Multidirectional eigenvalue-based coherence attribute for discontinuity detection

The conventional coherence attribute is typically applied to migrated full-stacked seismic data volumes to detect geological discontinuities. Recently, multispectral, multiazimuth, and multioffset coherence attributes have been proposed and implemented with different seismic data volumes of specific frequencies, azimuths, and offsets to enhance discontinuities. Generally, geological anomalies, such as faults and channels, will be better illuminated by a perpendicular rather than a parallel direction for computation. Therefore, we propose a multidirectional eigenvalue-based coherence attribute by establishing multiple covariance matrices along certain different directions on a single post-stack volume. We adopt two methods to compute multidirectional coherence attribute. One is to compute multiple coherence volumes in different directions and to define the minimum as the final multidirectional coherence. This method is time-consuming, but could provide partial and overall discontinuity simultaneously. The other method obtains one coherence volume by summing covariance matrices in different directions, which is computationally efficient, but only provides overall discontinuity. The performance of 3D physical model and field data volumes demonstrates that multidirectional coherence can highlight subtle geologic structures with a higher resolution than conventional coherence. This suggests that multidirectional coherence attribute may serve as an effective tool for detecting the distribution of geologic discontinuities in seismic interpretation.