Environmental constraints are critical issues for land seismic operations and require the development of appropriate equipment and methods. In 1998, Compagnie Générale de Géophysique (CGG) acquired and processed seamless seismic data on behalf of the Bahrain Petroleum Company (Bapco) from a complex field on the Island of Bahrain. The operation ran smoothly in spite of the difficulty of working amongst pipelines, wells, and other oil and gas installations; scarps and sabkhas; urban areas; a zoo, golf course, and racecourse; an oil refinery and aluminum smelter; and other industrial and commercial facilities. The survey deployed a wide range of recording equipment including two sets of vibrators and a combination of radio and cable telemetry recording systems, and a fleet-management system.
The successful completion of the survey was achieved through dedicated teamwork. The key elements were as follows: (1) collecting the correct baseline information; (2) detailed planning of the timeframe, human and material resources, adaptation of acquisition techniques to varying circumstances, and health and safety requirements; (3) continuous monitoring of external conditions and the impact of the operation on the environment, in full consultation with the appropriate authorities; (4) maximum flexibility in the field operations so as to allow for changing conditions and unforeseen events; and (5) close partnership between CGG and Bapco at all stages of the project.
A joint CGG/Bapco team designed sophisticated processing methods, with top priority being given to the numerous surface heterogeneities. In particular, the recording of an intensive up-hole survey allowed a dedicated team at CGG in France to compute a robust subsurface model and a precise set of primary statics. Seven velocity layers from the Quaternary sandstone to the base of the calcareous and dolomitic Eocene were mapped. These showed significant lateral and vertical velocity variations within identified geological units. Surface-consistent signal processing with calibration to well data was used to compensate for the variations. Other critical processing steps included detailed velocity control, the application of dip moveout routines with acquisition-irregularity-compensation schemes, Radon demultiples and 3-D algorithms, designed for optimal de-noising and imaging.
These combined efforts led to the delivery of a final 3-D migrated block. As a result, a complete reinterpretation of the area was possible that significantly increased the estimated recoverable oil reserves (particularly by-passed oil) in the Awali field.
The paper is based on presentations made at the GEO 2000 Conference in Bahrain in March 2000 by Postel and Mukhtar (2000) on the acquisition of the 3-D seismic data, and by Crémiere et al. (2000) on a complementary study of the data processing.
Bahrain Island has a surface area of about 586 sq km. It is the topographic expression of the northerly oriented Awali dome that is affected by a complex horst-and-graben fault system. The island has low relief and its highest point is Jebel ad Dukhan (134 m) in the center of the dome. The geological and topographical variety in the survey area (Figure 1), together with the many surface obstructions (Figure 2), made the survey a combined acquisition/processing challenge.
The Eocene Rus Formation crops out in the central erosional depression (the ‘Interior Basin’) of the dome (Figure 1). The Interior Basin is completely surrounded by the Eocene Dammam Formation that forms a low outward-dipping escarpment. Low-angle dip slopes of the Dammam Formation merge into the unconsolidated Quaternary deposits of the coastal lowlands that are most widespread in the north, west and south of the Island. In the center of the Awali dome, the Miocene Dam Formation (Jebel Cap formation) unconformably overlies the Rus Formation to form Jebel ad Dukhan. The geology of Bahrain Island is summarized in a 1996 Exploration & Production Highlights feature in GeoArabia (v. 1, no. 3, p. 376–379).
The 3-D survey of 1998 was the first such survey of Bahrain Island. It covered an area of 325 sq km, or about 55 percent of the Island’s surface area, and was centered on the main petroleum producing area of the Awali (or Bahrain) field. The history of the field dates from 1932 when it was the first oil field to be discovered in the Gulf region. More than 600 oil and gas wells have been drilled.
The aims of the survey were as follows:
(a) Cretaceous carbonates and clastics at a depth of about 2,000 ft (600 ms Two-way Time: TWT)— detection of small-throw faults and fractures and their orientation, and the identification of thin sand bodies and fluid contacts;
(b) Permian Khuff carbonates at a depth of about 10,000 ft (1,400 ms TWT)—detection of faults and their orientation, and the identification of porous zones and fluid contacts.
(a) Jurassic carbonates at a depth of about 4,000 ft (900 ms TWT)—detection of faults, fractures and their orientation, and the identification of facies changes and fluid contacts;
(b) Paleozoic clastics at depths of at least 12,000 ft (>1,600 ms TWT)—identification of facies changes and fluid contacts.
The survey area (Figure 2) was characterized by three distinct zones:
The southern zone is mostly desert and covers restricted areas. Some villas and farms are located in this area, which is also a natural wildlife reserve inhabited by animals such as sand gazelles (rheem). A sabkha is located on the western side of the zone.
The central zone covers the oil and gas field and includes 600 wells, many pipelines, and several production plants. The Al-Areen Wildlife Park, the University of Bahrain, and several large private properties are located on the western side of the zone.
The northern zone has urban areas such as Awali, Riffa, Isa, and Sanad. Several compounds and many large private properties are also located in this area. In addition, the Bapco oil refinery and the Alba aluminum smelter are located in the east of the zone.
As the survey took place during the summer (from June to November), high temperatures made working conditions difficult. In June, the average temperature was 37°C, which was 2°C higher than the record temperatures for the previous 96 years. July was also the hottest for 96 years. Temperatures were very high throughout the month, with a maximum of 54°C being recorded for 12 days.
The base camp for the survey was located 10 km from Manama, and Bapco’s Awali facilities were used for mechanical and geophone and cable repairs (Figure 2).
The extremely sensitive nature of the environment required the preparation of a complete health and safety report that detailed and rated all the risks associated with the operation. The major risk identified was road transportation. Special training sessions were given at the start of the survey to about 200 people who worked on the project. The course was divided into the following topics:
general safety information given by the Safety Advisor focusing on the objectives of the crew, the definition of an accident, and the risk matrix;
awareness of the major risks;
working conditions in the field;
awareness of the environmental impact; and
distribution of personal protective equipment.
In order to fulfill the technical, economic and logistical constraints of this survey, the 3-D parameters were designed as follows:
Figures 3 and 4 show the parameters in detail and the offset and azimuth distributions. We noted a uniform distribution of the offsets from 500 m to 3,500 m. Receiver lines were oriented E-W and vibrator lines N-S. Although the crew was equipped with two sets of vibrators, the time to move from one salvo to the next was on average equal to the time to record eight Vibrator Points (VP). This explains the selection of an 8-VP salvo and a roll of two receiver lines between successive swaths to reduce non-acquisition time between salvos.
In order to define the standard vibration parameters (4 sweeps linear 16 sec, 10–80 Hz), at the beginning of the survey, the following extensive set of sweep tests was recorded:
Low frequency limit: 8, 10, 12 Hertz (Hz).
High frequency limit: 60, 70, 80, 90, 100 Hz.
Sweep length: 10, 12, 16, 18, 24 sec.
Sweep number per Vibration Points: 4, 6, 8.
Sweep type: linear, log, random.
In accordance with CGG safety rules for urban vibroseis operations, special measures were taken when the vibrators were operated in the towns of Riffa and Hamad. Three types of VPs were then defined, the two last ones being used in urban areas:
Normal: 4 vibrators, 80 percent high drive, four sweeps (16 sec) range 10–80 Hz.
Low drive: 4 vibrators, 40 percent low drive, eight sweeps (16 sec) range 10–80 Hz.
Random: 4 vibrators, 95 percent high drive, eight random sweeps (16 sec) range 10–80 Hz.
The in-line vibrator pattern was 16 m between each vibrator and 12.5 m between each vibrating position. The geophone pattern consisted of two lines of 24 geophones set 4 m apart with 4.17 m between geophones.
Source: 10 Mertz M27 50,000 lb vibrators in two fleets.
Receivers: 4,400 strings of 12 SM4U LD 10 Hz geophones.
Recorder: Sercel SN388 with 2,500 cable channels and 300 Opseis radio channels, plus WRU microwave systems.
Surveying: Manpack Differential Global Positioning System (DGPS) surveying systems.
Quality Control: Geoland In-field Management System and In-field Geovecteur Processing System.
In order to conduct this type of operation in a timely and cost-effective manner, the required official permits had to be obtained well in advance, and close contact maintained with the local authorities. For some areas, several scenarios were prepared and the crew had to be able to quickly adapt equipment and techniques at only a few hours’ notice. Close liaison with the local authorities was established with the help of Bapco.
The survey was recorded in several parts as follows (Figure 5):
Part 1: Very soft ground in the sabkha on the western side (although it was still possible to vibrate), but very hard and stony ground in the east.
Parts 2 and 4: Recorded in block (or zipper) mode to reduce geophone redeployment from swath to swath. With such a technique, a receiver line on each block was deployed only once. In part 2, intense activity around the limestone quarry (where there were 800 truck movements per day) caused many line problems. In part 4, special bridges had to be built over pipelines so that the vibrator trucks could cross without wasting time on detours.
Part 3: Not in the original prospect, having been planned after the start of operations. It was recorded as a single patch. The receiver equipment was deployed once and all the shots were recorded on the same template.
Part 5: Included industrial and urban areas and required extensive use of Wireline Remote Units and Eagle radio-telemetric recorders with adapted sweep parameters in low drive or random sweep.
A two-week night shift was set up for the surveys of the towns of Riffa and Hamad. The survey was shot in multi-swath mode with more receiver lines than normal laid out, in order to minimize the amount of time the equipment remained in the area during recording operations.
To optimize planning and permitting, all possible non-seismic information was stored in the field management system database.
Roads and pipelines were surveyed by DGPS in the field. The rest of the files were obtained by digitizing 1:10,000-scale and 1:2,000-scale maps for areas such as the oil refinery. In terms of work time, digitization of the coast took two days, and Hamad town one week. These represented only a small part of the non-seismic information and give an idea of the difficulty of the task. By bringing together the tasks performed by survey and digitization, personnel in the field could work with an accurate map that showed receiver seismic point positions and all surface obstructions. It was a very useful planning tool.
Figure 6 is part of the map that was distributed to each line assistant, observer, field manager, and surveyor. It shows the positions of VPs (green crosses) and the location of pipelines (orange lines). Time information extracted from daily observers’ reports showed that these positions were vibrated continuously for 4 hours and 30 minutes. The time between two VPs did not exceed 4 minutes, including all vibrator movements and pipe crossings that, as the map illustrates, were particularly troublesome in this area.
In order to improve safety and real-time field planning, a Fleet Management System was set up at the start of the survey. This lightweight vehicle-tracking system based on DGPS and Syledis buoys offered continual and accurate positioning with a high tracking rate of the survey’s large vehicle fleet.
Each vehicle tracked was identified by a frequency channel number used to transmit its position. All vehicles could be seen on a satellite-image display (Figure 7). This was very useful for monitoring field operations from the base camp and the recording unit.
Quality Control and In-field Processing
The Quality Control (QC) process was fully integrated and included the following systems:
preplanning and field management system;
24-bit recording system with a two-way connection to the field management system;
real-time vibrator positioning system synchronized with the recording system; and
field processing and attribute QC analysis system.
All systems exchanged support data in SEG’s standard Shell Processing Support format.
The first major check made on a 3-D survey is to ensure that the subsurface bin coverage has been respected. Figure 8a shows the results of the fold for all offsets: a reasonably good distribution is present over the entire area with a fold minimum of 50 instead of the nominal 68. On the shallow offset maps (Figures 8b,c) that correspond to the shallow targets, holes in the coverage are areas (such as the central part of the refinery) where recordings were restricted.
Each vibrator was equipped with a DGPS receiver connected to the vibrator control system. The receivers recorded satellite-positioning information and also the differential corrections transmitted by the reference station. Each vibrator computed the base-plate coordinates, and transmitted them to the recorder at the end of each vibration. On the recorder-user interface, the software read the surveyed coordinates and displayed the location at the beginning of the day. During recording (and immediately after each vibration), it computed the location of the center of gravity, controlled the validity of the position, and compared it with the as-laid-out position. The threshold for this survey was set at 5 m.
The quality control of the vibrator’s seismic behavior was also displayed. Bar graphs indicated the average and maximum values of the phase error; the distortion and force for each vibrator for the current vibration and the mean value of these attributes over the last 50 vibrations. This kind of visual QC was of great help to the observer who was immediately able to detect a vibrator that did not comply with the specifications.
Although a real-time positioning system was available with the recording system, coordinates on first-break stacks were checked to verify receiver positions and confirm file transfers. To control the position of a receiver, traces were gathered into a circle centered on that receiver. A linear moveout was then applied by time shifting the data using the distance divided by the refraction velocity, and the traces were displayed in four quadrants. All traces belonging to the same quadrant were stacked and the program picked the first breaks in the four quadrants. An example of positioning QC is illustrated in Figure 9.
When a receiver was incorrectly positioned, the first breaks were distorted and differences occurred between the four picked times and surrounding ones. These time picks were used to compute a new position. A 3-D base map showed the position before and after repositioning. The survey team checked and corrected any positional anomaly before final approval.
DATA PROCESSING: ISSUES AND ACHIEVEMENTS
An in-field processing system received seismic data on cartridges, and the SPS and vibrator QC data on floppy disks. A coarse grid of data was extracted to produce in-line and cross-line stack sections. Several seismic and non-seismic attributes were also computed and loaded into a database for quick and efficient control of the overall quality (Figure 10).
With the results of the in-field processing, the processing team was able to concentrate on the critical challenges of optimal structural delineation of both shallow and deep oil- and gas-bearing targets with a special emphasis on the compensation of near-surface heterogeneities. In particular, the various outcrops had a highly variable weathering zone. This caused significant static anomalies, signal de-phasing with variable bandwidth, and amplitude discrepancies. The result was poor stacking and the loss of temporal resolution. This problem and the implemented solutions are discussed in the next section, followed by a brief summary of the critical steps that significantly improved data quality; for example:
compensation of acquisition irregularities due to the numerous obstacles in the survey area;
removal of multiple energy generated in relation to strong impedance interfaces; and
appropriate 3-D imaging of reflections.
Near-surface Heterogeneities: an Innovative Solution for Primary Statics
The 3-D survey area covered the Awali dome that has the surface expression of a wide internal depression with Jebel ad Dukhan at its center and surrounded by a low escarpment. The rimrock of the escarpment is composed of limestone, dolomite, marl, and shale. Diagenetic heterogeneity is evident in places and there are locally developed surficial Quaternary deposits. From the beginning, the complex morphology prevented the use of refraction seismology and led to the decision to build a 3-D near-surface velocity model in order to compute the primary statics.
The model integrated the following information that was either newly acquired or supplied from archives:
uphole data from the 1982 2-D seismic acquisition program;
uphole data acquired from the 3-D seismic acquisition program;
elevations for the 3-D survey;
structure contour maps of outcropping members of the Dammam Formation—and top and base of the Rus Formation;
geologic map of Bahrain Island at 1:50,000 scale.
The following three-step approach was used to produce the model:
Elevations, depth of tops and base of the reference formations (see above), coordinates, and thickness of upholes and velocities of interpreted velocity layers were entered into a 3-D database and mapped for control.
Iterative 2-D interpretations on E-W and N-S lines.
Inputting model parameters into the database as follows: digitizing the six horizons (see above) and inputting of interval velocities at all up-hole locations and along extrapolated points on the 3-D borders; gridding at 200 m spacing; and mapping (horizon depth, interval thickness, and interval velocity) as quality controls.
The resulting velocity model is shown as Figure 11.
The ‘modeled statics’ were derived from the database by arithmetic calculations performed for every shot point and receiver point of the 3-D acquisition. The elevation component of the static (elevation divided by the correction velocity) was then added to the model statics to obtain the primary statics. The reference datum was mean sea level (0 m). The validity of this solution was checked against elevation statics and field statics derived from up-hole times, to produce the following results:
Elevation statics have both long and medium wavelength biases that justify the use of dense uphole data measurements. This served as another demonstration that elevation statics may create structural interpretation bias.
The field statics computed by means of a simple arithmetical approach had organized medium-wavelength biases, so demonstrating the limitation of the arithmetical interpolation technique. The wavelength and the magnitude of this bias were locally out of the reach of standard residual statics solutions based on reflection-stack optimization. Static residues may therefore affect the stack image in this case.
In summary, the positive impact of the geological model was two-fold:
The primary static field allowed improved structural control and stack image.
The iterative process of static computation and geological reinterpretation provided greater insight into the geometry and properties of the surficial layers. This information was then available for further use and updating, such as depth-domain processing or geotechnical analysis.
A Solution for Signal Discrepancies
Initial diagnostics (raw shot-point and raw stack displays) showed that the recorded data appeared to lack frequency bandwidth and temporal resolution. Application of a surface consistent spiking deconvolution was the preferred solution to take into account the variable nature of the outcrops present in the survey area and to achieve improved temporal resolution. Tests confirmed the superiority of the spiking deconvolution against the alternative gapped deconvolution approach in terms of output signal bandwidth and phase stability. The selected approach was based on the assumption that input data was of minimum phase and was workable with careful parameter selection. The three-step approach is described below:
Minimum phase shaping of the data using the pilot sweep was purely deterministic. The vibrator signal was assumed to be the pilot sweep autocorrelation (zero phase signal emitted). It was transformed into minimum phase and processed accordingly (including minimum phase deconvolution). The minimum phase filter was calculated by spectral replacement (Cambois, 2000).
Surface-consistent spiking deconvolution. This is a statistical method where the whitening operator (in phase and amplitude) is estimated from the recorded data. The surface decomposition of the components compensates for outcrop variability. The critical aspect of this process is to achieve good operator stability. This is ensured with appropriate software and correct parameterization:
Wavelet estimation was performed in the time gate (0.50–2.5 sec) with an acceptable signal-to-noise ratio; the ground roll was removed from the estimation by means of a dedicated mute.
A deterministic extrapolation of the signal bandwidth outside the low frequencies of the recorded data was done in accordance with the minimum phase shaping described above. De-absorption had not been used prior to whitening deconvolution.
Control versus well data. The product was found to be zero-phased through cross-correlation techniques against vertical seismic profile data supplied in the early stage of the processing by Bapco. This supported the final decision on signal processing production parameters. Improvements in signal bandwidth and good phase matching were apparent.
In summary, the selected approach made it possible to broaden the usable bandwidth in all areas, whatever the outcrop, with good phase control against well data. This was achieved from the early stages of the processing, so making it possible to perform velocity picking and further processing tests on zero-phase data.
A Failed Attempt to Solve Amplitude Discrepancies
Processing in preserved amplitude mode for the purpose of reservoir characterization is common practice for marine seismic data. However, for land data, amplitudes are strongly affected by noise, weathering, and coupling effects. A source-receiver decomposition of the measured amplitudes in the analyzed gate can compensate for these discrepancies. This procedure is performed after spherical divergence compensation, and is completed by a time and offset correction.
In this particular case, because of the heterogeneity of the ground and the noise-generating conditions, the ideal compensation could not be achieved. Instead, pre-stack trace balancing had to be implemented. The failure in compensation was because the average amplitude measured on single traces was far from being representative of reflected signal amplitude, as is often the case with a land dataset.
Other Critical Processes
Removing Multiple Energy
Multiples were clearly identified on both stacked data and velocity spectra, as it is often the case in the Gulf region. Well velocities were used to calibrate the seismic velocity picks. This is an important aspect to consider, particularly in view of applying demultiple tools based on velocity discrimination. The demultiple technique selected was based on Radon decomposition, and was applied with a significant improvement in the stack image at target level.
3-D Algorithms for Imaging and De-noising
The main goal for 3-D dip moveout was to remove the dip component from the offset in the source-receiver plane. The use of a band limited spatial interpolator was also highly beneficial for data regularization prior to stack. Poststack random noise attenuation using a 3-D spatial deconvolution technique also proved beneficial. Final poststack 3-D migration was applied using a finite difference algorithm in the Fourier transform domain. It was a critical step toward properly imaging faults and lineaments.
Data acquisition was achieved on schedule and with good cost control. The survey had minimum environmental impact with no long-term effects. The whole operation (totaling 350,000 exposure hours and 700,000 vehicle-km) was completed without any loss due to injuries, and produced high-quality data for successful processing. Thanks to the early quality control performed at crew level the team could focus on the high-end processing issues.
The numerous surface heterogeneities and obstacles resulted in noisy and irregularly recorded data. However, the use of original and proprietary methods in addition to high standards in processing led to a clear structural delineation of the geological features. The signal-to-noise ratio of the data did not allow for a full ‘true amplitude’ processing stream, but amplitude balancing in a long time gate was used in order to minimize the compromise.
The success of the survey was due to the close partnership forged between Bapco, CGG, and the local authorities, from the planning stage right up to final product delivery. The continuous monitoring of external conditions and of crew impact allowed maximum flexibility of the survey parameters. In addition, the gathering of the correct baseline information, the highly detailed planning and the use of state-of-the-art equipment and an integrated QC system were key factors in the survey’s success.
The final data quality was rated as good, as shown in Figure 12. The survey has had a considerable impact on the understanding of the subsurface geology of the Awali field. The acquisition of the 3-D migrated block has enabled a complete reinterpretation of the area to be made that has significantly increased the estimated recoverable oil reserves in the field. In particular, it has delineated potential accumulations of by-passed oil (Mukhtar and Khalaf, 2000).
ABOUT THE AUTHORS
Jean-Jacques Postel is Applied Technology Manager for the Land and Shallow Water Acquisition Strategic Business Unit of Companie Générale de Géophysique (CGG). He graduated as a Civil Engineer from the École Centrale de Lyon in 1978. He joined CGG in 1980, spending five years in the field as Party Manager and Party Chief in the USA and Argentina. In 1986, Jean-Jacques moved to processing and occupied several positions from Group Leader to Center Manager in China. In 1990, he joined the R&D team as Area Geophysicist and was appointed R&D Manager for Land Acquisition in 1995. Jean-Jacques’ interests are seismic acquisition equipment and seismic processing. He is a member of SEG, EAGE and AFTP.
Abdul Nabi Mukhtar is a Senior Geophysicist with the Bahrain Petroleum Company (Bapco). He received his BSc in Geological Engineering/Petroleum Engineering from the University of Texas at Austin in 1986. Abdul Nabi then joined Bapco and has worked mainly in exploration and development geophysics. His professional interests are structural geology and reservoir characterization using seismic information.
Philippe Feugère was recently appointed as Manager of the Dedicated Processing Centre of Petroleum Development Oman on secondment from Companie Générale de Géophysique (CGG). He graduated as a Geological Engineer from the École Nationale de Géologie de Nancy in 1985. After working as an Asset Geologist in oil and gas exploration, he joined CGG Processing in 1990 and has gained processing experience in technical and managerial positions in various environments. These have ranged from Processing Geophysicist to Center Manager on CGG or Client sites worldwide, including Oslo, Kuala Lumpur, and Pau. Until recently, Philippe was Land Processing Department Manager for CGG in Massy, France.