Calibration Method Helps In Seismic Velocity Interpretation

C.E. Guzman, H.A. Davenport, R. Wilhelm
Shell Offshore Inc.
New Orleans

Acoustic velocities derived from seismic reflection data, when properly calibrated to subsurface measurements, help interpreters make pure velocity predictions. A method of calibrating seismic to measured velocities has improved interpretation of subsurface features in the Gulf of Mexico.

In this method, the interpreter in essence creates a kind of gauge. Properly calibrated, the gauge enables the interpreter to match predicted velocities to velocities measured at wells.

Slow-velocity zones are of special interest because they sometimes appear near hydrocarbon accumulations. Changes in velocity vary in strength with location; the structural picture is hidden unless the variations are accounted for by mapping in depth instead of time. Preliminary observations suggest that the presence of hydrocarbons alters the lithology in the neighborhood of the trap; this hydrocarbon effect may be reflected in the rock velocity. The effect indicates a direct use of seismic velocity in exploration.

This article uses the terms "seismic velocity" and "seismic stacking velocity" interchangeably. It uses "ground velocity," "checkshot average velocity," and "well velocity" interchangeably. Interval velocities are derived from seismic stacking velocities or well average velocities; they refer to velocities of subsurface intervals or zones. Interval travel time (ITT) is the reciprocal of interval velocity in microseconds per foot.

Seismic velocity, image

Seismic velocity is essential in the stacking or summing of multifold data. Basically, if the seismic velocity is incorrect, the power of stacking or summing is diminished, and the image is degraded. The greater the error in the velocity, the greater the image degradation.

The result of using an incorrect velocity is illustrated with several examples from the Mississippi Canyon salt canopy area of the Gulf of Mexico. No attempt was made to purposefully create images with the wrong velocities; the degraded images were available as a result of routine processing. The incorrect velocities led to incorrect interpretations of the subsurface.

The first example comes from Mississippi Canyon 211 (MC 211). Fig. 1a [172,207 bytes]; shows a south-to-north migrated seismic profile created with incorrect seismic velocities. Fig. 1b is the same profile based on a velocity interpretation that honored the base of salt. Note the base-of-salt and subsalt targets penetrated by the Exxon MC 211 No. 1. The base of salt is not imaged in Fig. 1a in fact, one can see a shear and salt intrabedded multiple instead of the base of salt and subsalt targets of Fig. 1b [172,207 bytes].

Fig. 2a [244,193 bytes] shows a seismic velocity panel in the basin south of the salt tablet. The seismic velocity necessary to stack or sum the data properly is given by the "semblance clouds," which indicate the velocity for which the stack energy will be maximized. The solid line through the semblance clouds is an interpretation of the seismic velocities in the basin.

Fig. 2b [244,193 bytes] shows a velocity panel over the salt tablet near the well location. The solid line at about 2.7 sec indicates the base of salt interpretation. The degraded image of Fig. 1a used the basin seismic velocity over the length of the salt tablet; this velocity, indicated by the dashed line in Fig. 2b, follows the noise train and misses the base of the salt and subsalt targets.

Fig. 3 [154,055 bytes] shows a seismic profile generated with incorrect (Fig. 3a) and correct (Fig. 3b) seismic velocities. The profile is from a salt body southeast of Mississippi Canyon 952 (MC 952). Again, compare the incorrectly imaged base of salt, the shear, and the salt intrabedded multiple of Fig. 3a with the clear base of salt shown on Fig. 3b. Although the seismic velocity panels are not shown for this profile, the panels indicate similar observations as discussed in the previous example.

Fig. 4a [165,362 bytes] shows a migrated seismic profile near MC 952, where the Shell MC 952 No. 1 was drilled near the location labeled SSP 2600. Note the base of the salt and the seismic energy below the salt body; this energy can be correlated to seismic events in the basin. The seismic horizons are "pulled up" in time by the faster rock-the salt. Correction for this pull-up is made by converting the profile to depth and will not be discussed here.

Fig. 4b [165,362 bytes] shows the same seismic profile prior to migration; observe the base of the salt and the pulled up seismic energy near the location labeled SSP 2200. The seismic velocity panels for the well location in the basin, SSP 2600, and for the pulled up seismic energy, labeled SSP 2200, are shown in Fig. 5 [106,725 bytes]. The red crosses indicate the velocity interpretations that generated the seismic profile of Fig. 4b (and of Fig. 4a after migration).

The seismic velocity must be sped up to image the subsalt energy. The seismic velocity at location 2200 indicates that the seismic reflections are feeling the effect of the faster rock (salt). If the velocity interpreter had ignored the measurement at location 2200 the subsalt energy would not have been present in the final image. Tighter spatial sampling of the seismic velocity panels is important in this case.

Figs. 6a and 6b [160,519 bytes]] from Mississippi Canyon 854 (MC 854) echo the same theme discussed in the foregoing examples. The base of salt is clearly imaged in Fig. 6b. The subsalt sedimentary section needs to be viewed with caution pending depth migration.

Predicting ground velocities

The calibrations discussed here use seismic stacking velocities, interpreted from semblance clouds mentioned previously, and ground average velocities measured with standard checkshots at well locations. Highly deviated wells were avoided in our work.

The calibration may also be done with seismic stacking velocities and well checkshot velocities converted to interval velocities. Calibration in the interval velocity domain requires extra care because conversion to interval velocities depends upon temporal sampling of seismic and well velocities. The effect of sampling will be illustrated later.

Seismic stacking velocities are extracted from dip lines near their respective wells and are then compared to checkshot average velocities. Over a wide geographical area, encompassing Pleistocene to Miocene rocks, the seismic stacking velocities were found to be, in general, 2-13% faster than the well average velocities; this difference between seismic and well velocities was found to increase with depth.

The top part of Fig. 7 [76,417 bytes] shows a plot of seismic stacking velocities (black) versus well average velocities (red) for two wells in the Green Canyon area-GC 65 No. 1 and GC 109 No. 2. The figure also shows a plot of the difference between the seismic and well velocities and a plot of this difference expressed as a percent of the well average velocity. The wells are located about 4 miles apart; the percent difference between seismic and well velocities is consistent at approximately 8%.

The bottom part of Fig. 7 displays interval velocities as a function of depth for the same wells. The horizontal axis of the plots is labeled interval travel time (ITT), a reciprocal of interval velocity in microseconds per foot. The solid lines correspond to the well interval velocities; the seismic interval velocity is larger than the corresponding well velocity, as is to be expected. The interval velocity for the GC 65 No. 1 is slower than the interval velocity for the GC 109 No. 2; this trend is preserved by the seismic measurement.

Fig. 8 [34,666 bytes] shows an interval velocity calibration from Shell's Auger field (Garden Banks). The solid line corresponds to the well interval velocity; the seismic interval velocities track the shape of the well velocity curve very well.

One set of measurements stands out as erratic: These measurements correspond to a seismic velocity that is oversampled or sampled too closely in time. The tight sampling gives rise to instabilities.

Within a given area such as the Mars-Ursa-Flathead basin, the deviation between well and seismic velocities is consistent to within 2% (Fig. 9 [72,146 bytes]). This consistency in the calibration allows the interpreter to make reasonable predictions of the ground velocities for depth estimation and interval velocity predictions. If the calibration shows inconsistent seismic-to-well velocity differences, the predictive power of the seismic velocity is reduced. Nevertheless, even in the case of rank wildcats far away from well control, the seismic velocity has proved to be a powerful exploration tool.

In late 1989, Shell drilled the Green Canyon 908 No. 1 wildcat. Prior to drilling, the seismic velocity was calibrated to the nearest well at GC 736 No. 1, approximately 25 miles northeast of the wildcat location (Fig. 10 [107,441 bytes]). The calibrated seismic velocity predicted that rock velocities in the Pliocene to Pleistocene section would be faster than any encountered previously in our deepwater wells. This observation was confirmed by the well results and is significant in that these fast background shales, bounding the slower sand, gave a very large reflectivity that was mistaken for a hydrocarbon indicator.

Fig. 11 [92,858 bytes] displays the GC 908 No. 1 average and interval velocities relative to the velocities at two other locations, MC 807 and GB 471. The velocities are compared from the mud line; note the 10,000 fps interval velocities measured 12,000 ft below the mud line. Also, the section 2,000-6,000 ft below the mud line is much faster in the GC 908 No. 1 well.

The information can be used to risk other fast basins in the area. False hydrocarbon indicators are possible in the neighborhood of these fast shale zones; the calibrated seismic velocity is essential in risking such amplitude anomalies.

In early 1990, Shell drilled the Garden Banks wildcat GB 594 No. 1 to test several high amplitude events in the basin (Fig. 12 [110,242 bytes]).Fig. 13 [29,135 bytes] shows the seismic interval velocity, the well interval velocity, and a smoothed curve through the well velocity. The seismic and well velocities track one another. Interestingly, the high amplitude target reflections occurred on either side of the fast zones at 12,000-13,000 ft and 16,000-18,000 ft subsea. The high reflectivity response was not due to hydrocarbons but to the contrast between fast bounding shales and slow sands.

It may be possible to extract higher frequency information from seismic velocities. This type of work is resource (interpreter)-intensive and requires areal interpretation of a large number of quality measurements.

Near hydrocarbons

In 1977, while working in the Edwards reef trend onshore Texas, our colleague, H.A. Davenport, made an interesting observation between seismic velocities and the occurrence of hydrocarbons in part of the trend: The seismic velocity diminished remarkably in the neighborhood of the wells with pay. Fig. 14 [193,035 bytes] shows seismic profiles with the seismic stacking velocities as color overlays. Compare the velocity response for the dry-hole seismic (Fig. 14a) with the response for the hole with pay (Fig. 14b).

We have been able to observe similar response of seismic velocities near some hydrocarbon accumulations in the Gulf of Mexico. Fig. 15 [67,915 bytes] shows a 3D profile in the vicinity of the Mars field; the seismic velocity slows over the area with stacked pays.

Fig. 16 [183,672 bytes] shows a 2D profile with a seismic velocity overlay from the Green Canyon area. A well encountered hydrocarbons in the zone indicated.

The slowdown in velocity is also important because it causes the structural picture to be hidden unless a lateral velocity correction is applied, essentially converting the image to depth. Fig. 17 [91,394 bytes] shows a 2D profile from Shell's Mensa field converted to depth with measured varying calibrated seismic velocities (top), and the same 2D profile converted to depth with a constant velocity over the area (bottom). Correcting to depth with the varying velocities places the crest approximately 400 ft higher.

Fig. 18 [61,858 bytes] shows a seismic velocity scan (black) together with a well velocity scan (red) tracking a seismic horizon from a line over Shell's Fandango field in the South Texas onshore. The well average velocity slows down 5-7%/mile over the crest of the field, and the seismic velocity mimics the well velocity. The strong velocity variation distorts the depth anticline, and the field looks like a saddle in time. Time maps are converted to depth with calibrated seismic velocity maps to restore the picture in depth.

Interpretation tool

Seismic velocity measurements are an integral part of seismic-based interpretation. The interpretation of seismic velocities creates images, both in time and depth, that can alter the view of the subsurface, and hence impact seismic-based models of the subsurface.

Calibrated seismic velocities can be used to predict ground velocities and help assess risks of prospective areas. The areal analysis of velocity measurements is essential.

The velocity response near hydrocarbon accumulations shown in the examples is in itself important because it can mask the true depth structure. The nature of the slow response is not fully understood. We speculate, however, that a seal is not an impermeable barrier and that leakage in the neighborhood of the trap could alter the nearby rocks. If the observation is confirmed through more measurements, the seismic velocity might gain a direct role in exploration.

Acknowledgment

We thank the management at Shell Oil Co. for permission to publish this work. We also thank our colleagues, Paul Garber of Enron Oil & Gas and Murali Ramaswamy of Shell Offshore.

Bibliography

Chalabi, M. Al., "Velocity Determination from Seismic Reflection Data." Chalabi, M. Al., "Seismic Velocities- A Critique," First Break, Vol. 12, No. 12, Dec. 1994/589. Davidson, M. J., "Toward a General Theory of Vertical Migration," OGJ, June 21, 1982, p. 288. Guzman, C. E., Ramaswamy, M., Wright, B. K., Lawler, K. P., "Depth Maps from Seismic Velocities Help Wilcox
Exploration," OGJ, Oct. 28, 1996, p. 62. Mastoris, S., "3D Seismic Interpretation Techniques Define Shallow Gas Reservoirs," OGJ, Jan. 9, 1989, p. 69. Mastoris, S., "Correction of Seismic Velocity Distortion Due to Gas," The Leading Edge, February 1990, Vol. 9, No. 2,
pp. 26-29. Neidell, N. S., and Cook, E. E., "Use of Seismic Derived Velocities for Stratigraphic Exploration on Land: Seismic
Porosity and Direct Gas Detection," AAPG Memoir 39, Seismic Stratigraphy II, pp. 49-77. Saunders, F. D., et al, "Alabama Ferry Field Detectable by Hydrocarbon Microseepages and Related Alterations,"
OGJ, Nov. 6, 1989, p. 53. Sherwood, J. C., "Depth Sections and Interval Velocities from Surface Seismic Data," The Leading Edge, September
1989.

The Authors

Carlos E. Guzmanhas worked with Shell Oil Co. since 1977. He holds a BS in physics from the University of New Orleans and an MS in physics from Purdue University.

Howard A. Davenport has worked at Shell Oil Co. since 1972. He holds a BS in mathematics from the University of New Orleans.

Rudy Wilhelmretired from Shell Oil Co. with 20 years of experience and is now a consulting geophysicist. He holds an MS in physics from the University of Texas and an MS in petroleum engineering from Tulane University.