The eigenstructure-based coherence attribute is a type of efficient and mature tool for mapping geologic edges such as faults and/or channels in the 3D seismic interpretation. However, the eigenstructure-based coherence algorithm is sensitive to low signal-to-noise ratio seismic data, and the coherence results are affected by the dipping structures. Due to the large energy gap between the low- and high-frequency components, the low-frequency components play the principal role in coherence estimation. In contrast, the spectral variance balances the difference between the low- and high-frequency components at a fixed depth. The coherence estimation based on amplitude spectra avoids the effect of the time delays resulting from the dipping structures. Combining the spectral variance with the amplitude spectra avoids the effect of dipping structures and enhances the antinoise performance of the high-frequency components. First, we apply the short-time Fourier transform to obtain the time-frequency spectra of seismic data. Next, we compute the variance values of amplitude spectra. Then, we apply the fast Fourier transform to obtain the amplitude spectra of spectral variance. Finally, we calculate the eigenstructure coherence by using the amplitude spectra of spectral variance as the input. We apply the method to the theoretical models and practical seismic data. In the Marmousi velocity model, the coherence estimation using the amplitude spectra of the spectral variance as input shows more subtle discontinuities, especially in deeper layers. The results from field-data examples demonstrate that the proposed method is helpful for mapping faults and for improving the narrow channel edges’ resolution of interest. Therefore, the coherence algorithm based on the spectral variance analysis may be conducive to the seismic interpretation.
Bias in magnitude-squared coherence estimation due to misalignment