposition technology is an effective tool to analyze non-stationary signals. Improving resolution of spectral decomposition techniques is important to extract more useful information from the received signal. Wigner-Ville distribution (WVD) has been widely applied in seismic signal analysis, it can better analyzes seismic signals due to many excellent mathematical properties, but this method has a drawback that cross terms interference exists in the analyzing of multi-component signals, which severely limits its application. The combination of the complex domain matching-pursuit (CDMP) with this approach effectively solves this problem. However, the conventional CDMP-WVD does not take the influence of the scale parameter on the Morlet wavelet waveform into account, which reduces the time-frequency resolution of CDMP-WVD. Therefore, in order to correct the defect that the atomic waveforms change only with the frequency parameter, we propose an improved spectral decomposition method ICDMP-WVD that considers the scale parameter. In this study, we first analyze influences of the scale parameter on Morlet wavelet waveform and make the scale parameter as search parameter, that improves the computational efficiency and time-frequency resolution of the traditional CDMP-WVD method. Accordingly, the seismic dispersion-dependent attributes are calculated via combing the improved CDMP-WVD algorithm and the frequency-dependent AVO inversion. We adopt a two-step frequency-dependent AVO inversion method to improve the stability of the conventional frequency-dependent AVO inversion. Theoretical data and real data application show that the approach in this study can identify gas reservoirs efficiently and accurately.