Incontrovertible Evidence of Anisotropy in Crosswell Data Douglas E. Miller* and ChristopherH. Chapman, Schlumberger Cambridge Research, England SB1.4 Summary. Crosswell seismic data collected at BP’s test site near Devine, Texas show clear evidenceof P-wave aniso- tropy (v,/u, up to 1.3) confined to shalelayersin a shale/car- bonate sequence. Tomograms made assumingisotropy are severely degraded, but could be mistakenly interpreted as indicating (nonexistent) low velocity zones within the car- bonates. The effectsof the anisotropy are directly evident in the raw data once headwaves and internally reflected waves are correctly identified. Inversion for a layered, transversely isotropic, anisotropic medium yields a model that fits the data to the accuracy of the time picking. 1. Introduction. Crosswell data were collected using a piezoelectric pressure source with hydrophone receivers at BP’s test site near Devine, Texas. Sourceand receiver wells were vertical with a separationof 100m. The peak frequency of the received signal was approximately 800Hz. A full 56 by 56 “tomographic scan” was run with sourceand receiver depths at 1.5m intervals between 771.0 and 853.5m. Figure 1 showsa sonic log, a common-shot gather, and a common- depth gather (z,,,,,, = z,,,,,,,,), all plotted on the same ver- tical scale. From top to bottom, the formations encountered were the Austin Chalk, Eagle Ford Shale (top at 777m), Buda Limestone (796m), Del Rio Clay (a shale) (823m), and Georgetown Limestone (S4Sm). In the limestone zones, the direct arrival times predicted by the log agree well with the first arrivals in the common- depth gather; in the shales,the first arrivals are headwaves. Figure 2 shows an expanded view of the zone around the Del Rio Clay, with the log superimposedon the common- depth gather at a scale to allow direct comparison. In this zone, headwave arrivals are clearly evident in both the shot gather and the common-depth gather (marked “H” in Figure 2). The vertical event at 33ms marked “B” on the common- depth gather fits well with the direct arrival time predicted by the log and has been interpreted as such in a previous study (Harris, 1988) when only a common-depth gather was available. However, there is no vertical event evident in the shot gather at this time In the shot gather, the earlier event (“A”) has the moveout pattern one would expect for a direct arrival if the formation had a velocity 15 to 30 percent higher than that suggested by the log (and increasingslightly with depth). When direct arrival times (following “A” in Figure 2) are picked and inverted by standard time tomographic methods, the resulting velocity tomogram (Figure 3) showsa disturb- ing deviation from the expected layered solution. The prob- lem is clear when one considersresidual times obtained by subtracting from measuredtimes the times predicted by ray- tracing through a layered model built from the times picked for the common-depth traces. The residuals are all posi- tive; they increase as the ray angles (measured from hori- zontal) increase and reach highly significant values (3.6ms, almost three wavelengths). In the tomographic inversion, these residuals translate to excess slowness backprojected along the steepest rays. The aboveobservations raise a number of questions:If event “A” is the direct arrival in the shale layer, what is event “B”? Why does the model built from these times disagree with the log? Why does it do such a bad job of predicting times for source-receiver pairs at significantoffsetsin depth? If event “B” is the direct arrival, what is event “A”? Why does “B” look like the intersection of a downgoing event with an upgoing event in the shot gather? 2. Analysis. In this section we will argue that simple, self-consistent answers to these questions can be obtained if, and only if, we admit the possibility that the shalesare anisotropic. Consider first the question of whether the direct arrival is event “A” or “B”. Note that event “H” is continuous with the first arrival in the Georgetown limestone in both the shot gather and in the common-depth gather and that its apparent vertical slowness ( At/AZ ) in the common-depth gather is twice as large a.s in the shot gather. Event “H” is undoubtedly a headwave generatedat the boundary between the slow Del Rio clay and the faster Georgetownlimestone. Note further that event “A” has a smaller apparent vertical slowness in the shot gather than does “H”. The underlying wave event must, therefore, have a larger horizontal slowness (and a more horizontal ray vector). However, the headwave has a maximum horizontal slownessamong (longitudinal) waves propagating in the limestone (i.e. the P-wave slow- nessof the limestone). PSP headwaves are ruled out by the known shear velocity (0.38ms/m) of limestone. It follows that the wave giving rise to “A” must have propagated en- tirely within the Del Rio clay. As it is the first such event in the data, it must be labeled “direct arrival”. Now consider P-wave energy propagating in the shale that is multiply-reflected from the top and bottom of the layer. Writing “I)” for the direct wave, “T” for the wave reflecting once at the top, “TB” for the wave reflecting once at the top and once at the bottom, etc., the arrival times for these 825 Downloaded 06 Dec 2010 to 18.111.24.240. Redistribution subject to SEG license or copyright; see Terms of Use at http://segdl.org/