3D seismic image processing for unconformities

Abstract

In seismic images, an unconformity can be first identified by reflector terminations (i.e., truncation, toplap, onlap, or downlap), and then it can be traced downdip to its corresponding correlative conformity, or updip to a parallel unconformity; for example, in topsets. Unconformity detection is a significant aspect of seismic stratigraphic interpretation, but most automatic methods work only in 2D and can only detect angular unconformities with reflector terminations. Moreover, unconformities pose challenges for automatic techniques used in seismic interpretation. First, it is difficult to accurately estimate normal vectors or slopes of seismic reflectors at an unconformity with multioriented structures due to reflector terminations. Second, seismic flattening methods cannot correctly flatten reflectors at unconformities that represent hiatuses or geologic age gaps. We have developed a 3D unconformity attribute computed from a seismic amplitude image to detect unconformities by highlighting the angular unconformities and corresponding parallel unconformities or correlative conformities. These detected unconformity surfaces were further used as constraints for a structure-tensor method to more accurately estimate seismic normal vectors at unconformities. Finally, using detected unconformities as constraints and more accurate normal vectors, we could better flatten seismic images with unconformities.

Simultaneous multiple well-seismic ties using flattened synthetic and real seismograms

“…However, our correlation of seismic traces and synthetic traces would likely yield inaccurate results in these cases. To improve seismic trace correlation, an alternative strategy could then be used to better exploit imaging information by flattening the whole seismic image (Lomask et al, 2006;Fomel, 2010;Wu and Zhong, 2012;Wu and Hale, 2015a, 2015b, instead of flattening only the seismograms extracted at well locations. To improve well correlations, one could also replace our DTW-based synthetic trace correlation by expert-based manual correlation or correlation using various logs or rules (Lallier et al, , 2016.…”

Automatically interpreting all faults, unconformities, and horizons from 3D seismic images

“…For 3D seismic images, this method processes inline and crossline slices separately to compute an unconformity probability volume. Wu and Hale (2015b) propose 3D image processing methods to (1) compute an unconformity likelihood image that highlights the termination areas and the corresponding parallel unconformities or correlative conformities, (2) extract unconformity surfaces from the unconformity likelihood image, and, finally, (3) use the unconformity surfaces as constraints to accurately estimate seismic normal vectors at unconformities and to compute a flattened image with vertical gaps corresponding to the unconformities.…”

“…However, these methods are unable to match horizons across faults unless additional information such as fault slips (Luo and Hale, 2013) and control points across faults (Wu and Hale, 2015c) are provided. Also, these methods cannot adequately deal with horizons terminated at unconformities unless the unconformity surfaces are provided (Wu and Zhong, 2012;Wu and Hale, 2015b).…”

Seismic feature extraction using steiner tree methods

“…Both kinds of features contain valuable spatio-temporal information and accurate localization of such features is of considerable interest. A host of automatic approaches for extraction of both types of features have been proposed in the literature, and we highlight a few representative works (for a more complete list of references, see the recent papers [6,7]). In the context of fault extraction, these include texture classification approaches [8], coherence-based methods [9], and evolutionary computation-based search heuristics [10].…”

“…In the context of fault extraction, these include texture classification approaches [8], coherence-based methods [9], and evolutionary computation-based search heuristics [10]. In the context of unconformity extraction, these include edge-detection based methods [11] and structure-tensor fields [7]. Most of these methods seem to suffer from a low tolerance to noise, high computational complexity, or both.…”