A directional coherence attribute for seismic interpretation

Yazeed Alaudah, Ghassan AlRegibarXiv 2017 · DOI · arXiv

Abstract

The coherence attribute is one of the most commonly used attributes in seismic interpretation. In this paper, we propose building on the recently introduced Generalized Tensor-based Coherence (GTC) attribute to make it directionally selective. This directional selectivity is achieved by selecting a directional Gaussian preprocessing kernel and applying a 3D rotational matrix to its covariance matrix. By weighing traces, or voxels, in the analysis cube by their relative proximity to the reference trace or voxel, this approach greatly enhances the clarity of the attribute. Furthermore, by making these weights directional, the proposed attribute gives interpreters greater freedom in exploring and understanding the seismic data. Various results from the Netherlands North Sea F3 block show that this approach greatly enhances the clarity of the coherence attribute and can highlight structures that are not visible using the traditional C3, or GTC coherence.

Introduction

In seismic interpretation, "coherence" is a vital attribute used to identify structural discontinuities such as faults, fractures, and stratigraphic channels. Traditional methods, like the widely used C3 coherence, analyze the similarity between neighboring seismic traces. However, standard algorithms often treat all neighboring data points equally, regardless of their orientation or distance from the point of interest. This "one-size-fits-all" approach can introduce noise and blur features that have a strong directional component. When geological structures are subtle or aligned in specific directions, they often get washed out in the final image, leaving interpreters in the dark.

To address this, researchers have developed a "Directional Coherence" attribute. By applying a rotatable 3D Gaussian kernel to tensor-based coherence calculations, this method reveals subtle geological features—like hidden channels and fractures—that remain invisible to traditional imaging techniques.

Methodology

The proposed method builds upon the Generalized Tensor-based Coherence (GTC) framework. The process involves four key steps:

  1. Tensor Unfolding: Instead of looking at traces individually, the 3D "analysis cube" of seismic data is treated as a third-order tensor. This tensor is "unfolded" along three modes (time/depth, inline, and crossline) to compute covariance matrices for each dimension.
  2. Gaussian Preprocessing: To reduce noise from distant voxels, a 3D Gaussian kernel is applied. This weighs the data points based on their proximity to the center of the analysis window.
  3. Directional Selectivity: By applying 3D rotation matrices to the covariance matrix of the Gaussian kernel, the weighting window can be "squashed" or "stretched" and rotated to any arbitrary angle. This allows the attribute to act like a polarized lens, tuning into specific orientations.
  4. Attribute Extraction: The final coherence value is derived from the ratio of the largest eigenvalue to the trace of the covariance matrix for each mode.
3D visualization of the directional Gaussian kernel used for trace weighting.

Results and Discussion

The study validates the method using the Netherlands North Sea F3 block seismic volume, focusing on a 1600ms time section.

The Power of Weighting and Orientation

Treating all voxels in an analysis window equally is often a liability. By prioritizing voxels closer to the reference point using the Gaussian kernel, the clarity of the attribute increases significantly. Furthermore, making the weighting kernel directional allows interpreters to "tune in" to specific features. If a fault is suspected at a certain angle, the kernel can be rotated to match, effectively filtering out misaligned noise.

Comparative Performance

The qualitative improvements are striking when compared to industry standards like C3 coherence. As shown in the comparison grid below, the directional GTC attribute clearly maps features that are virtually invisible in standard seismic amplitude and C3 coherence maps.

Comparison of seismic amplitude (a), C3 coherence (b), GTC coherence (c), and the proposed directional coherence at various angles: 40° (d), 75° (e), and 150° (f).

Key observations include:

  • Revealing Hidden Channels: The directional attribute successfully mapped the boundaries of a channel formation that was obscured in traditional views.
  • Sharper Fractures: When the kernel was rotated (e.g., to 150°), vertical faults and fractures became significantly sharper and more continuous.
  • Interpreter Freedom: Giving interpreters the ability to "steer" the attribute provides a much more flexible tool for exploring complex post-migration seismic data.

Limitations and Future Work

While powerful, the method currently requires manual exploration of rotation angles to find the best visualization. Future work could involve developing an "adaptive" directional coherence that automatically detects local dominant orientations. Additionally, the computational overhead of rotating kernels and performing tensor unfolding for every voxel in a massive 3D volume is higher than simpler cross-correlation methods, which may be a consideration for very large datasets.

Conclusion

The directional coherence attribute represents a significant advancement over traditional C3 and standard GTC methods. By leveraging tensor-based analysis and rotatable Gaussian kernels, it provides a clearer, more tunable view of the subsurface, empowering geoscientists to see between the lines of complex seismic data.

Summary updated

Abstract

The coherence attribute is one of the most commonly used attributes in seismic interpretation. In this paper, we propose building on the recently introduced Generalized Tensor-based Coherence (GTC) attribute to make it directionally selective. This directional selectivity is achieved by selecting a directional Gaussian preprocessing kernel and applying a 3D rotational matrix to its covariance matrix. By weighing traces, or voxels, in the analysis cube by their relative proximity to the reference trace or voxel, this approach greatly enhances the clarity of the attribute. Furthermore, by making these weights directional, the proposed attribute gives interpreters greater freedom in exploring and understanding the seismic data. Various results from the Netherlands North Sea F3 block show that this approach greatly enhances the clarity of the coherence attribute and can highlight structures that are not visible using the traditional C3, or GTC coherence. Presentation Date: Wednesday, September 27, 2017 Start Time: 3:30 PM Location: 340A Presentation Type: ORAL
BibTeX
@misc{alaudah2017directional,
  author = {Yazeed Alaudah and Ghassan AlRegib},
  title = {{A directional coherence attribute for seismic interpretation}},
  year = {2017},
  eprint = {1901.05307},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1901.05307},
}
Figure