Professional Documents
Culture Documents
net/publication/301509491
CITATIONS READS
5 2,798
4 authors, including:
John C. Davis
Heinemann Oil GmbH
92 PUBLICATIONS 8,859 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Christoph Eichkitz on 22 November 2017.
Abstract
We have used poststack seismic attributes to describe the fracture network of the naturally fractured
Tensleep Formation at Teapot Dome, Wyoming, USA. The attributes include coherence, coherence based on
spectral decomposed seismic data, attributes based on curvature, and textural attributes based on the gray-level
co-occurrence matrix (GLCM). Results were compared with image log interpretations of four wells. Seismic
attribute analysis allowed determination of strikes and dips as well as the intensity of fractures. The GLCM-
based attributes proved especially valuable for building a discrete fracture network.
1
MOL Group, Budapest, Hungary. E-mail: sschneider@mol.hu.
2
Geo5 GmbH, Leoben, Austria. E-mail: christoph.eichkitz@geo-5.at; marcellus.schreilechner@geo-5.at.
3
Heinemann Oil GmbH, Leoben, Austria. E-mail: jcdbaldwin@gmail.com.
Manuscript received by the Editor 10 December 2015; revised manuscript received 5 February 2016; published online 18 April 2016. This paper
appears in Interpretation, Vol. 4, No. 2 (May 2016); p. T273–T284, 11 FIGS.
http://dx.doi.org/10.1190/INT-2015-0210.1. © 2016 Society of Exploration Geophysicists and American Association of Petroleum Geologists. All rights reserved.
tures for the Tensleep Formation based mainly on horizon), and Tensleep Formation (yellow horizon).
seismic attributes and later on use this information The top of the Fremont Canyon Sandstone (orange hori-
for a DFN. The Rocky Mountain Oilfield Testing Center zon) was assigned to a reflector based on interpreta-
(RMOTC) provides public access to an extensive collec- tions by Wilson et al. (2013b). In step 2 (Figure 4b),
tion of data about Teapot Dome for scientific and tech- these five horizons were picked manually on every fifth
nical research. Three-dimensional seismic data, well inline and crossline. The 3D autotracking was then used
headers, well deviation surveys, formation tops, petro- to create continuous horizons on every line, and the fi-
physical logs of 27 wells, and GIS coordinates were nal tracked horizons served as input to generate 3D sur-
used in this study. faces. Faults were interpreted by picking every fifth
To establish a DFN, a workflow of seven steps has inline or crossline.
been generated (Figure 4). The first five steps consist Structural modeling in step 3 (Figure 4c) involved
of the generation of synthetics, seismic interpretation, building a framework of grid cells to be assigned prop-
time-depth conversion, and structural modeling. The erties required for reservoir characterization. “Fault
sixth step (Figure 4f), seismic attribute calculation, is modeling” is a process of representing the shapes of
the most time-consuming part of this study. The last picked faults by key pillars. In this study, each key pillar
step (Figure 4g) of the workflow is the creation of a has three shape points that can be edited separately to
DFN based on the seismic attributes and log data. tilt the pillars and model listric faults.
Input data are a 3D poststack migrated seismic vol- After fault modeling, the skeletal framework of the
ume, with 345 inlines and 188 crosslines, and digital structural model was built by pillar gridding. Distances
well data from 27 wells. The well logs used include between inlines and crosslines defined the dimensions
sonic and density logs, plus image logs from four wells. of grid cells.
Figure 1. Location of Teapot Dome relative to the Powder River Basin (modified after Cooper et al., 2006).
Figure 2. Stratigraphic column of Teapot Dome (modified Figure 3. Detailed overview of the Tensleep Formation and
after Cooper et al. [2006] and Wilson et al. [2015]). its subdivisions (modified after Wilson et al., 2015).
Figure 5. Series of coherence time slices (1100 ms). (a) Coherence of the raw seismic data, (b) coherence of dip-steered (event
steering) seismic data, (c) coherence of dip steered and fault enhancement filtered seismic data, and (d) results from Petrel’s ant-
tracking algorithm applied to coherence cube of panel (c). The enhancement of fault (red arrow) and circular structures (blue
arrows) could be improved by dip steering and filtering of the raw seismic data.
Figure 7. (a) Karst feature on seismic section and coherence time slice, (b) most positive curvature attribute (red) indicates
flanks of the karst feature, (c) most negative curvature attribute (blue) indicates thalweg of the karst feature, (d) fault on seismic
section and coherence time slice, (e) most positive curvature attribute (red) indicates upthrown fault part, and (f) most negative
curvature attribute (blue) indicates downthrown fault part.
ure 5c) has reduced noise and shows sharper faults. To and bottoms of circular patterns (Figure 7a–7c) that
improve fault and fracture visualization (Figure 5d), may be karst features in dolomitic intervals in the upper
Figure 8. Spectral decomposition images at (a) 30, (b) 40, and (c) 50 Hz. (d) Seismic amplitude coherence on spectral decom-
position cubes of (e) 30, (f) 40, and (g) 50 Hz. Arrows indicate faults (red) and karst structures (blue) visible at particular frequen-
cies. (h) RGB blending of these coherence slices (red: 30 Hz, green: 40 Hz, and blue: 50 Hz) clearly indicates faults.
CWT analysis using a Morlet wavelet. The output fre- based attribute energy. This attribute shows highest val-
quency cubes were generated in 5 Hz steps from 5 to ues when calculated along the strike of a feature and is
90 Hz and were then filtered with a dip-steered median therefore a useful indicator of the strike direction of
filter. The filtered cubes were used as input for coher- fractured zones (Figure 9a). Figure 10 shows GLCM-
ence calculations. based energy that indicates areas with high directional
Figure 8 shows semblance-based coherence time sli- variability. These areas are zones of fractures that have
ces based on spectral decomposition cubes of 30, 40, different fracture azimuths as indicated by colors.
The effectiveness of an attribute in DFN modeling
can be judged by comparison with image log data (Tha-
chaparambil, 2015). The GLCM-based attribute calcula-
tions were used to generate fracture strikes and dips
that were plotted as rose diagrams to compare with dia-
grams based on image logs (Figure 11). It was first nec-
essary to upscale the fault and fracture estimates. The
GLCM energy attribute produced a good match (Fig-
ure 11) and is a potential indicator for fracture strike
and dip estimation. For further calculation of a DFN,
this information can be used as driver for fracture inten-
sities modeling.
Conclusions
Seismic attribute analyses demonstrated consistent
delineation and enhancement of faults and fractures
in this 3D seismic survey case study. The attribute is
a good way to calculate volumetric estimation of frac-
ture intensities, which can later be used for the con-
struction of a DFN model of the Tensleep Formation
Figure 9. GLCM-based energy calculated for (a) maximum at Teapot Dome, Wyoming.
and (b) minimum values, corresponding to the highest and Prior dip steering and filtering of the raw seismic
lowest directional variation of seismic response. data created the high-quality seismic coherence cubes.
Figure 11. Rose diagrams for well 48-X-28. Comparison of fracture strike and dip of image log data (orange) with maximum azimuth
and dip of GLCM-based attributes energy (blue), entropy (green), homogeneity (rose), contrast (turquoise), variance (violet), and
dissimilarity (yellow). The last column shows the results of Thachaparambil (2015), where spectral edge attribute calculations were
converted into discrete objects called seismic discontinuity planes (SDPs) to compare their azimuth and dip with image log data.
In addition to noise reduction and sharpening of faults, based energy attribute proved especially valuable for
seismic attributes also highlighted geologic features building a DFN, as the fracture model closely compared
other than faults and fractures, such as karst features. with fracture models from previous studies.
Most positive and negative curvature attributes helped
to determine the upthrown and downthrown sides of
faults and could detect minor faults that were not ap- Acknowledgments
parent on coherence time slices. This makes informa- We would like to thank the RMOTC for providing the
tion from curvature attributes a useful supplement to data and the permission to publish the results of this
coherence interpretations. Coherence based on spec- study. Critical and constructive reviews by S. Yuan
tral decomposition cubes revealed faults and karst fea- and two anonymous reviewers helped to improve an
tures at different frequencies and could delineate early version of this paper.
anomalies otherwise buried in broadband data. Use
of two different attributes offered the possibility of References
enhancing geologic features not clearly recognizable Bahorich, M. S., and S. Farmer, 1995, 3-D seismic disconti-
on simple coherence slices. The GCLM-based attributes nuity for faults and stratigraphic features: The coher-
mapped fractured zones as areas of high directional ence cube: The Leading Edge, 14, 1053–1058, doi: 10
variability; directional color coding indicates fracture .1190/1.1437077.
strike trends. The resulting interpretation was con- Bakker, P., 2002, Image structure analysis for seismic in-
firmed by plotting the fracture trends as rose diagrams terpretation: Ph.D. dissertation, Technische Universiteit
and comparing them with image log data. The GLCM- Delft.
Chiaramonte, L., M. D. Zoback, J. Friedmann, and V. matrix and its application to seismic data: First Break,
Stamp, 2008, Seal integrity and feasibility of CO2 33, 71–77.
sequestration in the Teapot Dome EOR pilot: Geome- Eichkitz, C. G., M. G. Schreilechner, P. de Groot, and
chanical site characterization: Environmental Geology, J. Amtmann, 2014, Mapping directional variations in
54, 1667–1675, doi: 10.1007/s00254-007-0948-7. seismic character using gray-level co-occurrence ma-
Chiaramonte, L., M. D. Zoback, J. Friedmann, V. Stamp, trix-based attributes: Interpretation, 3, no. 1, T13–
and C. Zahm, 2011, Fracture characterization and the T23, doi: 10.1190/INT-2014-0099.1.
fluid flow simulation with geomechanical constraints Friedmann, S. J., and V. Stamp, 2006, Teapot Dome: Char-
for a CO2-EOR and sequestration project Teapot Dome acterization of a CO2-enhanced oil recovery and storage
Oil Field, Wyoming, USA: Energy Procedia, 4, 3973– site in eastern Wyoming: Environmental Geosciences,
3980, doi: 10.1016/j.egypro.2011.02.337. 13, 181–199, doi: 10.1306/eg.01200605017.
Chopra, S., and K. J. Marfurt, 2005, Seismic attributes — A Gao, D., 2003, Volume texture extraction for 3D seismic
historical perspective: Geophysics, 70, no. 5, 3SO– visualization and interpretation: Geophysics, 68, 1294–
28SO, doi: 10.1190/1.2098670. 1302, doi: 10.1190/1.1598122.
Chopra, S., and K. J. Marfurt, 2007, Volumetric curvature Gao, D., 2007, Application of three-dimensional seismic
attributes for fault/fracture characterization: First texture analysis with special reference to deep-marine
Break, 25, 35–46, doi: 10.3997/1365-2397.2007019. facies discrimination and interpretation: Offshore An-
Cohen, I., and R. R. Coifman, 2002, Local discontinuity gola, West Africa: AAPG Bulletin, 91, 1665–1683, doi:
measures for 3-D seismic data: Geophysics, 67, 1933– 10.1306/08020706101.
1945, doi: 10.1190/1.1527094. Gao, D., 2009, 3D seismic volume visualization and
Cooper, S. P., 2000, Deformation within a basement-cored interpretation: An integrated workflow with case stud-
anticline Teapot Dome, Wyoming: Master’s thesis, New ies: Geophysics, 74, no. 1, W1–W24, doi: 10.1190/1
Mexico Institute of Mining and Technology. .3002915.
Cooper, S. P., L. B. Goodwin, and J. C. Lorenz, 2006, Gao, D., 2011, Latest developments in seismic texture
Fracture and fault patterns associated with basement- analysis for subsurface structure, facies, and reservoir
cored anticlines: The example of Teapot Dome, Wyo- characterization: A review: Geophysics, 76, no. 2, W1–
ming: AAPG Bulletin, 90, 1903–1920, doi: 10.1306/ W13, doi: 10.1190/1.3553479.
06020605197. Gao, D., 2013, Integrating 3D seismic curvature and curva-
Cooper, S. P., and J. C. Lorenz, 2007, Tensleep Formation ture gradient attributes for fracture characterization:
fracture study compendium: Enhanced Oil Recovery In- Methodologies and interpretational implications: Geo-
stitute, University of Wyoming, 406. physics, 78, no. 2, O21–O31, doi: 10.1190/geo2012-0190.1.
Cooper, S. P., J. C. Lorenz, and L. B. Goodwin, 2001, Litho- Gao, D., T. Wilson, L. Zhu, and K. J. Marfurt, 2011, 3D seis-
logic and structural controls on natural fracture charac- mic curvature and curvature gradient for fractured res-
teristics Teapot Dome, Wyoming: Technical Report, ervoir characterization at Teapot Dome (Wyoming), in
Sand2001-1786, Sandia National Laboratories. K. J. Marfurt, D. Gao, A. Barnes, S. Chopra, A. Corrao, B.
de Matos, M. C., M. Yenugu, S. M. Angelo, and K. L. Marfurt, Hart, H. James, J. Pacht, and N. C. Rosen, eds., Attrib-
2011, Integrated seismic texture segmentation and clus- utes: New views on seismic imaging — Their use in
ter analysis applied to channel delineation and chert exploration and production: 31st Annual Gulf Coast
reservoir characterization: Geophysics, 76, no. 5, Section Society of Economic Paleontologists and
P11–P21, doi: 10.1190/geo2010-0150.1. Mineralogists Foundation Bob F. Perkins Research
Di, H., and D. Gao, 2014, Gray-level transformation and Conference Proceeding: GCSSEPM 31, 750–775.
Canny edge detection for 3D seismic discontinuity en- Gersztenkorn, A., and K. J. Marfurt, 1999, Eigenstructure-
hancement: Computers and Geosciences, 72, 192–200, based coherence computations as an aid to 3-D
doi: 10.1016/j.cageo.2014.07.011. structural and stratigraphic mapping: Geophysics, 64,
Eichkitz, C. G., J. Amtmann, and M. G. Schreilechner, 2013, 1468–1479, doi: 10.1190/1.1444651.
Calculation of grey level co-occurrence matrix-based Gilbertson, N. J., 2006, 3D geologic modeling and frac-
seismic attributes in three dimensions: Computers ture interpretation of the Tensleep Sandstone, Alcova
and Geosciences, 60, 176–183, doi: 10.1016/j.cageo anticline, Wyoming: M.S. thesis, Colorado School of
.2013.07.006. Mines.