By WALTER E. JOHNSON
The Geophysical Corner is a regular column in the EXPLORER, produced cooperatively by the AAPG Geophysical Integration and SEG Interpretation committees, and edited by R. Randy Ray. This month’s column is titled “The Power of Multiple Seismic Attributes in Revealing Geologic Understanding.”
Seismic Has Its Multiple Attributes
Editor’s note: Walter Johnson is a consulting geophysicist, Denver; Richard O. Louden is a consulting geologist affiliated with Thomasson Partner Associates, Denver; Donald Lehman, former exploration manager for Miller Energy, is now with VuCom Data Services in Kalamazoo, Mich.; and Duncan Edwards is an independent consulting geophysicist in Rutherfordton, N.C.
Click on image to enlarge.
Analysis of 3-D seismic data with several types of seismic attributes can reveal geologic factors that control the location of productive algal mound reservoirs in the Paradox Basin.
Routine seismic mapping of the producing interval did not reveal the presence of a regional strike-slip fault that is clearly shown using attribute analysis. This previously unknown strike-slip fault controls the local stratigraphy and extends for over 30 miles.
Although each attribute when used alone has some level of ambiguity, it is important to note that when a number of attributes with different mathematical algorithms yield similar results, the reliability of the geologic interpretation is enhanced.
Several commercially available seismic attributes, along with attributes generally incorporated within most workstations, collectively have had a significant impact on the understanding of how and where the algal mound reservoirs form, and indicate that their stratigraphic development is often not a random act. The purpose of this article is to demonstrate that the small cost of time and money required to perform attribute analysis is far outweighed by the increased understanding of the geologic dynamics of an area.
The ability to map more geologic detail will ultimately result in reduced risk of drilling dry holes.
Miller Energy, of Kalamazoo, Mich., and its partners completed the Miller Horse Canyon 1-10 in 1998 for an initial potential of 960 barrels of oil per day and 940 MCFG/D from a depth of 5,850 feet from an algal mound reservoir. The well was featured in the 1999 EXPLORER.
Figure 1 shows the location of the Paradox Basin in the southwestern United States, where shelf carbonate buildups within the Pennsylvanian basin have produced oil and gas since the 1950s.
Generally these carbonate buildups are the result of algal debris from Ivanovia (these Ivanovia algal skeletons are much the same size as cereal corn flakes). In the 1980s it became generally known that there could be a seismic expression resulting from these carbonate buildups, but the 2-D seismic data that existed then added little to the understanding of the genesis of these features.
Figure 2, a geologic cross section through the discovery well, is a good reference for the various geologic strata that will be mentioned throughout this article. The well was completed in an Upper Ismay carbonate buildup; however, the test also encountered a deeper Desert Creek carbonate buildup that is further revealed with attribute analysis.
A seismic tuning effect owing to thickness changes in the Hovenweep shale creates a mappable high amplitude that indicates where the shale is extra thick, as shown in
Figure 3. This thick resulted in a positive “island” that existed during the time of deposition of the Upper Ismay. The island is outlined in black and will be shown on subsequent displays.
Note in Figure 2 how the Hovenweep thick is located to the left of the discovery well.
The isochron thickness map of the Upper Ismay (from the top of the Hovenweep to the top of the Upper Ismay) shown in Figure 4 demonstrates that there is an atoll shape of increased thickness of Upper Ismay surrounding the Hovenweep island.
A series of different seismic attribute displays are shown in Figures 5 through 10 (5, 6, 7, 8, 9, 10), which reveals a linear fault zone that affected the formation of the Hovenweep thick and thus the locations for mound development. The attribute analyses include several types of methods, including multivariate analysis, edge detection, spectral decomposition, wavelet cross-correlation and wavelet classification.
The amplitude map of the Desert Creek horizon demonstrates a distinctive NW-SE trend (Figure 5). Originally it was believed that this NW-SE trending amplitude anomaly was related only to the Desert Creek carbonate buildups.
An attribute process that could be characterized as “wavelet classification” was run on the shallower Upper Ismay seismic horizon - and again, the NW-SE trend was observed. This wavelet classification technique divides the horizon wiggle into 12 characteristic wavelets, and then makes a color map indicating areas of similar wavelets. Small changes in the wavelet character often reveal significant geologic features like porosity and thickness changes (Figure 6). Field analyses from other fields on trend with the fault demonstrate that the “atoll model” is not unique to this discovery.
Another attribute, which can be classified as multivariate attribute analysis, combines the properties of autocorrelation, amplitude and phase of seismic traces. The output from the multivariate attribute analysis of the Upper Ismay is shown in Figure 7, and also demonstrates this same strong NW-SE trend.
It is important to note that both the wavelet classification map and the multivariate attribute analysis have entirely different mathematical algorithms, and yet both show the same NW-SE lineament. One therefore has to give more credibility to the lineament being the result of a geologic phenomena rather than some “black box” artifact.
The multivariate attribute analysis creates a cube output that allows one to evaluate a lineament (fault?) spatially throughout a seismic data volume.
Edge detection technology has been used in the seismic industry for several years, and is designed to compare each trace of the data with its neighbors in order to map dissimilarities. Figure 8 (previous page) is the edge detection horizon slice of the Upper Ismay horizon. A fault was interpreted through the linear discontinuity on the horizon slice. This package also has the advantage that it creates a cube of data output that can be loaded on the workstation to be interpreted at different time or horizon slices.
Figure 9 is another coherency slice that is approximately 1,500 feet below the Upper Ismay, and again this same lineament can be mapped. Two-D seismic has been shot in the area for decades, and this very small-offset fault (or perhaps zone of weakness) has not been observed. In fact it was not even observed on the 3-D data until output was displayed and interpreted using the various attribute packages.
Figure 10 is an amplitude map of the top of the Akah salt, and it is believed that the dim shown in green and red is the result of salt dissolution.
Note how the amplitude dim conforms with the overlying Hovenweep thick. We postulate that the fault may have been a conduit to cause dissolution of the Akah salt. A horizon slice slightly above the Akah salt has a very distinctive linear pattern that also could be the result of differential salt dissolution.
As another attribute comparison, spectral decomposition is a very precise frequency measurement tool that can also be viewed as a wavelet measurement tool. Figure 11 is the result of spectral decomposition over the Upper Ismay horizon, and again one can see the fault, Hovenweep “island,” and the flanking atolls.
Many workstations have add-on packages that allow some trace processing. A horizon windowed seismic wiggle from the area of the producing well was cross-correlated with the entire 3-D volume, and those areas that had similar correlation values were mapped on the workstation to produce Figure 12.
Note the similarity between Figures 12 and 6 . Both the cross-correlation and wiggle classification mapping techniques indicate strong potential for new drill site locations on the south side of the Hovenweep “island.”
The enhanced geologic understanding gained from the various attribute analyses far outweighs the small costs and interpretation time required trying them.
The similar patterns that emerge by running many different techniques reinforces the interpreter’s confidence when mapping subtle geologic variations. This improved geologic understanding leads to reduced risk in both development and exploration projects.
Often seismic interpreters are discouraged from mapping anything but the primary producing zone. Our experience on this stratigraphic play shows that attribute analysis of the section surrounding the producing reservoir leads to a better understanding of the porosity development in the productive zones, as well as showing how to project the extension of the trend for further prospecting.