3D full tensor gradient method improves subsalt interpretation

Gary W. Coburn
Bell Geospace Inc.
Houston

Imagine you're working the deepwater Gulf of Mexico, looking for potential subsalt prospects to guide your company's bidding in an upcoming lease sale. There are no speculative 3D surveys in the area, just 2D seismic and a few well logs. So you obtain some regional 2D lines across a number of promising salt features and begin your initial structural interpretation.

The top of salt is pretty easy to pick. But, not surprisingly, the base is fuzzy in many areas. Large shadow zones wipe out the image, making it difficult to tell how thick the salt may be and whether sediments continue beneath the salt or truncate at the edges. With the limited data you have available, you could pick the base of salt in several different places, all of them reasonable.

How do you decide?

One option is an expensive reprocessing job. But it would be nice to have another choice-a way to independently test your salt interpretation against high-quality data not derived from seismic. Three-dimensional full tensor gradient (FTG) data can provide such an alternative.

This new geophysical technology was introduced to the oil industry at the annual meeting of the Society of Exploration Geophysicists (SEG) in November 1997. Recent articles have described the origins and potential of the technology.¹

This article focuses on an actual test study done on a regional 2D seismic line across the Green Canyon area of the Gulf of Mexico. The purpose of the study was to determine how well 3D FTG data could identify the base of salt, where standard seismic interpretation was ambiguous.

Initial seismic interpretation

In 1995, Bell Geospace acquired 3D FTG data over 46 blocks in the Green Canyon area, providing a 3D data set of high-quality gravity gradients. In late 1997, a single 2D seismic line was obtained from TGS-Calibre for this test study.

The line was approximately 86 miles long and ran northwest to southeast across the 3D FTG grid for about one third of its length. Water depths ranged from approximately 1,000 ft in the northwest to over 8,000 ft in the southeast. The line had already been interpreted. But for the purposes of this study, a fresh seismic interpretation was made by the author, an experienced deepwater explorationist.

To begin, the regional 2D line and cultural data file were loaded to a standard Landmark seismic interpretation system. The line crossed four major salt bodies separated by distinct sedimentary basins. Seismic data quality above the salt was good; therefore, the top of salt was easily defined and was not altered during the subsequent modeling process.

Data quality below the salt was extremely poor. The base of salt to the northwest appeared to be very deeply rooted, while salt structures in the southeast were thinner and more elongated. A reasonable interpretation of the base and edges of salt was made (Fig. 1 [330,722 bytes]). Because this study focused on salt thickness, no attempt was made to do a detailed interpretation of the overlying sediments or adjoining basins.

Interpreted time horizons for the sea bottom, top of salt, and base of salt were then converted to depth according to a velocity gradient derived from sonic logs in three nearby wells. According to this data, velocities were fairly constant within the sediments overlying the salt. Depth-converted horizons then were exported as an Ascii data file and loaded into a system specially designed for modeling gravity gradient data.

Density modeling

An initial depth model was constructed on the basis of the interpreted seismic horizons. Next, a number of density horizons were added to the model ( Fig. 2, top [114,058 bytes]). Using this data, the modeling software could calculate pseudogradients, which could then be compared directly with the measured gradients to identify anomalies-areas of too much or too little mass: The further the model from geological reality, the greater the differences between calculated and actual gravity gradients.

Instead of mapping specific reflectors, as in a typical seismic interpretation, density packages above and adjacent to the salt bodies were identified from well logs in the area. Only where distinct density contrasts appeared were they used in the model. Seawater density was set at 1.03 g/cc and salt at 2.19 g/cc.

Deep layers down to 150,000 ft, as well as the oceanic crust, were also added to the base of the model. Densities were assigned based on published literature. Originally, these deeper layers of the model were kept horizontal.

In addition to density intervals along the plane of the seismic section, certain assumptions had to be made concerning densities out of plane. The model constructed for this study was, in fact, a 2.5D model. To accomplish this, salt bodies were divided into layers corresponding with adjacent density packages. Then the software made it possible to restrict the estimated extent of salt in either direction out of the plane, based on all available data. Thus, at the outer edges of any particular salt body, densities were set to that of the surrounding sediments. Without taking this precaution, calculated pseudogradients would have been erroneous, making it difficult to match them with measured gradients from the 3D data set.

FTG modeling

At this point, a 2D traverse corresponding with the regional seismic line was extracted from the 3D FTG survey. Gradient data was available for only 30 miles along the 86 mile seismic section (boundaries of the 3D FTG survey are indicated by the two vertical lines in Fig. 1).

Nevertheless, salt bodies along the entire 2D line were incorporated into the geologic model because 3D gravity gradients are always affected by masses outside of the survey area. To avoid distortion at the edges of the gradient data set, it was important to extend the structural interpretation laterally.

The extracted 2D FTG data was converted to the appropriate format and imported to the gradient modeling system. Pseudogradients were calculated on the basis of the original model and compared with actual gradients measured in the field.

Those measurements came from Bell Geospace's 3D gradiometer survey system, which employs a suite of 12 highly accurate accelerometers rotating in three separate planes. This unique instrument directly measures five independent components of the 3D gravity field. Each gradient measurement contains directional information related to the geometry of subsurface features.

In this study, three of those gradients-Tzz, Tzx, and Txx-were most helpful in refining the salt interpretation. Tzz is shorthand for the vertical component of the 3D gravity field as it varies in the vertical direction. Tzx stands for the vertical component as it varies in the east-west direction. And Txx stands for the east-west component as it varies in the east-west direction.

The most important of the three turned out to be Tzz because it provided vertical gradient information within the plane of the 2D seismic line. The enhanced gravity, Tze-the mathematical sum of all the measured gradients-also proved useful. Had this been a 3D seismic data set, all of the 3D full tensor gradients would have been of roughly equal value, depending on the orientation of salt masses.

When pseudogradients calculated from the model were compared with these measured gradients, significant differences could be seen. Only the Tze and Tzz values are shown in Figs. 2 and 3, to simplify the comparison.

Notice how the calculated curve (thin line) falls below the measured values (heavy line) in some places and rises above it in others. These indicate areas in the model of either deficient or excessive mass. To make the two curves match, interpretive modifications to the size, shape, and thickness of salt bodies and sediments were necessary. Each time the model was altered, gradients were recalculated. In this manner, a good match was achieved.

Model modifications

Enhanced standard gravity measurements (Tze) did not provide sufficient detail at the depths of interest to make any contribution to the salt interpretation ( Fig. 3 [130,701 bytes]). The mostly long wavelengths of the standard gravity meant they were useful mainly in the deeper portion of the model.

Comparing calculated gravity values with measured gravity indicated basement layers were not horizontal, as originally modeled. They were actually dipping somewhat to the northwest, which was consistent with published literature. The model was altered accordingly.

In the main section of the model, FTG data made it possible to refine both the salt and sediment interpretations. First, the shape of shallow sediments had to be altered in order to get a good match between the actual gradients and pseudogradients. The gradient data indirectly confirmed several of the more significant faults above and adjacent to the salt. The attitudes of various layers in the basin separating the salt bodies were also adjusted.

Second, the edges and base of the salt itself were iteratively modified until the gradient curves matched. Corrections up to 18,000 ft were made to salt thickness in different places. The top of salt was not revised, however, as noted earlier. It was felt that the seismic data accurately identified this interface.

In the final revised model (Fig. 4 [232,541 bytes]), the Tzz and Tze curves matched very well. Tzx and Txx curves (not shown) also matched well over the northwest portion of the data set. The few remaining discrepancies between calculated and measured gradients appear to be due to either very shallow density anomalies (above the salt), or uncertainties in the extent of salt bodies out of the plane. In more typical interpretation projects, additional 2D crosslines likely would be available. A similar interpretation and modeling process on those lines would help narrow down most of the residual anomalies.

Revised seismic interpretation

Once the final gradient model was complete, all of the depth horizons were exported back to the Landmark interpretation system. There they were converted to time and redisplayed on the original seismic section. The revised base and edges of salt fit the seismic very well ( Fig. 5 [279,171 bytes]).

The most important refinement to the structural model is that some of the horizons apparently do not turn up as steep as they appear on the seismic. Indeed, the gradient data suggests they continue right through the seismic wipeout zones under the edges of salt bodies, truncating against the base of salt (note orange interval in Fig. 4).

At one point, the salt is less than 2,500 ft thick-thin enough to drill through-creating a potential subsalt objective.

Implications

Obviously, many aspects of this test study are different from a typical interpretation scenario. But the implications are clear.

In places where seismic data quality is poor and the base of salt uncertain, 3D FTG data provides an independent, highly accurate means of making the pick. What's more, the modeling process is simple enough for the average seismic interpreter to perform on his or her own desktop.

FTG data improves and verifies existing seismic interpretations, both 2D and 3D. An enhanced 2D salt model, such as the one created in this study, could be used to reprocess the 2D line more accurately or plan a more effective 3D survey. Also, because 3D FTG data is acquired as a true three-dimensional data set, it is particularly well suited for 3D interpretation projects. Enhanced 3D seismic interpretations that result from integration with 3D FTG can then be used to improve subsequent processing flows, including prestack depth migrations.

Wherever seismic data has difficulty imaging due to significant density contrasts-whether it's a subsalt play in the Gulf of Mexico or a basalt flow in the North Sea-3D FTG information could make the difference between success and failure.

Acknowledgment

Special thanks to John Adamick of TGS-Calibre for permission to publish the Green Canyon regional 2D seismic line.

References

1. Bell, R.E., Anderson, R., Pratson, L., "Gravity gradiometry resurfaces," The Leading Edge, January 1997; Pawlowski, B., "Gravity gradiometry in resource exploration," The Leading Edge, January 1997; "3D FTG enhances seismic," Hart's Oil & Gas World, January 1998; Bell, R.E., "Gravity gradiometry," Scientific American, June 1998.

The Author