New Dimensions in Land Seismic Technology

Transcription

1 New Dimensions in Land Seismic Technology Advanced seismic acquisition and processing technology has come onshore. A high-fidelity, high-resolution, integrated single-sensor system is now available for use on land. This technology marks a significant step forward for exploration, field development and production. Malik Ait-Messaoud Mohamed-Zerrouk Boulegroun Aziza Gribi Rachid Kasmi Mahieddine Touami Sonatrach Algiers, Algeria Boff Anderson Peter Van Baaren WesternGeco Dubai, UAE Adel El-Emam Ghassan Rached Kuwait Oil Company Kuwait Andreas Laake Stephen Pickering WesternGeco Gatwick, England Nick Moldoveanu WesternGeco Houston, Texas, USA Ali Özbek Cambridge, England For help in preparation of this article, thanks to Mark Daly, Jean-Michel Pascal Gehenn, Will Grace, Dominic Lowden and Tony McGlue, Gatwick, England; Mark Egan and Norm Pedersen, Houston, Texas; Zied Ben Hamad, Lagos, Nigeria; Mahmoud Korba, Algiers, Algeria; and Andrew Smart, Kuwait. DSI (Dipole Shear Sonic Imager), Q-Borehole and VSI (Versatile Seismic Imager) are marks of Schlumberger. Omega2, Q-Land, Q-Marine, VIVID and Well-Driven Seismic are marks of WesternGeco. Seismic technology has achieved amazing feats in exploration and production activities in the past few decades. The advance to threedimensional (3D) seismic acquisition and imaging of the subsurface, introduced in the 198s, was perhaps the most important step. 1 Another was development of four-dimensional (4D), or time-lapse, seismic data to monitor how reservoir properties such as fluids, temperature and pressure change during the productive life of a field. 2 Introduction of multicomponent seismic data acquisition with the recording of shear wave signals, in addition to compressional wave data, provided a tool for rock characterization and identification of pore-fluid types. 3 With the world s ever-growing demand for oil and gas, the emphasis in the oil and gas industry has shifted to exploring deeper, more complex reservoirs and to enhancing production from existing assets. Field life can be extended by delineating bypassed oil and gas and by placing production and injection wells optimally. The proactive monitoring of reservoir fluid behavior saturation and pressure over time, allows remedial actions to be implemented before production is affected. For all these applications, the geophysicist, geologist and reservoir engineer require reliable and repeatable data of exceptional resolution that can be fine-tuned to a specific reservoir objective. Exceptional data resolution means having data with increased frequency bandwidth and low coherent and noncoherent noise, while preserving signal fidelity. 4 For decades, the battle between signal and noise has driven the seismic industry to seek ways to suppress noise and to enhance signal. The signal is a true representation of the actual reflection that corresponds to changes in rock characteristics such as lithology, porosity and subsurface structure. Both noise, which can be coherent or noncoherent, and absorption of higher frequencies by the Earth obscure the true nature of the signal. This article examines a new, integrated singlesensor acquisition and processing system that delivers measurements that were previously unobtainable with conventional recording of seismic data. Examples from producing assets in Kuwait and Algeria illustrate the superior quality of these data in terms of signal fidelity and frequency bandwidth in comparison with data acquired with conventional methods. Challenges in Conventional Land Acquisition Single-sensor seismic recording has been available since the early days of seismic exploration. The principle is simple. An impulse source such as dynamite or a controlledfrequency source such as a vibrating plate on a truck sends acoustic energy into the Earth. 5 This energy propagates in many different directions. Downward traveling energy reflects and refracts when it encounters boundaries between two materials with different acoustic properties. Sensors or geophones placed on the surface measure the reflected acoustic energy, 42 Oilfield Review

2 Digital Group Forming Field Tape Hard Disk/ Processing Conventional Data Q-Land Data converting it into an electrical signal that is displayed as a seismic trace. 6 A complication in land acquisition is that, unlike marine data, a seismic line is rarely shot in a straight line because of the presence of natural and man-made obstructions such as lakes, buildings and roads. More importantly, variation in ground elevation causes sound waves to reach the recording geophones with different traveltimes. The Earth s near-surface layer may also vary greatly in composition, from soft alluvial sediments to hard rocks. This means that the velocity of sound waves transmitted through this surface layer may be highly variable. Static corrections a bulk time shift applied to a seismic trace are typically used in seismic processing to compensate for these differences in the elevations of sources and receivers and near-surface velocity variations. 7 Another major problem in land acquisition is that land sources typically generate energy that travels horizontally near the surface, also known as airwaves and ground-roll noise. Conventional sensor arrays comprising strings of geophones 1. Beckett C, Brooks T, Parker G, Bjoroy R, Pajot D, Taylor P, Deitz D, Flaten T, Jaarvik LJ, Jack I, Nunn K, Strudley A and Walker R: Reducing 3D Seismic Turnaround, Oilfield Review 7, no. 1 (January 1995): Aronsen HA, Osdal B, Dahl T, Eiken O, Goto R, Khazanehdari J, Pickering S and Smith P: Time Will Tell: New Insights from Time-Lapse Seismic Data, Oilfield Review 16, no. 2 (Summer 24): Barkved O, Bartman B, Compani B, Gaiser J, Van Dok R, Johns T, Kristiansen P, Probert T and Thompson M: The Many Facets of Multicomponent Seismic Data, Oilfield Review 16, no. 2 (Summer 24): Coherent noise is unwanted seismic energy that shows consistent phase from one seismic trace to another. This may consist of waves that travel through the air at very low velocities such as airwaves or air blast, and ground roll that travels through the top of the surface layer, also known as the weathering layer. The energy trapped within a layer, also known as multiples, is another form of coherent energy. Noncoherent energy is typically nonseismic-generated noise, such as noise from wind, moving vehicles, overhead power line or high-voltage pickup, gas flares and water injection plants. 5. A vibrator source sends a controlled-frequency sweep into the Earth. The recorded data are then convolved with the original sweep to produce a usable signal. 6. Each trace consists of one recording corresponding to a single source-receiver pair. In practice, traces from one source are simultaneously recorded at several receivers. Then, sources and receivers are moved along the survey line and another set of recordings is made. When a seismic wave travels from a source to a reflector and then back to the receiver, the elapsed time is the twoway traveltime. The common depth point (CDP) is the halfway point of the path; it is situated vertically below the common midpoint. Sorting of traces by collecting traces that have the same subsurface midpoint is called a common-midpoint (CMP) gather. The number of traces summed or stacked is called a fold. For instance, in 24-fold data, every stacked trace represents the average of 24 traces. In the case of dipping beds, there is no common depth point shared by multiple sources and receivers, so dip-moveout (DMO) processing becomes necessary to reduce smearing or inappropriate mixing of data. For more on seismic recording: Farmer P, Gray S, Whitmore D, Hodgkiss G, Pieprzak A, Ratcliff D and Whitcombe D: Structural Imaging: Toward a Sharper Subsurface View, Oilfield Review 5, no. 1 (January 1993): Ashton CP, Bacon B, Mann A, Moldoveanu N, Déplanté C, Ireson D, Sinclair T and Redekop G: 3D Seismic Survey Design, Oilfield Review 6, no. 2 (April 1994): Ongkiehong L and Askin HJ: Towards the Universal Seismic Acquisition Technique, First Break 6, no. 2 (1988): Autumn 25 43

3 7 m Seismic source Geophone 2.8 m 4.16 m Receiver array length: m > Conventional acquisition onshore. Seismic energy recorded at the receivers arrives at different times because of elevation differences and near-surface velocity variations (top). In conventional acquisition, strings of geophones hard-wired together average the individual sensor measurements and deliver one output trace, whose location is denoted by the center of gravity of the array, indicated by the red dot (bottom). The resulting output trace has a generally lower frequency than each of the input signals, and the amplitude is smaller than the sum of the individual amplitudes, a phenomenon known as the array effect. t sampling rate Undersampled seismic signal Properly sampled seismic signal Receiver array > Aliasing effect. Sampling at a frequency lower than the highest frequency present in the signal (red curve) results in insufficient samples to capture all the peaks and troughs present in the data. Not only will inadequate sampling miss the information on higher frequencies but the signal will be incorrectly defined (blue curve). are based on the assumption that the upward traveling energy, or the reflected wave, arrives at the array essentially vertically and simultaneously, while the surface-wave noise arrives mainly horizontally and sequentially. To cancel this source-generated noise, spatially distributed receiver groups arrays are summed. 8 Ideally, this process results in an attenuation of the noise and an improvement of the signal. However, there are problems associated with conventional arrays. In reality, the sensor array is often not located on flat and homogeneous ground, so local changes in elevation and surface geology lead to fluctuations in the signal arrival time (left). These fluctuations are called intraarray perturbations. The hard-wired sensor array instantaneously sums all traces, and in the case of intra-array perturbations, this would lead to a partial cancellation of signal. The resulting output trace would then be at a lower frequency than each of the input signals, and the amplitude would be smaller than the sum of the individual amplitudes, a phenomenon known as the array effect. Aliasing is a well-known problem that arises when the sampling rate of a signal is inadequate to capture the higher frequencies in the signal. 9 Not only is the information contained in higher frequencies lost, it is incorrectly represented (below left). Aliasing is a consideration for spatial sampling too, not just temporal sampling. Ground roll typically contains many different wavelengths related to the distance between successive peaks in a waveform that are shorter than the typical group interval or the distance between receiver array centers of gravity in a conventional survey. Due to undersampling of ground-roll energy, this energy is aliased and passed on to the signal bandwidth, causing ambiguity between signal and noise. 8. Newman P and Mahoney JT: Patterns With a Pinch of Salt, Geophysical Prospecting 21, no. 2 (1973): Aliasing is the ambiguity that arises because of insufficient sampling. It occurs when the signal is sampled less than twice the cycle. The highest frequency defined by a sampling interval is termed the Nyquist frequency and is equal to inverse of 2 t, where t is the sampling interval. Frequencies higher than the Nyquist frequency will be folded back or wrapped back. This condition can be observed in video or motion pictures: the spoke wheels on horse-drawn wagons sometimes appear to be turning backward instead of forward. Aliasing can be avoided by a finer spatial sampling that is at least twice the Nyquist frequency of the waveform. 1. El-Emam A, Moore I and Shabrawi A: Interbed Multiple Prediction and Attenuation: Case History from Kuwait, presented at the 25 SEG International Exposition and 75th Annual Meeting, Houston, (November 6 11, 25). 11. Roden R and Latimer R: An Introduction Rock Geophysics/AVO, The Leading Edge 22, no. 1 (October 23): Oilfield Review

4 Tests of varying array lengths have shown the degradation in signal quality caused by increasing the array size (right). For longer offset receiver arrays, the arrival time of the signal can vary significantly at either end of the array, smearing the higher frequencies when summed within the group. Therefore, just as adequate temporal sampling of the recorded trace is needed to successfully record a given frequency, a sufficiently small group interval is required to record a particular spatial frequency. A problem common to all seismic acquisition is energy trapped between subsurface layers, known as interbed multiples, caused by a strong velocity contrast between layers. This occurs when the energy from the source reflects more than once in its travel path. Interbed multiples are analogous to a bouncing ball trapped between two layers, which continues to bounce until it loses its energy. Borehole seismic data, which are acquired when sources are placed on the surface and receivers are anchored in a borehole, help identify the interfaces that generate these interbed multiples. Recent developments in datadriven methods and the use of vertical seismic profile (VSP) data to guide surface-seismic multiple attenuation, such as Interbed Multiple Prediction (IMP), seem promising. 1 The quality of the raw seismic dataset is fundamental to achieving superior frequency resolution and high signal-to-noise ratio. Amplitude and phase preservation of the input signals is critical in all facets of stratigraphic interpretation, including prestack seismic inversion, amplitude variation with offset (AVO) and amplitude variation with angle (AVA) interpretation. An analysis of the variation in reflection amplitudes with source-geophone distance or offset gives some valuable insights into reservoir properties such as lithology, porosity and pore fluid. 11 Because land data often exhibit poor signal-to noise ratios arising from irregular geometries and noise contamination, a fundamental change in acquisition and processing methods was required. Two-way traveltime, s Point Sensors at 2 m Airwave 2 4 Offset, m First array First break Reflected waves Ground roll Group interval 16-m Array 2 4 Offset, m Second array 32-m Array 2 4 Offset, m Third array Channel 1 Channel 2 Channel 3 > Signal degradation with an increase in array size. A point-source, point-receiver test was carried out with single sensors spaced 2 m [6.6 ft] apart and a single vibrator source. A 16-m [53-ft] array was formed by summing groups of nine consecutive geophones and assigning the summed signal to a channel placed in the center of gravity of the array (bottom). The group interval is the distance between consecutive channels. Similarly, a 32-m [15-ft] array was formed by summing groups of 17 consecutive geophones. By using 2-m sensor spacing, wave types were recorded without aliasing (top left). As the sensors were grouped into arrays of increasing length of 16 m (top middle) and 32 m (top right), first the airwave, then the ground roll and finally the first breaks became aliased, manifested as cross-banded signal areas in the shot domain. Aliasing also manifests as a wrap-around of the noise energy in the frequency-wavenumber domain, not shown here. (Data courtesy of Shell.) Autumn 25 45

5 A Change in Acquisition Philosophy In the early 199s, WesternGeco began extensive research on compressional wave (P-wave) sensitivity that led to a fundamental change in acquisition philosophy. Experiments conducted on synthetic signals revealed the effects of source and receiver statics, recording electronics specifications, source phase distortion and receiver sensitivity on P-waves (below). Excluding coherent source-generated noise, ambient noise and source frequency sweep, the dominant effects on the signal-to-noise ratio are due to perturbations that could not be corrected for within an analog array. Factors such as source and receiver statics, coupling of the geophone to the ground, geophone position and tilt, source positioning, and amplitude and phase distortion in the sources were more important than hardware changes in the geophone or the recording system itself. A small error of 1 ms in receiver statics results in the introduction of -29 db of noise relative to the signal. Such static errors are commonly found within a conventional analog receiver group. System Receiver Source Statics Perturbation Distortion Gain tolerance Synchronization Harmonic distortion Geophone sensitivity Natural frequency Temperature Coupling Tilt Sensor position Harmonic distortion Amplitude Phase Source position Receiver statics Source statics Knowledge gained from these experiments was used to design and build the Q-Land singlesensor land seismic system to reduce the effects of these perturbations while addressing the issue of coherent noise removal such as ground roll. A receiver spacing that is half (or less) of the ground-roll wavelength would be adequate to sample ground-roll noise without aliasing. Just as temporal aliasing arises from insufficient sampling in time, a large receiver interval leads to spatial aliasing. The new Q-Land system digitizes each sensor at the recording location (next page, top). To achieve this fine spatial sampling, the recording system requires a massive increase in the number of live channels. A live channel means that the receivers are connected to record simultaneously. Compared with a typical high channel-count conventional system, which may have 4, to 5, channels that record live, the new point-receiver acquisition system has 2, or more live channels. The Q-Land system is the first to implement an integrated point-receiver acquisition and processing methodology. Signal error in db Signal error, %, 95% confidence interval > P-wave sensitivity chart for land acquisition. Experiments were conducted on synthetic signals to understand the effect of perturbations such as source and receiver statics, recording electronics, source phase distortion, and receiver sensitivity. The chart shows that hardware changes in the receiver or the recording system have low signal error, as compared with other factors that cause a significantly higher signal error. The ability to correct for these higher order perturbations allows for the preservation of signal fidelity and bandwidth within the seismic data. -1 The same concept is applicable to the seismic sources. The source array can be replaced by point sources. In addition, to prevent aliasing in the common midpoint domain, the source interval should be small, ideally equal to the receiver interval. The new point-source, pointreceiver recording technique replaces the conventional method of using sensor and source arrays to attenuate noise and to improve signalto-noise ratio. 12 Recording seismic data through point receivers rather than analog receiver arrays has several potential advantages, including better static solutions, velocity estimation, amplitude preservation, bandwidth retention and noise attenuation. This point-source, point-receiver methodology increases data volume by more than an order of magnitude. Advances in data transmission and computing power have enabled the development and deployment of this cost-effective, high channel-count recording system. A New Integrated Acquisition and Processing System The new Q-Land system is a 2, live-channel seismic acquisition and processing technology. The typical sampling rate for the system is 2 ms. However, the Q-Land system can record with 3, live channels when the sample rate is changed to 4 ms. Digital recording of the incoming wavefield at densely spaced receiver positions ensures that the recorded signal and noise are properly sampled and are therefore unaliased. In Q-Land acquisition geometry, one source line and one receiver line that are orthogonal to each other form a cross-spread. These are then repeated spatially within the acquisition area. Each source-receiver pair generates a trace that corresponds to a subsurface midpoint. If the midpoints corresponding to all source-receiver pairs are binned, with a bin size equal to half the receiver by half the source interval, every bin will be one midpoint corresponding to single-fold coverage. Thus, the cross-spreads provide singlefold subsets of the continuous wavefield, sampled finely enough to prevent aliasing of the coherent noise, through which a cross-spread volume is generated (next page, bottom). 12. Ongkiehong and Askin, reference Christie P, Nichols D, Özbek A, Curtis T, Larsen L, Strudley A, Davis R and Svendsen M: Raising the Standards of Seismic Data Quality, Oilfield Review 13, no. 2 (Summer 21): Oilfield Review

6 Sophisticated algorithms are then applied in a processing technique called digital group forming (DGF). DGF comprises three main steps. The first is correction to each geophone for intraarray perturbations such as amplitude, elevation differences and near-surface velocity variations. After the geophone outputs are grouped, the result is a signal with a frequency bandwidth similar to that of the individual traces and an amplitude almost equal to the sum of the individual amplitudes. This step is similar to the one applied in the Q-Marine single-sensor marine seismic system. 13 Sensors 1,824 receivers per line > The Q-Land acquisition and processing system. A line of receivers is laid out perpendicular to a line of sources and every source point is recorded by every receiver point. The example shows 1 receiver lines that are 2 m [656 ft] apart, with 1,824 point receivers per receiver line that result in 18,24 live receivers (top). In digital group forming using the Omega2 software processing system, the seismic traces from individual geophones have perturbation corrections made to each geophone (bottom). Data-adaptive filters are then applied over a number of traces to suppress coherent noise. An output trace from a number of sensors can then be produced at the desired spatial sampling. Receiver lines 2 m apart Receiver line Source line Sources Digital signals from individual sensors Field acquisition system Source line Receiver line Time Area of common midpoint coverage Receiver line Source line Digital group forming Hard disk/ processing > A three-dimensional (3D) display of the cross-spread volume. A cross-spread configuration is achieved by deploying receivers along a line in one direction and placing sources along an orthogonal line (right). Each source-receiver pair generates information from a subsurface point that, for a flat surface, is located at the midpoint between source and receiver (gray area). In this example of cross-spread configuration, with receiver sampling at 5 m [16 ft] and source sampling at 2 m [66 ft], the subsurface coverage is single fold. A 3D view of the cross-spread volume shows that the ground-roll noise is confined within a conical shape volume, making its removal or attenuation by 3D filters in the frequency-wavenumber domain more effective (left). Autumn 25 47

7 The second step applies data-adaptive filters for noise suppression. Noise attenuation can include, but is not limited to, coherent and ambient-noise attenuation, high-voltage power line pickup cancellation, and airwave and flarenoise attenuation. There are different ways to attenuate noise using digital filtering techniques. However, the design of optimal 3D digital filters is important to realize the potential of pointreceiver recording. An ideal filter would pass all the desired frequencies in the pass band with no distortion, and completely reject all frequencies outside the range of interest, called the stop band. The ideal spatial anti-alias filter response would also be azimuthally isotropic, that is the array response would be the same for energy arriving from all angles. There are two problems associated with anti-alias filter performance for conventional data acquisition: imperfect rejection of azimuthally varying levels of noise in the stop band and an imperfect flat response in the pass band (below). The Q-Land technique of forming an orthogonal acquisition geometry into crossspreads is well suited to the application of threedimensional anti-alias filters. A filtering technique based on the APOCS method alternating projections onto convex sets is an effective approach that works optimally on crossspread geometry. 14 Magnitude, db Wavenumber k y, 1/m Wavenumber k x, 1/m db > Three-dimensional spatial anti-alias filter response. The problem of unwanted noise contaminating the signal bandwidth area is illustrated. The spatial anti-alias filter response displays amplitude on the vertical axis, and wavenumbers along the two horizontal axes, k x and k y, in x and y directions. The color represents the magnitude in db. An efficient filter would pass the signal that is at around k=, and stop or reject any noise for all other directions for k. For a conventional 16-m receiver array, noise leaks into the signal from almost all azimuths (left). In contrast, for point-receiver data, the anti-alias filter using the APOCS filter design technique shows the effectiveness of the filter in rejecting noise (right). Magnitude, db Wavenumber k y, 1/m Wavenumber k x, 1/m 14. A well-known mathematical technique, APOCS is an iterative technique that derives filter parameters to remove coherent noise. The algorithm, working in 3D space, switches constantly between sample domain with time on one axis and x and y on the other two axes and frequency transform domain with frequency on one axis and wavenumber in x and y directions, k x and k y, on the other two. Wavenumber is the inverse of wavelength and represents the frequency of the wave in space. For more on APOCS: Özbek A, Hoteit L and Dumitru G: 3-D Filter Design on a Hexagonal Grid for Point-Receiver Land Acquisition, EAGE Research Workshop, Advances in Seismic Acquisition Technology, Rhodes, Greece, September 2 23, 24. Quigley J: An Integrated 3D Acquisition and Processing Technique Using Point Sources and Point Receivers, Expanded Abstracts, 24 SEG International Exposition and 74th Annual Meeting, Denver, (October 1 15, 24): Shabrawi A, Smart A, Anderson B, Rached G and El-Emam A: How Single-Sensor Seismic Improved Image of Kuwait s Minagish Field, First Break 23, no. 2 (February 25): The Q-Borehole system optimizes all aspects of borehole seismic services, from job planning through data acquisition, processing and interpretation. Well logs, VSP and surface-seismic data are combined to build a property model of vertical velocities, frequency attenuation factors, anisotropy related to vertical variations in interval velocities and the multiples wavefield. The model is then used for enhanced surface-seismic processing and calibration in the Well-Driven Seismic process. 17. A zero-offset VSP is acquired when a seismic source is placed on the surface close to the wellhead and receivers are positioned at different depths in a borehole. In a walkaway VSP, an array of receivers collects data for multiple source positions located along a line that extends from the wellhead. For more on VSP and walkaway VSPs: Arroyo JL, Breton P, Dijkerman H, Dingwall S, Guerra R, Hope R, Hornby B, Williams M, Jimenez RR, Lastennet T, Tulett J, Leaney S, Lim T, Menkiti H, Puech J-C, Tcherkashnev S, Burg TT and Verliac M: Superior Seismic Data from the Borehole, Oilfield Review 15, no. 1 (Spring 23): Oilfield Review

8 The last step is spatial resampling of the output data according to the desired group interval. Analog arrays, once laid out in the field, have almost no flexibility to adjust the outputsampling interval, whereas with digital group forming, any output sampling is possible down to the granularity of the individual sensors. While data from conventional arrays may provide reasonable results for structural interpretation, detailed reservoir analysis using seismic inversion or AVO techniques is limited to a narrow frequency band because of aliased noise wrapping back in the frequency range of interest (below right). With such reduced bandwidth, AVO and inversion are unlikely to produce meaningful results. The densely spaced point receivers employed by Q-Land methodology provide alias-free data, and hence, a more complete bandwidth for AVO interpretation. In complex geological settings where conventional array data are unable to deliver the required results, single-sensor data provide significant improvement in signal fidelity and frequency bandwidth. This improvement enables interpretation of subtle stratigraphic features and increased vertical and lateral resolution of the seismic response as demonstrated in the following two examples from Kuwait and Algeria. In addition, the Minagish field posed an unusual operational hazard. The area was strewn with unexploded cluster bombs and mines from earlier military activity. A detailed understanding of the internal reservoir structure was essential for a planned water injection scheme to work. Forward seismic modeling using rock properties from core samples and logs has shown that a 5 to 95% change in water saturation could result in a 5% difference in acoustic impedance a product of velocity and density of the rock. However, a previous 4D study in 1998 shows the inability to detect these small changes due to the background noise level in conventional seismic data. Some of the limiting factors were frequency resolution, inferior noise attenuation and a low signal-to-noise ratio. To allow monitoring of minute changes in reservoir behavior, it was obvious that a step change in acquisition methodology was necessary to reduce noncoherent signal and coherent noise m Array Four tightly grouped vibrators in a rectangle of 12.5 m [41 ft] by 5 m [16.4 ft] vibrated in synchrony at 6% of their peak force capability of 8, lbf [356 kn]. Driving at less than peak force provided a low distortion in the seismic source. The vibrators were set as close as possible to resemble a point source while maximizing energy input into the Earth. The Q-Land system recorded 14,94 channels at a 2-ms sampling rate. 15 Perturbation corrections were made to each receiver and each source prior to summation in the DGF process. In addition, a Q-Borehole integrated borehole seismic study was planned at the inception of the Q-Land pilot program. 16 A zero-offset VSP and two walkaway VSPs recorded the data around the center of the survey area. 17 The integration of surface seismic and borehole geophysical data was vital to ensure that all steps in the processing sequence, from digital group forming through to the final migrated stack, were optimally calibrated utilizing some of the latest 6 Point Sensors at 2 m Pioneering New Technologies in Kuwait The Minagish field in southwestern Kuwait was selected for a Q-Land pilot study in 24, to address several development and exploration objectives. One goal was to provide a detailed image of multiple reservoir intervals within the Cretaceous for fluid-front monitoring. Discovered in 1959, the Minagish field is one of the country s main producers, with production primarily from carbonate rocks, including the Minagish oolites. A waterflood program resulted in water influx overriding oil in layers with high permeabilities. A previous 3D seismic survey in 1996, with source and receiver arrays of 5-m [165-ft] spacing, provided poor imaging of deeper prospects and limited the vertical and lateral resolution at principal reservoir zones. Characterizing fracture density and orientation, necessary to optimally place horizontal wells and maximize production, was also a problem. Noise arising from gas flares and water injection plants, coupled with seismic-generated noise such as air blast, ground roll and multiples, caused extreme distortions in the seismic data. Frequency, Hz Aliased noise Usable bandwidth Wavenumber, 1/km Wavenumber, 1/km > Impact of aliasing on frequency bandwidth. A test conducted with a 16-m receiver array displays aliasing of ground roll and airwave because of the wrap-around effect seen in the frequencywavenumber (fk) domain (left). The airwave (solid black line) is completely aliased. However, the ground-roll noise (dashed black line) is folded back into the signal frequency band above the frequency where the dashed lines intersect. The signal, which is expected to dominate the central area of the fk plot as k approaches zero, becomes contaminated. This means that data-adaptive spatial filtering can no longer remove the coherent noise without damage to the signal. The usable frequency bandwidth for AVO processing, for example, is substantially reduced for conventional array data because aliasing distorts the higher frequencies in both amplitude and phase. In contrast to this, point-receiver data clearly show an unaliased response that permits processing of the entire useful frequency range without coherent noise contamination (right). (Data courtesy of Shell.) Frequency, Hz Usable bandwidth Autumn 25 49

9 Conventional 3D Data 1,4 Q-Land Data Two-way traveltime, ms 1,5 1,6 1,7 1,8 > Comparison of conventional 3D seismic data and Q-Land data in the Minagish field, Kuwait. The Q-Land data (right) show a much higher lateral and vertical resolution as compared with conventional seismic data (left). The Minagish target reservoir appears at about 1,5 ms. developments in the Well-Driven Seismic process. 18 Restoration of true amplitude and phase, effective multiple suppression and compensation for frequency absorption with depth provided superior imaging and resolution (above). The zero-offset VSPs at two control wells, an injector and a producer, resolved seven intrareservoir zones. Conventional seismic data, with a frequency bandwidth of 1 to 45 Hz, showed only three of these events, which led to a flawed interpretation that there were no obstructions between the two wells and injected fluids could flow freely between two wells. The Q-Land volume imaged the same seven intrareservoir zones seen on the VSPs. The enhanced resolution from Q-Land data, with a frequency bandwidth of 6 to 7 Hz, enabled the seismic interpreters to map stratigraphic features. Also identified were tar mats at the injector well that act as baffles and inhibit fluid movement. In addition, minor faults and deeper gas targets obscured by interbed multiples energy were now imaged The Well-Driven Seismic process uses borehole seismic data for true amplitude and phase recovery, velocity analysis, multiple attenuation, anisotropic migration and angle-based muting. For more on the Well-Driven Seismic technique: Morice SP, Anderson J, Boulegroun M and Decombes O: Integrated Borehole and Surface Seismic: New Technologies for Acquisition, Processing and Reservoir Characterization; Hassi Messaoud Field, presented at the 13th SPE Middle East Oil and Gas Show and Conference (MEOS), Bahrain, June 9 12, El-Emam et al, reference 1. Encouraged by the results of this Q-Land pilot study, the operator is in the process of planning a full-field survey using the Q-Land system. Plans to reevaluate pore pressure and fracture characterization incorporating the new Q-Land data are also under consideration. The Seismic Challenge in Algeria An oil field in Algeria, known to be one of the most seismically challenging fields in the world, was selected for a Q-land study. Since the discovery of this field in the 195s, many wells have been drilled. Oil and gas production is mainly from Cambro-Ordovician fluviomarine, clastic reservoirs. Despite the large number of wells drilled, abrupt changes in lithology and fault compartmentalization have made full-field reservoir characterization from well data alone difficult. Few seismic surveys have been attempted in the past because of a poor seismic response and failure to image the reservoir zones. As a result, reservoir zones were identified from petrophysical and pressure data. In addition, a weak correlation between well log and core permeability indicated that the permeability could be strongly influenced by fractures. There are several geophysical and geological challenges. The main producing reservoir, a braided-channel fluvial system, has a highly heterogeneous distribution of sand and shale. The oil field has also been affected by multiple episodes of fault deformation and reactivation, resulting in complex fault and fracture patterns that are difficult to image. Added to these problems, a small velocity and density contrast at the top of the reservoir and within the reservoir units makes detection of reservoir units difficult. In addition, the influence of strong interbed multiples obscures the signal, and the presence of a thick evaporite layer above the reservoir causes severe attenuation of higher frequencies, resulting in a poor signal-to-noise ratio. All these problems lead to a poor tie to wells, making it extremely difficult to map the interwell region. Typically, the maximum usable frequency obtained from the target reservoir has been around 35 to 4 Hz. This translates into a maximum vertical resolution of 4 m [131 ft]. However, mapping the reservoir units with any degree of confidence requires a vertical resolution of less than 2 m and much lower noise levels. 5 Oilfield Review

10 A pilot survey with the Q-Land system was initiated to address these geophysical and geological challenges. Integration of boreholeseismic data and surface-seismic data was planned at the onset of the project, and the acquisition parameters were optimized through presurvey planning and testing. Q-Land seismic data were acquired over an area covering 44 km 2 [17 mi 2 ], with a dense grid of sensors, equating to a density of 2, sensors per km 2. Borehole geophysical data included measurements of zero-offset VSP, a twodimensional (2D) walkaway VSP using the VSI Versatile Seismic Imager with 154 geophone positions in the borehole, and sonic measurements using the DSI Dipole Shear Sonic Imager. The Q-Borehole seismic system aided in Well-Driven Seismic processing. The surface-seismic processing results were compared with well data at key stages in the processing sequence, so that the processing parameters were optimized to tie the final seismic data to the wells. The bandwidth obtained ranged from 6 Hz to 8 Hz, nearly double the previously recorded high-resolution 2D seismic results. For the first time, the frequency resolution obtained from surface seismic data matched that obtained from a VSP, providing an excellent well tie (below). High-Resolution 2D Data Q-Land Data X,5 X,6 Two-way traveltime, ms X,7 X,8 X,9 Y, Y,1 Distance Distance VSP X.4 X.5 Two-way traveltime, s X.6 X.7 X.8 Distance Power, db Signal Frequency, Hz > A Q-Land example from Algeria. Exceptional resolution (frequencies above 8 Hz) was obtained with the Q-Land survey (top right), in which the frequency bandwidth has nearly doubled in comparison to a high-resolution 2D survey (top left). In addition, the excellent match between the vertical seismic profile (VSP) data (shown within the red box, bottom) and Q-Land data will permit advanced reservoir characterization studies. (Data courtesy of Sonatrach.) Autumn 25 51

11 Normalized acoustic impedance Low High Two-way traveltime, s X.9 Y. Y.1 > Acoustic Impedance (AI) cross section. The Hercynian unconformity forms the top of the reservoir zone (dashed line). The vertical thickness of the low AI interval within the reservoir section indicates a thickness in the range of 1 to 15 m [33 to 49 ft]. (Data courtesy of Sonatrach.) Seismic amplitude was inverted to compute absolute acoustic impedance (AI) volume (above). Low AI correlates reasonably well with highporosity sands. At 8-Hz frequency, for an interval velocity of about 4,5 m/s [14,765 ft/s], this low AI zone equates to a thickness resolution of about 14 m [46 ft]. This degree of resolution has never been achieved in this geological environment. To assess the relationship between permeability and fault proximity, which is generally associated with higher fracture density, several seismic attributes were computed. Extraction of fractures and faults from the seismic data involved several steps. Seismic attribute cubes that enhance discontinuities in the data, also known as edge-enhancing attributes, were computed. The edge-detection seismic volumes include variance, dip and deviation. The ant-tracking algorithm was then applied to the edge-detection cube to highlight 2. The ant-tracking algorithm delineates discontinuities in a seismic cube and maps faults and fractures. The algorithm tracks discontinuities built upon previous knowledge, mimicking the behavior of ants when they find the shortest path between their nest and their food source. The ants communicate using pheromones, a chemical substance that attracts other ants. Therefore, the shortest path to the food source will be marked with more pheromones than the longest path, so the next ant is more likely to choose the shortest route, and so on. The idea is to distribute a large number of these electronic ants in a seismic volume. Ants deployed Well Acoustic Impedance Distance 33 to 49 ft Reservoir section XX, Depth, ft XY, the discontinuities in the seismic data, and to map faults and fractures. 2 Distance to fault (DTF) attributes were then generated from filtered sets of faults from the ant-track cube and mapped into the 3D geocellular grid (next page). The DTF attribute helps identify zones that are highly fractured. A crossplot between permeability and DTF confirms the trend: higher well log permeability when close to faults. A strong inverse relationship between core permeability and DTF was observed on about 7% of wells. However, to answer questions about whether those fractures and small-scale faults enhance or degrade permeability, grid cells were extracted in the vicinity of seismic faults with greater length, that is, those that intercept both basement and the overlying Hercynian unconformity. Seismic acoustic impedance was then mapped into these cells to discriminate along a fault should be able to trace the fault surface for some distance before being terminated. The algorithm then automatically extracts the result as a set of faultpatches, a highly detailed mapping of discontinuities. Fault discrimination is based on the size of the fault, its orientation, and on the amplitude of vertical displacement. For more on ant tracking: Pedersen SI, Randen T, Sønneland L and Steen Ø: Automatic Fault Extraction Using Artificial Ants, Expanded Abstracts, 22 SEG International Exposition and 72nd Annual Meeting, Salt Lake City, Utah, USA (October 6 11, 22): between sealing and draining faults. A high cellaverage acoustic impedance in the vicinity of a fault suggests that the fractures act as flow barriers, because the fractures were cemented with pyrite or shale. Conversely, a low acoustic impedance in the vicinity of a fault suggests a higher proportion of open, fluid-filled fractures, which have lower density than rock. This may suggest that tectonically induced fractures enhance draining of hydrocarbon. Wells are continually being drilled in this area, and additional drilling is planned in 26 guided by the interpretation results of the Q-Land data. Toward Fit-For-Purpose Seismic Data The fundamentally improved measurements delivered by Q-Land technology radically expand the potential of seismic data. With the lower noise associated with single-sensor acquisition and processing, and the ability to correct for perturbations within a group, array design and fold are no longer the dominant factors in improving signal-to-noise ratio. Rather, sensor spacing and a requirement to adequately sample coherent noise become the main drivers in designing the acquisition geometry. Since it is now possible to recover a signal more faithfully, the vibrator source can also be reevaluated, making it possible to record shorter, single sweeps of frequency with a better sampling of the wavefield. These design considerations now offer the possibility of acquiring point-source and pointreceiver exploration surveys with a lower field effort, compared with equivalent surveys using conventional source and receiver arrays. The Q-Land surveys acquired to date suggest that the use of smaller groups of vibrators can provide data that are equal to or better than larger arrays of vibrators and geophones. Smaller groups of vibrators allow for more efficient operation. The Q-Land VIVID imaging services enhance the value of seismic data recorded throughout the life of a field. In the exploration phase, the low-noise Q-Land data make it possible to acquire high-quality seismic surveys with wider line spacing and lower fold than a survey acquired with conventional technology and still meet or exceed the imaging expectations. In subsequent surveys for appraisal or development, it is possible to acquire the data by interleaving lines between the previous surveys to build up the fold. The data from the original surveys and the current survey are processed together using 52 Oilfield Review

12 Combined Acoustic Impedance and Distance to Fault Attributes High Acoustic Impedance Along Faults Low Acoustic Impedance Along Faults Flow barriers Fractures enhancing permeability > Relationship between seismic acoustic impedance and permeability. Seismic acoustic impedance (AI) and distance to fault (DTF) attributes are combined and mapped onto a geocellular volume (top). A dual filter based on proximity to fault and seismic AI threshold value is applied to the volume. The filtering assumes that fractures that are open and fluid-saturated have lower velocity and density, and therefore lower AI (bottom right). These are discriminated against faults that are cemented, with higher AI caused by silicification or pyrite filling (bottom left). (Data courtesy of Sonatrach.) the required group interval to image the target correctly. This is the concept of uncommitted seismic data for the life of a field. Exploration leads can be pursued during the same survey. This results in a lower cumulative environmental footprint of the overall seismic program, in addition to reduced development time. As the data have a high signal-to-noise ratio and fidelity, they can be reused at each stage of a field s development, ensuring that the exploration investment is not lost. With unsurpassed seismic data quality and a versatile approach to acquisition geometry and innovations in processing, the Q-Land acquisition and processing system will have a major impact on the life of the field, in exploration, development and reservoir monitoring. RG Autumn 25 53