OTC-18671-MS

Copyright 2007, Offshore Technology Conference This paper was prepared for presentation at the 2007 Offshore Technology Conference held in Houston, Texas, U.S.A., 30 April3 May 2007. This paper was selected for presentation by an OTC Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the Offshore Technology Conference, its officers, or members. Papers presented at OTC are subject to publication review by Sponsor Society Committees of the Offshore Technology Conference. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Offshore Technology Conference is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, OTC, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435. Abstract The signal to noise problems inherent in towed streamer data associated with mud volcanoes, subsurface heterogeneities and distributed gas in the Azeri, and Gunashli structures of the Caspian sea prompted the use of three dimensional four component ocean bottom seismic (3D/4C OBS) to improve imaging. The introduction of several innovative enhancements to the traditional ocean bottom cable technique, when applied cohesively across both acquisition and processing, resulted in cost savings compared to conventional OBS acquisition and improved final data quality compared to towed streamer seismic. To further guide cost effective designs of future ocean bottom seismic (OBS) for 3D, 4D or permanent sensor time-lapse, surveys over the Azeri, Chirag and Gunashli reservoirs, a

multi-faceted study was subsequently conducted evaluating both real OBC seismic data from Azeri and synthetic data from numerical modeling. This study evaluated sensor density requirements to properly acquire the PZ component of the wave field, as this component provides the best reflector S/N and, depending on rock physics, is sensitive to production-related fluid changes in the reservoir. Critical spatial sampling and acquisition parameters were established through a series of comparison tests including: Fold decimation tests - to establish the sensor density required to insure image quality and signal-to-noise necessary for time-lapse monitoring. Migration aperture tests - to establish the permanent array line length across the structure necessary to adequately image the hydrocarbon column. Repeatability tests - to establish if the noise floor from repeated data acquired in the Azeri OBC survey is adequate to allow effective time-lapse monitoring from OBC. Comparisons included data from fixed shots recorded on different receivers, and also data from reoccupied receiver locations recorded with different shots. Spatial resolution testing through numerical modeling - to understand the relationship of the shot and receiver spacing and resulting image resolution for different acquisition geometries including surface tow, roll-along OBC, and

OTC 18671

3D/4C and 4D Ocean-Bottom Seismic Surveys in the Caspian Sea Jack Bouska, BP

Figure 1: Location of study area shown within red box: Azeri Chirag Gunashli PSA, (ACG) Caspian Sea, Azerbiajan.

Figure 2: Examples of areas where towed streamer data is compromised over the structure crest due to seabed scarps, mud volcanoes, and near surface distributed gas.

2 OTC 18671

permanent OBS designs. Qualitative and quantitative analysis of: seismic volumes, horizon amplitude maps, difference sections, S/N estimates and comparisons to synthetic seismograms were used to evaluate the adequacy of the various decimation levels. Results of these evaluations indicate that future OBS designs employing relaxed spatial sampling specifications, compared to the original OBS survey, would provide adequate balance between minimum seabed equipment (cost) and recording of high quality structural, or time-lapse seismic data.

Introduction The Azeri-Chirag-Gunashli (ACG project) is a world-class

oilfield development (15 billion barrels oil in place) located in the South Caspian Sea; offshore Azerbaijan (Fig. 1) The ACG anticline extends in a northwest to southeast direction, in water depth of 120m to 350m. The structure is asymmetric with steep dips (40deg.) on the north flank and gentler (25deg.) on the south flank. The reservoir consists of 9 laterally extensive stacked Pliocene sandstone intervals; in the Pereriv and overlying Balakhany formations. Mud volcanoes of varying size penetrate the structure near the crest. The mud volcanoes are characterized by debris cones on the seabed fed by over-pressured shale from strata below the target reservoirs. The existing towed streamer seismic data, while generally of good quality, does contain areas of weak reflections over the crest of the structures, especially in the vicinity of the mud volcanoes, to the extent that accurate structural mapping over the crest has been seriously impaired. (Fig. 2) The cause of the poor data has been postulated as a combination of a number of factors:

P-wave absorption/attenuation through distributed gas in the overburden sediments

Disturbed/disrupted sediments in the vicinity of the mud volcano plume (Fig. 2)

Backscattered shot generated noise from near surface heterogeneities.

Following a 2D test of 4C seismic over Azeri in 2001, a pair of 3D/4C-OBS surveys were acquired by a Caspian

Geophysical crew in 2002, and processed in 2003, to image the Azeri (160 sq.km.), and Gunashli (114 sq.km.) areas of the ACG PSA. Both PZ (summed pressure phone & vertical geophone), and PS (converted shear wave) surveys were planned. The PS image was intended to provide improved data quality, by virtue of reduced attenuation through gas in the up going shear leg. Early processing demonstrated that the P-wave image was of markedly better quality than the existing towed streamer, to the point where the P-wave image from the 3D/4C OBS is now vital to the overall value derived from the seismic. (Fig. 3)

Survey Acquisition Method Data acquisition costs typically dominate the

overall seismic budget, however from an interpreters point of view, survey acquisition / processing / interpretation are strongly interrelated. The danger of treating acquisition and processing as sequential (separate) steps, is that assumptions related to data processing requirements, can lead to an unbalanced acquisition design, with overemphasis of expensive field parameters

The tasks of Acquisition design, processing

Figure 3: OBS vs Tow streamer data comparison.

Figure 4: Acquisition pattern for Azeri OBS survey. Left: source configuration around two active 6km rcvr lines (shown in blue) Right top: schematic showing areal extent of receiver lines used in decimation study. Right bottom full source (blue) and receiver map (red) for full area.

OTC 18671 3

management, and interpretation were carried out by a single core team of specialists. This unified approach resulted in two notable advantages:

Knowledge of the deficiencies or idiosyncrasies in the acquisition were carried seamlessly forward in the project, allowing processing flow, and program parameter adjustments, avoiding known problems

Innovations, which are intentionally imbedded in the acquisition design, are well understood during the processing stage, and were properly exploited to maximum advantage.

The remote location of the Caspian Sea exaggerated the

high cost of 3D/4C OBS acquisition. Budget pressures prompted innovation in acquisition design to accommodate the conflicting requirements of tight spatial sampling (high fold) over the crest of the structure, while maintaining adequate aerial coverage over the migration aperture extent demanded by the steeply dipping reservoir strata at depth.

Apart from boats and air guns, the underlying acquisition design and processing techniques of 3D/4C OBS surveys share many features in common with land 3D seismic. Numerous parameter optimizations were applied during acquisition and processing however six major areas of advance stand out as unique innovations, pioneered in the Caspian 3D/4C OBS surveys:

1. Variable cross-line spatial sampling via receiver line interlacing. Deployed during acquisition to induce fold variability generating high fold on the crest (to improve S/N) grading to lower fold on the down dip flank to expand migration aperture.

2. Uniformly sampled shot wave field comprised of a wide patch (wide aperture), 75m x 75m grid of source points (4km X 10.4km) surrounding each receiver line pair, forming one half of a 3D symmetric sampled wavefield (Vermeer 1994)

3. First break refraction tomography (made viable with the wide patch acquisition scheme) used to estimate P- wave (and indirectly S-wave) receiver statics, and near surface velocity model definition for joint inversion depth migration.

4. Pre-stack noise attenuation in the common receiver domain using 3D-FXY-Decon (3D Random noise attenuation, RNA, made possible via the large, regularly sampled source grid around each receiver point.)

5. Additional pre-stack noise attenuation via a second pass of 3D RNA in the single fold common offset domain, a technique borrowed from land processing.

6. Kirchoff 3D pre-stack time migration, with pre-migration fold normalization, and post migration offset dependant fold weight restoration, providing improved attenuation of backscatter noise (Bouska 1998), acquisition/processing footprint, and multiples.

OBC Imaging Examples The Azeri and Gunashli OBS surveys were designed to use

a non-uniform, interlaced, receiver line patch layout, creating a distribution of high fold coverage over the difficult zone near the crest of the structure, grading to lower fold over the better quality, deeper data in the flanks of the structure. Arranging a greater concentration of receivers over the poor data quality area served two purposes: first, it helped attack some of the noise associated with backscatter, and

compensates for weak signal penetration. Second, it maintained stack fold consistent along stratigraphy, rather than constant at one depth.

The advantages of OBC wide-patch, wide-azimuth

acquisition are readily apparent in figure 3, which shows a comparison between towed streamer (right) and OBC (left). The towed streamer depth slice (top right) is seen to suffer from significant disturbance in the core of the Azeri structure, while the OBC depth slice (top left) yields dramatic improvements in clarity, continuity and S/N. The full shape of the structure is now easily interpretable, including the large circular shaped rim of the mud volcanoes cordillera which pierces the S-W flank of the structure.

Design optimization using decimation testing Various seismic data decimation case history studies have

shown the value of using spatial sampling analysis so that the acquisition parameters can be adjusted to achieve appropriate balance between cost and final interpretation quality. The earliest application of (fold) decimation tests on 3D seismic data was reported by Bouska (1995, 1996), and showed how post acquisition decimation testing during processing assisted in determining appropriate parameters for subsequent exploration 3D surveys over large areal extent. Later other authors, (e.g.: Schroeder, 1998) used the same decimation processing methodology to test bin size and fold in relation to interpretability of reservoir characteristics over smaller producing fields.

The use of decimation testing was extended to 4D data as

reported by Nolte (2004) using the Valhall permanent sensor array. All of the above case histories used land or permanent OBS surveys incorporating a wide recording patch, wide azimuth geometries, where cost is more directly proportional to density of sources and receivers per unit area, compared to towed streamer seismic surveying. The results from Nolte (2004) highlight how the 4D signal can be interpreted at lower levels of receiver coverage than were deployed in the current buried cable. Permanent sensor installations are currently very expensive with spatially over sampled designs. Reliance on close spaced sensors may force this 4D technique to be uneconomic.

In areas where permanent seismic array 4D deployment is

being planned, there is strong motivation for analysis, such as decimation testing, to assist in setting the most appropriate sensor sampling level to balance budget and quality of 4D signal. Decimation testing on towed streamer 4D is not directly applicable to OBS geometries. However decimation testing on existing high fold OBS surveys can be used to estimate the expected permanent sensor image quality and 4D signal detectability for a variety of acquisition geometries with different detector densities

In this case study, the spatial sampling requirement for

additional OBS surveys was estimated by performing decimation testing on an existing 3D 4C OBC survey, as part of the data processing phase.

4 OTC 18671

Fold Decimation Tests The decimation tests focused on determining the minimum

sampling requirement (and CDP fold) for receivers on the seabed. The key assumption in this study is that sources are less costly, and regardless of receiver spatial sampling, dense shots would be deployed in a well-sampled aerial pattern to compensate for the sparser receiver density. For decimation testing purposes, all the shots for each receiver were used, with no shot decimation applied in this test.

Future OBS surveys will require a pattern of cables to be

arranged on the seabed, to adequately cover the production area (planned well-bores). The two elements contributing to receiver spatial sampling in this context are the distance between cables, and the distance between receiver elements within each cable. The decimation testing was designed to determine the impact on image quality for cable spacing of 360m and 720m, and for receiver element spacing of 25m, 75m, 150m and 300m, for each of the two-cable spacing. This approach yields a spatial sampling analysis corresponding to a range between: 112 rcvr/sq.km and 5 rcvr/sq.km.

The decimation scheme used eight of the acquired swaths

(as highlighted in the map Figure 4). Eight swaths were required to produce a 1km wide zone of full fold coverage in the centre of the test area.

Previous 4COBS processing steps were not repeated and

only the Kirchhoff PSTM processing step was re-run,

employing all the pre-existing velocities and undecimated pre-processed data. The data were re-imaged with variable amounts of input data omitted from the process, and attributes extracted from each decimated output volume. All imaging was output to the bin size of 25m x 37.5m to facilitate comparison with the previous imaging as the base case reference image.

The decimation tests were processed in three steps: Step 1: Simulate 720m and 360m line spacing, with 25m-

rcvr intervals; Step 2: Simulate 75m, 150m and 300m station intervals; at

both 360m and 720m line intervals; Step 3: Investigate migration aperture requirements by

omitting down dip stations from one of the datasets in 2) above.

Analysis of Decimation Test Results The receiver station decimation tests (Fig 4, Fig 5.)

investigated the effects of reduced sampling and reduced fold on the data quality. The most general observation is that reduced receiver spatial sampling has a very large effect on amplitude fidelity in a signal-to-noise sense, where a reduction in receiver density correlates directly with increased levels of noise on the seismic images, and amplitude maps. However it was also noted that the structural image (and mapping) remained viable down to the very lowest fold levels.

Analysis revealed that both reflection amplitudes and S/N remain reliable at or above a decimation level of 75m station

Figure 5: Periviv amplitudes from four of the decimated volumes. Figure 4: Time slice sections from four of the decimated volumes.

OTC 18671 5

spacing across the crest of the structure. On the flanks, the amplitudes remained stable at or above a station spacing of 150m. All of the decimations with 720m-line spacing were judged inferior to those with 360m-line spacing over the crest of the structure, but acceptable on the flanks.

The Migration aperture tests indicated that line lengths of 10.5km or greater would have less than 1dB effect on the amplitude fidelity at the oil/water contact of the Pereriv horizon.

Conclusions The use of wide-patch, wide-azimuth OBC technology has

resulted in a step change in data quality, compared to prior towed steamer surveys, over the Azeri and Gunashli structures in the Caspian Sea. The dramatic improvements in interpretability brought with the OBC technology have prompted further plans to perform additional OBC surveys on adjacent fields, as well as install a permanent sensor array for 4D reservoir surveillance.

The Azeri OBS decimation study provided valuable insight into data quality and interpretability as a function of sensor spatial sampling density. Results indicate that future 4D OBS surveys may be acquired more economically, using slightly reduced levels of receiver density, while maintaining adequate quality in the final seismic image and extracted attributes through an increase in source effort.

The suite of decimation, migration aperture, and resolution tests implied the following acquisition parameters which achieve a regular fold of 100 (crest) and 50 (flanks) would be adequate for future targeted 4D OBS acquisition:

Line length: 10,500m. Line spacing: 480m (crest and flank). Inline sensor spacing: 75m for a zone of 6km over

the crest. Inline sensor spacing: 150m for 2.25km in two zones

flanking the crest, in the down-dip region. 50m x 50m regular spaced source grid spanning

entire acquisition area. 25m x 25m subsurface bin size.

References Bouska, J., 1995, Investigating the Effects of Reduced Surface Sampling

in 3D Data Acquisition., 1995 CSEG National Convention. Bouska, J., 1996, Cut to the quick: Techniques for effective use of sparse

3-D in exploration, 66th Ann. Internat. Mtg: Soc. of Expl. Geophys., 84-85. Bouska, J., Lyon, T., Johnston, R., Buddery, D., Howe, D., Mueller, M.,

Thomsen, L., Ebrom, D., 2004, Acquisition design of the first four component 3D ocean bottom seismic in the Caspian., 74th Ann. Int. Mtg: Soc. of Expl. Geophys., SEG Expanded Abstracts 23, 49 and- : 67th Meeting, EAGE, Expanded Abstracts , B003.

Bouska, J. G., J. Howie, B. Nolte, R. Johnston, and R. Walters, 2005,

Azeri 4C Time-lapse design using 3D 4C OBS imaging decimation tests: 75th Annual International Meeting, SEG, Expanded Abstracts , 2406-2409.

Bouska, J., and R. Johnston, 2005, The first 3D/4-C ocean bottom seismic

surveys in the Caspian Sea: Acquisition design and processing strategy: The Leading Edge, 24 , no.9, 910-921.

Crompton, R., K. Dodge, P. Whitfield, J. G. Bouska, and R. Johnston, 2005, Depth imaging of 3D, 4C OBS surveys in the Caspian Sea: 75th Annual International Meeting, SEG, Expanded Abstracts , 425-428.

Howie, J., Lyon, T., Thomas, R., Manley, D., Ford, A., Robinson, N.,

Riviere, M., and Stewart, S., 2004, Long-term seismic strategy for a major asset: Azeri-Chirag-Gunashli, South Caspian Sea, Azerbaijan., 74th Ann. Int. Mtg: Soc. of Expl. Geophys., SEG Expanded Abstracts 23, 472

Howie, J. M., N. Robinson, M. Riviere, T. Lyon, and D. Manley, 2005, Developing the long-term seismic strategy for Azeri-Chirag-Gunashli, South Caspian Sea, Azerbaijan: The Leading Edge, 24 , no.9, 934-939.

Johnston, R., Bouska, J., Lyon, T. Ashby, A., Walters, R., Whitfield, P., Crompton, R., Ebrom, D., Mueller, M., Thomsen, L., Azeri 4C: 2004, Processing the first 3D OBS survey in the Caspian Sea., 74th Ann. Int. Mtg: Soc. of Expl. Geophys.,SEG Expanded Abstracts 23, 845

Luke, E., J. G. Bouska, A. Ashby, R. Johnston, and T. Probert, 2005, PS

multicomponent time processing over a mud volcano in the Caspian Sea: 75th Annual International Meeting, SEG, Expanded Abstracts , 939-942.

Lyon, T., Bouska, J., Johnston, R., Mueller, M., Thomsen, L., Reducing

structural uncertainty on the Azeri field using Ocean Bottom Seismic: Offshore Azerbaijan., 74th Ann. Int. Mtg: Soc. of Expl. Geophys., SEG Expanded Abstracts 23, 468 (2004)

Manley, D.M., S.F. Mohammed, N.D. Robinson, and R.W. Thomas,

2005, Structural interpretation of the deepwater Gunashli Field, facilitated by 4-C OBS seismic data: The Leading Edge, 24 , no.9, 922-926.

Michael Mueller & Thomas Lyon, AZERI FIELD 2D 4C OBS TEST

RESULTS AND 3D 4C OBS BUSINESS CASE EVALUATION, Baku Geophysical Conference 2002

Morice, S., P. Whitfield, R. Crompton, M. Aitchison, M. C. Mueller, J. G.

Bouska, R. Johnston, T. Lyon, J. Puech, and S. Tcherkashnev, 2005, Azeri field: New frontiers for 3D4C and borehole-seismic integration: 75th Annual International Meeting, SEG, Expanded Abstracts , 421-424.

Nolte, B. Clarke, R. Read, R., 2004, Decimation tests using Vahall ocean-

bottom-cable data., 74th Ann. Int. Mtg: Soc. of Expl. Geophys., SEG Expanded Abstracts 23, 45.

Probert, T., Bryan, R., Underwood, D., Mueller, M., Lyon, T. and

Rowson, C., 2002, Multicomponent Seismic Challenges on a Mud Volcano - Imaging the Azeri Field, 64th Mtg.: Eur. Assn. Geosci. Eng., F017.

Robinson, N., A. Ford, J. M. Howie, D. Manley, M. Riviere, S. Stewart, and R. Thomas, 2005, 4D time-lapse monitoring of Chirag Field: The Leading Edge, 24 , no.9, 928-932.

Schroeder, F. W., Farrington, T. G., Balon, S. G. and Rapp, C. S., 1998,

How fold and bin size impact data interpretability: The Leading Edge, 17, no. 09, 1274-1284.

Acknowledgments BP operates ACG field on behalf of the shareholders of the Azerbaijan

International Oil Company (AIOC) which include the following companies: BP 34.14%, UNOCAL 10.28%, SOCAR 10%, INPEX 10%, Statoil 8.56%, ExxonMobil 8%, TPAO 6.75%, Devon 5.63%, Itochu 3.92% and Amerada Hess 2.72%.

The authors would like to thank the AIOC shareholders for permission to

publish this case study and their input to the planning and execution of the project.