Technology Alliance Improves 3D Seismic Survey Design In Egypt's Western Desert

Mike Bahorich, Craig Jarchow
Apache Corp.
Houston

Paul Baltensperger, Abu Bakr Ibrahim
Qarun Petroleum Co.
Cairo

Michael Fleming, Maurice Nessim, Craig J. Beasley, Fred J. Barr, Ron Chambers
Western Geophysical
Houston

Research and development available through a technology alliance with a large geophysical contractor enabled Apache Corp. to solve a noise problem that had plagued earlier 3D seismic data acquired on its Qarun concession in Egypt's Western Desert.

Needing good seismic images of complex structures in the area, Apache employed a technology alliance with Western Geophysical. Under a nonbinding, nonexclusive agreement, Western essentially worked as Apache's R&D department.

This new approach to the contractor-operator relationship produced a unique solution to a noise problem on the Qarun concession and provided images of the subsurface not available previously in the area.

The alliance overshot the area of interest in a pilot 3D survey and used the densely sampled data to analyze signal, noise, and the energy required to most effectively resolve subsurface features. Decimation of the data by offset and azimuth revealed where data quality was highest and set acquisition and processing parameters for subsequent shooting.

Fig. 1 [229,169 bytes] compares an image from the 119-fold pilot survey, shot in the fourth quarter of 1997, with a 32-fold 3D data set acquired over the same location in 1994. In an area such as this in the U.S., data sets of about 30-fold are not uncommon.

Results of the pilot survey will determine acquisition parameters of a 290 sq mile 3D survey to be shot this year.

Problem, solution

Although Apache has had considerable success in the Western Desert, like any other international operator it has had to break new technical ground in its lightly explored area of interest. In situations such as this, technology becomes a critical component of success.

As an independent, however, Apache does not allocate the overhead associated with a large number of research specialists.

It was for this reason that Apache decided to form a technology alliance for international and domestic projects with a geophysical contractor with demonstrated R&D strength. Western Geophysical suited the need. It has more than a 30% market share, R&D capability in seismic acquisition and processing comparable to that of a major operating company, and a geographic presence complementing Apache's.

In Egypt, Apache holds 27 million acres, the largest of any single independent operator active in the country. Most of the acreage is in the Western Desert. In 1997 alone, the company drilled 53 wells in Egypt, including 30 producing wells, eight of them discoveries.

On the Qarun concession, immediately southwest of Cairo, Apache, Seagull Energy Corp., and Egyptian partner Qarun Petroleum Co. produce 40,000 b/d of oil from Qarun field.

Poor seismic quality complicates exploration on the Qarun concession. Unconsolidated sands at the surface impair shot and receiver coupling and generate statics on the order of 100 ms within a spread length. Significant surface waves generate a near-offset noise cone, and a near-surface basalt with frequent outcrops introduces back-scatter, shot-generated surface waves. The shot record in Fig. 2 [226,877 bytes] illustrates the problem. Across a less noisy record, reflection patterns would form a hyperbolic shape.

Noise degraded 3D seismic data acquired in 1994 on the Qarun concession to the extent that Apache's explorationists could not accurately interpret the complex structure and faulting that characterize the target formation. Yet reserves potential is great.

To solve the geophysical problem and unlock that potential, Apache explorationists asked Western to work with them on a solution in the form of an overshot parameterization test. The pilot 3D survey became the first project conducted under the companies' technical alliance.

The concept

The aim of the overshot pilot survey was to study seismic acquisition problems in the Qarun area. Oversampling would enable the alliance to cut the amount of energy-or fold-in later surveys to strike the best possible balance between cost and survey quality.

The pilot survey used a broad azimuth distribution as well as high fold. This allowed for comparison of wide vs. narrow azimuth shooting. Oversampling allowed the alliance to test wide vs. narrow azimuth shooting fully through migration.

Table 1 [29,462 bytes] describes the spread geometry and other recording parameters of the pilot survey.

With source and receiver parameters fixed, based on Apache and Qarun's experience in the area, the pilot survey evaluated noise characteristics associated with offset, azimuth, and fold.

Offset stacks

A crucial finding of the pilot survey experiment was that the signal to noise ratio (S/N) varied greatly when the oversampled data set was decimated into six 20-fold stacks by offset range (Table 2 [6,984 bytes]).

In general, an S/N sweet spot emerged with offsets of 1,500-3,000 m (Fig. 3a). Data from the 1994 3D survey had few traces in this offset range.

Obviously, a source-generated noise cone exists in the near offsets. A shooting geometry focused on near offsets is destined to produce poor results for all but shallow targets.

A similar view emerges with accumulated offset distribution (0-1,200 m, 0-1,700 m, 0-2,050 m, etc.; Fig. 3b). Similarity between corresponding ranges in Figs. 3a and 3b [474,242 bytes] tends to confirm the conclusion that data quality is highest in the 1,500-3,000 m range. Again, near offsets add no apparent value to the stack and are, in some cases, detrimental.

Azimuth distribution

Decimating the 119 fold data among different azimuth ranges also affects data quality, as shown in Fig. 4 [110,431 bytes]. But results can be deceiving.

In these sections, data quality (S/N) from the widest-angle azimuths, ±60°-±120°, seems slightly poorer than that of the narrower azimuths. A natural conclusion is that narrow-azimuth acquisition geometry would be superior to a wide-azimuth geometry.

Such a conclusion would be premature because the data have not yet been migrated. Acquisition is never completely independent of processing.

In the pilot survey, in fact, data quality in the wider azimuth stacks was comparable to if not better than that of the narrower azimuth stacks following migration. Another advantage of wide-azimuth geometry is that it biases the offset distribution toward traces outside of the source-generated noise cone at the target level.

Fold decimation

The spider plots in Fig. 5 [335,409 bytes] ,Fig. 6 [382,776 bytes], Fig. 7 [331,244 bytes], Fig. 8 [342,895 bytes], Fig. 9 [330,172 bytes] show how the Apache-Western alliance decimated the oversampled data set by fold and azimuth.

For restricted-azimuth geometry, the alliance cut lines on the outside. For broad azimuths, it cut every other line.

One of the difficulties in testing a variety of geometries is the intensive data processing required. To successfully evaluate each geometry, the data must be fully processed through the final 3D imaging processes.

Processes such as dip moveout (DMO) and migration are designed to map reflections and diffractions to their true position in depth. Prior to this mapping, the stack images (Figs. 3 and 4) may show complexity in the form of interfering reflections, reflections from out of plane and mispositioned in time. That which appears to be noise prior to imaging may in fact be usable data.

Until migration unravels this puzzle, it is difficult to determine which result is actually superior for interpretation. In this case, the wider azimuth stacks appeared to be of inferior quality but and after migration were comparable if not superior to the narrower azimuth studies. This result underscores the need to fully understand and apply all salient data processing steps in the testing procedure.

The alliance ultimately chose the 85-fold, slightly restricted azimuth geometry as the best balance between cost, shooting efficiency, and data quality.

Velocities were picked on the 119-fold line and applied to all data sets. This issue received much attention. An original concern was that this technique would help the lower-fold data sets too much, making them similar to the full-fold case. On the other hand, picking velocities data set by data set multiplied the chance for human error.

The decision to treat velocity as a constant proved adequate in the Qarun case because of the tremendous variation in data quality from low-fold to high-fold. The results from the lower-fold data sets are certainly helped somewhat by the 119-fold velocities, but the higher fold data sets are still substantially improved. In another area, a different approach might be preferable.

With the oversampled Qarun data set, the alliance calculated and applied surface-consistent reflection statics independently for each data set. Since an algorithm controls the procedure, it involves no interpretation along with the associated potential for additional human error.

The alliance then migrated each data set independently with a 3D algorithm.

Migrated results for each fold decimation appear in the time slices and vertical sections of Figs. 5-9. Even at 119 fold, the data set is not overshot in that significant improvements are evident as fold increases.

Sweet spot

The Qarun pilot thus identified an S/N sweet spot in the 1,500-3,000 m offset range. S/N was very low in the 0-1,200 m range.

S/N improved with fold at a rate greater than the standard rate of reduction of random noise with increasing fold (square root of the fold number). Higher trace density enables better sampling of the wavefield with benefits of better array forming and further advantages to the processing, including calculation of reflection statics. This should improve velocity as well, but the effect was not tested. If anything, the low fold data set shown would get worse because it would not have the advantage of having the velocities picked on the higher fold data.

With the new seismic acquisition geometry developed by the alliance, Apache and Qarun Petroleum are now comfortable with the applicability of cost-effective 3D in the Qarun concession and are currently shooting an extensive program in the area and evaluating optimal processing parameters. Evidence of success should be represented with production increases in 1999.

Acknowledgment

The authors thank Gary Jones of Western Atlas for help in forming the technology alliance and Apache, Qarun Petroleum, and Western Geophysical for permission to publish this article.

Bibliography

Baltensperger, P., Abu Bakr Ibrahim, Bahorich, M., Nessim, M., Fleming, M., Optimization of 3D Design, From Theory to Practice: The Qarun Discoveries, Cairo '98, Africa/Middle East Second International Geophysical Conference and Exposition, Cairo, Egypt, 1998.

The Authors

Mike Bahorich is vice-president, exploration technology, of Apache Corp. He holds a BS in geology from the University of Missouri and an MS in geophysics from Virginia Polytechnic Institute. He joined Apache as chief geophysicist in 1996 after working with Amoco Corp. as a resource manager.

Craig Jarchow, is geophysical technology coordinator at Apache. He holds a BA in geological sciences from the University of California at Santa Barbara and an MS and PhD in geophysics from Stanford University. He worked at Amoco in 1991-97 and joined Apache in 1997.

Paul Baltensperger is exploration manager, Qarun Petroleum Co.. He holds a BS in geology from New Mexico State University and an MS in geology from the University of Texas. He previously worked for Oryx and Apache.

Abu Bakr Ibrahim is prospect sector manager of Qarun Petroleum. He holds a BS and MS in geoscience from Alexandria University. He earlier worked for Geisum Petroleum Co. and Khalda Petroleum Co.

Michael Fleming is manager of geophysics, Europe, Africa, and Middle East (land), for Western Geophysical. He holds a BSc in physics from Warwick University (U.K.) and an MSc in physics of materials from Bristol University.

Maurice Nessim is manager of Western Geophysical's data processing center in Cairo. He holds BSc and MSc degrees in physics.

Craig J. Beasley is vice-president, research and development, for Western Geophysical. He holds a BS from the University of Houston, an MS from Emory University, and a PhD from North Texas State University, all in mathematics.

Fred J. Barr is manager, data acquisition research and development for Western Geophysical. He holds BS, MS, and PhD degrees in electrical engineering and began his career with Petty Geophysical Co., which became part of Western through mergers.

Ron Chambers is geophysical advisor to the president of Western Geophysical. He holds a BSc in physics from Colorado State University.

Copyright 1998 Oil & Gas Journal. All Rights Reserved.