t

Special section: Multicomponent seismic interpretation

Downloaded 06/04/14 to 129.116.232.233. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Interpretation of fractures and stress anisotropy in Marcellus Shale using multicomponent seismic data Mehdi E. Far1 and Bob Hardage1 Abstract Using a data set from the Marcellus Shale, we evaluated the advantages of multicomponent seismic data for fracture and anisotropy studies over conventional P-wave data. Using traveltime and amplitude analysis on preand poststack seismic data, we concluded that PS-waves can provide more accurate information about the location, orientation, and intensity of natural fractures and stress anisotropy than P-waves. Our analysis indicated that regional stress was the main cause of velocity anisotropy. Amplitude variation with offset and azimuth appeared to be more useful for fracture studies, whereas traveltime variations (especially PS-waves) provided a better indication of regional stress orientations. Principal directions for amplitudes and traveltimes of PP- and PS-waves were different. Misalignment of PP- and PS-waves principal directions suggested that the simplest, most realistic anisotropy model for the fractured Marcellus is monoclinic symmetry.

Introduction Advancements in marine and land multicomponent acquisition and processing have led to several applications for PS-data. These applications include highresolution, near-surface imaging; fault imaging; seismic imaging of gas reservoirs; lithology estimation using V P /V S analysis; and direct hydrocarbon indication and seismic anisotropy (Stewart et al., 2003; Hardage et al., 2011). Hardage et al. (2012) report that multicomponent seismic data are better for exploiting the Marcellus Shale than single-component P-wave seismic data, with the latter (PP data) being the most common seismic data used across the Appalachian Basin. Specifically, they show that the converted mode (PSV mode) provides better spatial resolution of Marcellus Shale stratigraphy than its companion PP mode. The difference in resolution is significant, with PP wavelengths being longer than PSV wavelengths by 40% to 50%. The use of seismic waves to determine the orientation of fractures has received much attention. Lynn et al. (1995) and Lynn (2004a, 2004b) use azimuthal variations in the reflection amplitude of seismic Pwaves to characterize fractured reservoirs. Tsvankin and Grechka (2011) also analyze the AVO and moveout patterns in azimuthally anisotropic media. Reflection amplitudes have advantages over seismic velocities in characterizing fractured reservoirs because they have higher vertical resolution and are more sensitive to the properties of a reservoir (Far et al., 2013b, 2013c).

Evidence from outcrop studies indicates that the Marcellus Shale has at least two sets of near-vertical fracture sets (Figure 1) known as J1 and J2, which can be either orthogonal or nonorthogonal (Engelder et al., 2009; Engelder, 2011). Therefore, the simplest models for fractured Marcellus Shale can be either orthorhombic or monoclinic (with a horizontal symmetry plane, parallel to bedding) media (Far et al., 2013b, 2013c). Data description The site selected for this study is in Bradford County, Pennsylvania. Bradford County lies in the northeast part of the asymmetric Appalachian foreland basin. The study area traverses Marcellus Shale and Utica Shale geology as well as numerous brine-filled sandstones and carbonates that are potential water storage targets. A modern multicomponent 3C 3D seismic survey spans the site, and an exploratory well was drilled at the center point of the 3D seismic image space. Numerous attributes of seismic data, and particularly attributes of multicomponent seismic data, are affected by the source-receiver geometry that is deployed across a survey area and the field procedures that are used to acquire the data. Specifically, an acquisition geometry should create adequate stacking folds not only for common-midpoint P-P and S-S data but also for common-conversion point P-SV and SV-P data. In addition, a seismic data-acquisition geometry must create a full range of source-to-receiver offsets and azimuths for all P- and S-wave modes. Full-offset and full-azimuth

1 The University of Texas at Austin, Bureau of Economic Geology, Austin, Texas, USA. E-mail: [email protected]; bob.hardage@ beg.utexas.edu. Manuscript received by the Editor 18 July 2013; revised manuscript received 21 October 2013; published online 25 April 2014. This paper appears in Interpretation, Vol. 2, No. 2 (May 2014); p. SE105–SE115, 15 FIGS.

http://dx.doi.org/10.1190/INT-2013-0108.1. © 2014 Society of Exploration Geophysicists and American Association of Petroleum Geologists. All rights reserved.

Interpretation / May 2014 SE105

Downloaded 06/04/14 to 129.116.232.233. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

data are particularly important if fracture intervals are to be detected and quantified, or if stress fields and geomechanical rock properties are to be analyzed (Hardage et al., 2012). The 3D multicomponent seismic survey that was to be implemented in this research was intended to be an orthogonal brick pattern in which 13 receiver lines spaced 268-m (880 ft) apart were deployed northwest to southeast to form a 3.2 × 3.2 km (2 × 2 mi) square of 3C geophone stations, with 97 receiver stations spaced at intervals of 33.5 m (110 ft) along each receiver line. The total number of planned receiver stations was 1261. This receiver grid was to be positioned in the center of a 8 × 8 km (5 × 5 mi) square array of source stations arranged in a southwest–northeast brick pattern in which 41 source lines were spaced 201 m (660 ft) apart. Each source line consisted of a sequence of four source stations spaced at intervals of 67 m (220 ft) with a gap of 268 m (880 ft) between successive four-station groups. This source-station pattern created 60 source stations per line, with a total of 2460 source points across the survey area. Each source involved a 1 kg (2.2 lb) explosive positioned at a depth of 6 m (20 ft). Figure 2 shows survey design parameters for the Marcellus 3C 3D data set. The preserved elongate axis of the Appalachian Basin extends southwest–northeast across the west

half of Pennsylvania. The east margin of the basin is overthrust by the Appalachian Mountains, and the west margin extends into Ohio and Kentucky. Appalachian sedimentation is controlled by repetition of passivemargin environments, basin deepening, and sediment starvation, and advances of immature siliciclastic units in a general east–west direction. Much of the reconstruction of deep geology comes from projections of trends outside of Bradford County compiled by the Pennsylvania Geological Survey (Harper, 1990, 2008), the West Virginia Geological Survey (Roen and Walker, 1996), and the United States Geological Survey (Milici and Swezey, 2006). The stratigraphy and basin structure of northern Pennsylvania reflect Precambrian rifting and sediment deposition in a passive-margin setting during most of the Cambrian through Early/Middle Ordovician (Figure 3). Structural features EC, RR, RS, RT, and RW (Figure 3) are elements of the Precambrian Rome Trough. The RT arm of the Rome Trough extending across Pennsylvania is offset by several regional faults and passes in the immediate vicinity of our study site. Late Cambrian events included plate movement of the present-day Appalachian area into the evaporative subtropical trade-winds belt, where it remained until late Mississippian time (Miall and Blakely, 2009). Figure 4 shows lithology sequence and log data from a well drilled in the middle of the survey area. Prestack data analysis Azimuthally variant prestack seismic data contain valuable information about fractures and stress anisotropy. As seismic waves propagate through fractured media or a medium under an anisotropic stress state, seismic velocity and amplitude are affected by anisotropy. If reflected seismic waves are recorded at the surface in different azimuthal directions, a systematic variation can generally be seen on amplitude and

Figure 1. Top: Exposure of Marcellus Shale. This unit is stratified into thin layers and has two orthogonal joint sets, J1 and J2 (Engelder et al., 2009; Engelder, 2011). Bottom: Nonorthogonal J1 and J2 joint sets in Marcellus outcrop on the Appalachian Plateau. J1 joints maintain the same orientation around oroclinal bends, whereas J2 joints change orientation to remain orthogonal to oroclinal bends. SE106 Interpretation / May 2014

Figure 2. Map of study area showing location of VSP calibration well relative to planned positions of source and receiver stations used for 3D 3C seismic survey.

Downloaded 06/04/14 to 129.116.232.233. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

traveltime patterns as a function of azimuth. Seismic anisotropy becomes more important for PS-waves compared to PPand SS-waves (Thomsen, 2002). It has been shown that PS-waves are more sensitive to fractures than P-waves (Mueller, 1991; Beaudoin et al., 1997; Hardage et al., 2011). Figure 5 shows the variations of P-wave traveltime and amplitude in a common-offset section close to the interval of interest (top and base Marcellus). The top Marcellus trough is indicated by the upper arrow on the right and the base Marcellus by the lower arrow pointing to the peak. As shown, amplitude and traveltimes change in the azimuthal direction. The interpreted fast direction is approximately 60°. Figure 6 shows the variations of PS-wave traveltime and amplitude in a common-offset section (the same offset as Figure 5), close to the interval of interest. The top Marcellus peak is indicated by the upper arrow on the right and the base Marcellus by the lower arrow pointing to the trough. Again, amplitude and traveltimes change in the azimuthal direction. The interpreted fast

Figure 3. Study site in Bradford County, Pennsylvania (Harper, 2008).

Figure 4. Well log data (density, gamma ray, and S-wave and P-wave velocities) from a well in the middle of the survey. Synthetic and surface recorded seismic data are also shown. Top Marcellus is the interface between Stafford and the Upper Marcellus. Base Marcellus is the interface between the Lower Marcellus and Onondaga. Interpretation / May 2014 SE107

Downloaded 06/04/14 to 129.116.232.233. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

direction is approximately 80°. However, the azimuthal change for PS-wave traveltime is much more prominent than for the P-wave (Figure 5). The time difference between the highs and lows of traveltime for the Upper Marcellus is approximately 40 ms, whereas for P-waves, this difference is only approximately 10 ms. This difference between P- and PS-waves can be seen at all offsets. Figures 7 and 8 show the changes in amplitudes (Figures 7a, 7c, 8a, and 8c) and traveltimes (Figures 7b, 7d, 8b, and 8d) in the Upper and Lower Marcellus for Pand PS-waves, respectively. To facilitate interpretation of amplitude maps, interpreted trends are shown by white and black lines. The sinusoidal pattern in PStraveltimes is clearly detectable, whereas for P-waves, these changes are more subtle. It should be noted that a careful static correction was done on P- and PS-waves

to remove the effect of the near surface. Therefore, azimuthal anomalies seen in seismic data are assumed to be due to anisotropy. Another piece of evidence that proves these variations are not due to near-surface effects is that highs and lows (especially for PS-waves) in traveltimes occur every 90°. Principal directions for P-wave amplitudes and traveltimes are almost the same, and principal directions for PS-wave amplitudes and traveltimes are also the same. Highs and lows of amplitudes and traveltimes for P- and PS-waves do not coincide with one other. This misalignment in principal directions of P- and S-waves can be attributed to monoclinic or triclinic symmetry of rocks (Sayers, 1998). Sayers (1998) shows that in order for P- and S-wave principal directions to coincide, C 36 should be equal to zero. In terms of fracture properties, this misalignment can be caused by the differences in compliances of fracture sets (Sayers, 1998). Sayers (1998) shows that in a medium with multiple sets of vertical fractures, if normal and tangential compliances of all fractures are the same, the principal directions of P- and Swaves coincide. An important issue in fracture modeling for seismic exploration, which has not been sufficiently emphasized in the literature, is the choice of coordinate system. In real-world problems, the orientation of fractures is usually unknown. Therefore, the problem of characterizing fractured reservoirs should be looked upon without assuming that the orientations are known. In other Figure 5. P-wave amplitude and traveltime variation with azimuth at offset ¼ words, one should not assume that the 8000 ft. Troughs shown by upper arrow show reflections from the Upper fractures are aligned with the coordinate Marcellus, and peaks shown by lower arrow show reflections from the Lower axis (seismic acquisition coordinates), Marcellus. nor that the number of fracture sets is known (Far, 2011). As an example, consider one or more than one set of parallel and vertical penny-shaped or asymmetric fractures (either orthogonal or nonorthogonal) in an either isotropic or VTI background, where fractures are not aligned with the coordinate system. For such a medium, the stiffness matrix will have 13 nonzero components in the arbitrary coordinate system, including C 16 , C 26 , and C 36 (for more details, see Far, 2011). Therefore, even a simple medium such as HTI will appear as a monoclinic medium in the arbitrary coordinate system. Therefore, theories developed for fracture modeling must be valid for an arbitrary coordinate system and the appropriate tensor rotation should be applied Figure 6. PS-wave amplitude and traveltime variation with azimuth at offset ¼ to the stiffness matrix (assumed to be 8000 ft. Peaks shown by the upper arrow show reflections from the Upper inverted from geophysical data) before Marcellus and troughs shown by lower arrow show reflections from the Lower Marcellus. concluding the type of symmetry. If the SE108 Interpretation / May 2014

Downloaded 06/04/14 to 129.116.232.233. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 7. P-wave reflection amplitudes (left) and traveltimes (right, in ms) for the top and base of Marcellus. The offset is increasing from 0 to 14,000 ft.

Figure 8. PS-wave reflection amplitudes (left) and traveltimes (right, in ms) for the top and base of Marcellus. The offset is increasing from 0 to 14,000 ft. Interpretation / May 2014 SE109

Downloaded 06/04/14 to 129.116.232.233. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

medium has orthorhombic or HTI symmetry, the tensor rotation should reveal the zero and dependent components. Previous studies on the same data set confirm that the fast direction is approximately 80° which is in agreement with what PS-wave analysis suggests (e.g., Gaiser and Verm, 2012; Far et al., 2013a). FMI data (Figure 9) also show that orientations of dominant natural fractures are almost 80°. However, careful analysis of the common-offset section shows that this fast direction may not be due only to fractures in a specific layer. Figure 10 shows a larger common-offset PS section. This figure shows that the azimuthal variation in traveltime (and amplitude) is not restricted only to deeper layers, but it continues all the way to the surface (approximately 1120 ms at 8000 ft offset). This fact is confirmed by information from the World Stress Map (Figure 11). Data from the World Stress Map is mainly based on information from well breakouts and hydraulic fracturing. It is assumed that by computing the isochron, propagation effects from above layers are essentially stripped, similar to layer stripping in PS processing. If traveltime is sensitive enough to the presence of fractures, we should be able to see an azimuthal change in an isochron from the top to the bottom of a fractured layer. The isochron for the top and base of the fractured

Marcellus (time difference between top and base Marcellus horizons) is shown in Figure 12a. However, we can only observe subtle sinusoidal variations in Figure 12a at the far offset, which is hard to interpret with confidence. To facilitate interpretation of the isochron map, a traveltime map from the base Marcellus is also plotted (Figure 12b). What this behavior suggests is that observed azimuthally variant patterns in traveltimes can be caused by a large-scale factor such as regional stress anisotropy, not by fractures. In other words, although traveltimes show clearer trends than amplitudes, they are not sensitive enough to fractures.

Figure 11. Data from the World Stress Map showing that in the area studied, orientation of the maximum horizontal stress is approximately 80° from north.

Figure 9. Rose diagrams calculated from borehole image logs acquired in the central-image calibration well for Upper Marcellus (left) and Lower Marcellus (right).

Figure 10. PS-wave amplitude and traveltime variation with azimuth at offset ¼ 8000 ft in a larger scale. Peaks shown by the upper arrow (around 1520 ms) show reflections from Upper Marcellus and troughs shown by the lower arrow (around 1600 ms) show reflections from the Lower Marcellus. SE110 Interpretation / May 2014

Figure 12. (a) Isochron created in ms as the difference between the Upper and Lower Marcellus PS-supergathers. (b) Traveltimes in ms for the Lower Marcellus PS-wave.

Downloaded 06/04/14 to 129.116.232.233. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

On the other hand, amplitudes could be more reliable for fracture parameter determination (Far et al., 2013a). Poststack data analysis The effect of azimuthal anisotropy can also be seen on poststack data. P- and PS-waves respond differently to fractures and stress anisotropy. Numerous real data observations show that PS-waves are more sensitive to fractures and azimuthal anisotropy than P-waves; therefore, their frequency content decreases more

significantly than does that of P-waves (Mueller, 1991; Beaudoin et al., 1997; Hardage et al., 2011). Also, the amplitudes of the PS-waves are affected more than Pwaves. Figure 13a and 13b shows poststack amplitude maps, along a picked top Marcellus horizon, for P- and PS-waves, respectively. An arbitrary inline is shown (45° from north). Color bars were adjusted automatically to avoid any bias. As shown, certain areas (possibly channels) are highlighted on the PS-amplitude map that are not clear on the P-wave amplitude map.

Figure 13. (a) Poststack P-wave amplitude map. (b) Poststack PS-wave amplitude map. Red colors show low-amplitude areas, whereas blue colors show high-amplitude areas.

Interpretation / May 2014 SE111

Downloaded 06/04/14 to 129.116.232.233. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 14a and 14b shows poststack instantaneous frequency maps (Taner et al., 1979) along the same picked Marcellus horizon for P- and PS-waves, respectively. Interpreting the instantaneous frequency maps, one can also identify geologic features with lower frequency content (blue colors). This decrease in frequency can be attributed to natural fractures, which cause attenuation. We also looked at the local frequency attribute

Figure 14. (a) Poststack P-wave instantaneous frequency map in Hz. (b) Poststack PSwave instantaneous frequency map in Hz.

SE112 Interpretation / May 2014

(Fomel, 2007), which shows frequency variation on a local scale, as opposed to instantaneous attributes that consider each sample separately (Figure 15). Figure 15b shows a cleaner image of the low-frequency parts in the Marcellus Shale, whereas Figure 15a does not give a clear indication of low-frequency sections. The PS-wave local frequency map is in a good agreement with the curvature map that Hardage et al. (2012) study.

Downloaded 06/04/14 to 129.116.232.233. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Figure 15. (a) Poststack P-wave local frequency map in Hz. (b) Poststack PS-wave local frequency map in Hz.

Conclusions Our study shows that for the Marcellus Shale, PSwaves can provide more accurate information about the existence of natural fractures and seismic azimuthal anisotropy. Variations seen in traveltimes are caused mainly by regional stress rather than fractures. Seismic amplitudes can provide more relevant information about fractures than can traveltimes. However, PS-wave

traveltimes indicate the correct direction of maximum horizontal stress, which is in agreement with all other evidence, such as stress data and FMI logs. In addition to the correct direction being provided, the degree of apparent anisotropy in PS-waves is more prominent than in P-waves. More specifically, for the Marcellus Shale horizon, stress anisotropy causes about a 40-ms difference between the traveltimes of fast and slow Interpretation / May 2014 SE113

Downloaded 06/04/14 to 129.116.232.233. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

directions, whereas only a 10-ms difference can be seen for P-waves. Principal directions of P-wave amplitudes and traveltimes are nearly the same, and principal directions of PS-wave amplitudes and traveltimes are also the same. However, principal directions of P- and PS-waves are different. With considering the fact that the geology being flat in the study area, and assuming that fractures are vertical, the simplest, most realistic anisotropy model for the Marcellus is monoclinic symmetry, with a horizontal mirror symmetry. Complex combination of fractures and stress anisotropy could also cause such a misalignment. Our study shows that poststack PS-wave amplitude anomalies can be a better indicator of natural fractures than P-waves amplitudes. The local frequency attribute provides a better image of fractured areas than does the instantaneous frequency attribute. Acknowledgments We thank the sponsors of the Exploration Geophysics Lab for their support. The first author also thanks the sponsors of the Texas Consortium for Computational Seismology (TCCS) and S. Fomel for their support and comments. We thank Geokinetics, Geophysical Pursuit, and Chesapeake Energy for permission to use this data set. Publication authorized by the director of the Bureau of Economic Geology. References Beaudoin, G. J., T. A. Chaimov, W. W. Haggard, M. Muel, and L. Thomsen, 1997, The use of multi component seismology in CBM exploration: A case history: 59th Annual International Conference and Exhibition, EAGE, Extended Abstracts, B009. Engelder, T., 2011, Outcrops of the Marcellus Formation: Pennsylvania State University. Engelder, T., G. G. Lash, and S. Uzcategui, 2009, Joint sets that enhance production from Middle and Upper Devonian gas shales of the Appalachian Basin: AAPG Bulletin, 93, 857–889, doi: 10.1306/03230908032. Far, M. E., 2011, Seismic characterization of naturally fractured reservoirs: Ph.D. thesis, University of Houston. Far, M. E., B. Hardage, and D. Wagner, 2013a, Inversion of elastic properties of fractured rocks from AVOAZ data Marcellus Shale example: 83rd Annual International Meeting, SEG, Expanded Abstracts, 3133–3138. Far, M. E., C. M. Sayers, L. Thomsen, D. Han, and J. P. Castagna, 2013b, Seismic characterization of naturally fractured reservoirs using amplitude versus offset and azimuth analysis: Geophysical Prospecting, 61, 427– 447, doi: 10.1111/1365-2478.12011. Far, M. E., L. Thomsen, and C. M. Sayers, 2013c, Seismic characterization of naturally reservoirs with asymmetric fractures: Geophysics, 78, no. 2, N1–N10, doi: 10 .1190/geo2012-0319.1. Fomel, S., 2007, Local seismic attributes: Geophysics, 72, no. 3, A29–A33, doi: 10.1190/1.2437573. SE114 Interpretation / May 2014

Gaiser, J., and R. Verm, 2012, SS-wave reflections from P-wave sources in azimuthally anisotropic media: 82nd Annual International Meeting, SEG, Expanded Abstracts, doi: 10.1190/segam2012-1293.1. Hardage, B., E. Alkan, M. DeAngelo, D. Sava, C. Sullivan, and D. Wagner, 2012, Improving the monitoring verification and accounting of CO2 sequestrated in geologic systems with multicomponent seismic technology and rock physics modeling: United States Department of Energy (DOE) Report, DOE award no. DEFC2609FE0001317, Bureau of Economic Geology, The University of Texas at Austin. Hardage, B. A., M. V. DeAngelo, P. E. Murray, and D. Sava, 2011, Multicomponent seismic technology: SEG. Harper, J., 1990, Leidy Gas Field, Clinton and Potter Counties, Pennsylvania, in E. Beaumont, and N. H. Foster, compilers, Structural traps I. AAPG Treatise of Petroleum Geology, Atlas of oil and gas fields. Harper, J., 2008, The Marcellus Shale: An old new gas reservoir in Pennsylvania: Pennsylvania Bureau of Topographic and Economic Survey. Lynn, H. B., 2004a, The winds of change: Anisotropic rocks: Their preferred direction of fluid flow and their associated seismic signatures: Part 1: The Leading Edge, 23, 1156–1162, doi: 10.1190/1.1825938. Lynn, H. B., 2004b, The winds of change: Anisotropic rocks: Their preferred direction of fluid flow and their associated seismic signatures: Part 2: The Leading Edge, 23, 1258–1268, doi: 10.1190/leedff.23.1258_1. Lynn, H. B., C. R. Bates, M. Layman, and M. Jones, 1995, Natural fracture characterization using P-wave reflection seismic data, VSP, borehole imaging logs, and insitu stress field determination: Presented at SPE Low Permeability Reservoirs Symposium, 29595. Miall, A., and R. Blakely, 2009, The Phanerozoic tectonic and sedimentary evolution of North America, in A. Miall, ed., Sedimentary basins of the world, vol. 5: Elsevier Science, 1–28. Milici, R., and C. Swezey, 2006, Assessment of Appalachian Basin oil and gas resources: Devonian Shale. middle and upper Paleozoic total petroleum system: Open File Report 2996-1237, United States Geological Survey. Mueller, M. C., 1991, Prediction of lateral variability in fracture intensity using multicomponent shearwave surface seismic as a precursor to horizontal drilling in the Austin Chalk: Geophysical Journal International, 107, 409–415, doi: 10.1111/j.1365-246X .1991.tb01402.x. Roen, J., and B. Walker, 1996, The atlas of major Appalachian gas plays: West Virginia Geologic and Economic Survey Publication. Sayers, C. M., 1998, Misalignment of the orientation of fractures and the principal axes for P- and S-waves in rocks containing multiple non-orthogonal fracture sets: Geophysical Journal International, 133, 459–466, doi: 10 .1046/j.1365-246X.1998.00507.x.

Downloaded 06/04/14 to 129.116.232.233. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/

Stewart, R. R., J. E. Gaiser, R. J. Brown, and D. C. Lawton, 2003, Converted-wave seismic exploration: Applications: Geophysics, 68, 40–57, doi: 10.1190/1.1543193. Taner, M. T., F. Koehler, and R. Sheriff, 1979, Complex seismic trace analysis: Geophysics, 44, 1041–1063, doi: 10.1190/1.1440994. Thomsen, L., 2002, Understanding seismic anisotropy in exploration and exploitation: SEG.

Tsvankin, I., and V. Grechka, 2011, Seismology of azimuthally anisotropic media and seismic fracture characterization: SEG.

Biographies and photographs of the authors are not available.

Interpretation / May 2014 SE115