Methods to compute fault images, extract fault surfaces, and estimate fault throws from 3D seismic images

Abstract

Fault interpretation enhances our understanding of complex geologic structures and stratigraphy apparent in 3D seismic images. Common steps in this interpretation include image processing to highlight faults, the construction of fault surfaces, and estimation of fault throws. Although all three of these steps have been automated to some extent by others, fault interpretation today typically requires significant manual effort, suggesting that further improvements in automatic methods are feasible and worthwhile. I first used an efficient algorithm to compute images of fault likelihoods, strikes, and dips from a 3D seismic image. From these three fault images, I then automatically extracted fault surfaces as meshes of quadrilaterals that coincide with ridges of fault likelihood. A quadrilateral mesh is a simple data structure alongside which one can easily gather samples of the 3D seismic image. I automatically estimated fault throws by minimizing differences in values of samples gathered from opposite sides of a fault, while constraining the variation of throw within a fault surface. I tested the fidelity of estimated fault throws by using them to undo faulting. After unfaulting, reflectors in 3D seismic images were more continuous than those in the original 3D seismic image. In one example, this unfaulting test supported the observation that some extracted fault surfaces have unusual conical shapes.

Automatic fault localization using the generalized Earth Mover's distance

“…Considerable effort has been invested into discovering automatic methods for fault surface extraction; see, for example, the recent article [3] and references therein. Some example approaches include texture classification approaches [4], coherence-based methods [5], filtering-based methods [6], and greedy optimization algorithms to extract dominant paths in seismic images [7].…”

Automated fault detection without seismic processing

“…This iterative refinement is expensive in terms of human costs (picking out features/adjusting the velocity model) and computational costs (many repeated wave propagations and attributes computations). Most prior work focused on identifying features in already migrated images (Hale, 2012;Hale, 2013;Guillen, 2015;Addison, 2016;Bougher and Hermann, 2016). The literature is filled with refinements to this workflow, but ultimately, it remains largely the same.…”

A parametric fault displacement model to introduce kinematic control into modeling faults from sparse data

“…Displacement is usually quantified on the fault surface using displacement vectors (Liang et al, 2010;Hale, 2013), displacement maps (Freeman et al, 2010;Tvedt et al, 2013Tvedt et al, , 2016 or throw profiles along strike or down-dip (Willemse, 1997;Peacock, 2002;Nixon et al, 2011;Jackson and Rotevatn, 2013;Magee et al, 2015). However, the near-field displacement is not only expressed by the slip along the fault surface (Figure 1.b), but also by the displacement in the surrounding stratigraphy (Figure 1.c).…”

“…Several methods also exist to estimate fault geometry from seismic images based on semblance (e.g., Machado et al, 2016;Qi et al, 2017;Wu and Zhu, 2017). Image warping methods can also estimate fault throw (Hale, 2013), but may be limited by the seismic resolution and the quality of the seismic image. In all cases, the kinematic consistency of the resulting structural models then needs to be tested using rules such as displacement distance characteristics (Chapman and Meneilly, 1990), visual inspection of displacement profiles and maps (Freeman et al, 1990), by empirically assessing the strain in rocks adjacent to faults (Freeman et al, 2010) or by using restoration (Maerten and Maerten, 2015).…”