Seismic multi-attribute approach using visual saliency for subtle fault visualization

Abstract

This study improves a collection of attributes to detect subtle faults in three dimensional data obtained from the Krishna-Godavari (KG) basin, with results displayed on synthetic and real datasets. Seismic attributes, for instance, curvature and coherence, are often used to delineate discontinuities, such as faults and fractures where hydrocarbons may have been trapped. These attributes have their advantages subjective to the seismic data. In this paper, we propose a multi-attribute framework for identifying subtle faults inside seismic volumes. Curvature attribute is a powerful and popular technique to deal with these faults. The faulted horizon is fitted on the quadratic surface using the least-square method, and the most positive and most-negative curvature attributes are calculated, which are further used in saliency map calculations. Several signal processing techniques, such as Hough transform and ant tracking, have been used to delineate faults. Here, we have proposed a novel signal processing approach based on energy variations known as top-down saliency on the curvature attributes using 3D-FFT local spectra and multi-dimensional plane projections. To analyze the directional nature of seismic data, the directional center-surround technique is employed for visual attention. Furthermore, the log-Gabor filter and image erosion are applied to the saliency-rendered seismic volume to highlight the oriented amplitude discontinuities at faults. Most of the time, these discontinuities may not be very prominent to find the subtle faults and other trace-to-trace hidden geological features in three-dimensional seismic data. In our work, calculated attributes assist us in mapping these changes, because they are all differently sensitive to the faults and fractures in unique ways. Experimental results on real field seismic data from the Krishna-Godavari basin prove that the proposed algorithm is effective and efficient in tracking subtle and minor faults, better than previous works.

KEYWORDS:

• Curvature attribute
• fault detection
• log-Gabor filter
• visual saliency

Acknowledgements

We would like to thank the anonymous reviewers for reviewing our paper. The work is an outcome of research work sponsored by Oil & Natural Gas Corporation (ONGC) Limited, Dehradun, a Government of India enterprise. We’d also like to thank the organization for its assistance and for sharing the offshore Krishna-Godavari basin 3D seismic dataset as well as providing permission for publication. All the codes for the published work are implemented on the MATLAB platform.

Data availability statement

The data that support the findings of the study are confidential and proprietary to ONGC, India and hence can not be shared.

Disclosure statement

No potential conflict of interest was reported by the author(s).