Highlighting Discontinuities with Variational Mode Decomposition Based Coherence

Bin Lyu,Fangyu Li,Jie Qi,Tao Zhao,Kurt J. Marfurt

DOI: https://doi.org/10.1190/segam2018-2997710.1

2018-01-01

Abstract:The coherence attribute is a powerful tool to delineate structural boundaries and discontinuity features, such as faults, channels, and karst collapse. Multispectral coherence is calculated from different band-pass filtered data volumes, providing improved images of the edges and boundaries over conventional broadband coherence. However, the lower signal-to-noise ratio high-frequency components of the band-pass filtered results often give rise to artifacts that can contaminate the result. The solution to this problem is to only use those spectral components that contain useful information. An alternative solution is to somehow separate signal from noise data components. To address this issue, we evaluate a coherence calculation workflow using variational mode decomposition (VMD) components as input. VMD is a data-driven signal decomposition method, which adaptively decomposes a signal into a series of band-limited intrinsic mode functions exhibiting greater lateral continuity than spectral components. In general, noise is contained in the residual of the data, that part not represented by the leading components. We first decompose the broadband seismic data into several intrinsic mode functions, each with its own central frequency. We then compute energy ratio coherence from each VMD component. We finally combine these results together, providing the VMD based coherence. We evaluate this VMD-based coherence method with a 3D seismic survey acquired over the Red Fork Formation of the Anadarko Basin, Oklahoma. Presentation Date: Tuesday, October 16, 2018 Start Time: 1:50:00 PM Location: Poster Station 14 Presentation Type: Poster