Advances on Empirical Mode Decomposition-based Time-Frequency Analysis Methods in Hydrocarbon Detection

Abstract:

Empirical mode decomposition (EMD), which is a data-driven adaptive decomposition method and is not limited by time-frequency uncertainty spreading, is proved to be more suitable for seismic signals which are nonlinear and non-stationary. Compared with other Fourier-based and wavelet-based time-frequency methods, EMD-based time-frequency methods have higher temporal and spatial resolution and yield hydrocarbon interpretations with more statistical significance. Empirical mode decomposition algorithm has now evolved from EMD to Ensemble EMD (EEMD) to Complete Ensemble EMD (CEEMD). Even though EMD-based time-frequency methods offer many promising features for analyzing and processing geophysical data, there are some limitations or defects in EMD-based time-frequency methods. This presentation will present a comparative study on hydrocarbon detection using seven EMD-based time-frequency analysis methods, which include: (1) first, EMD combined with Hilbert transform (HT) as a time-frequency analysis method is used for hydrocarbon detection; and (2) second, Normalized Hilbert transform (NHT) and HU Methods respectively combined with HT as improved time-frequency analysis methods are applied for hydrocarbon detection; and (3) three, EMD combined with Teager-Kaiser energy (EMD/TK) is investigated for hydrocarbon detection; and (4) four, EMD combined with wavelet transform (EMDWave) as a seismic attenuation estimation method is comparatively studied; and (5) EEMD- and CEEMD- based time-frequency analysis methods used as highlight volumes technology are studied. The differences between these methods in hydrocarbon detection will be discussed. The question of getting a meaningful instantaneous frequency by HT and mode-mixing issues in EMD will be analysed. The work was supported by NSFC under grant Nos. 41430323, 41404102 and 41274128.