51381schD1R1.p2.ps 04/24/2002 07:44 PM Page 2 The Time for Depth Imaging Many of today’s exploration targets cannot be seen with conventional seismicimaging methods. Operators now are getting a clearer picture—even of the most complex features—using prestack depth imaging. The more accurate results reduce exploration risk and help delineate reserves. Uwe Albertin Jerry Kapoor Richard Randall Mart Smith Houston, Texas, USA Gillian Brown Chris Soufleris Phil Whitfield Gatwick, England Fiona Dewey Wintershall Noordzee BV The Hague, The Netherlands Jim Farnsworth BP Houston, Texas Gary Grubitz BHP Billiton Houston, Texas Mark Kemme Clyde Petroleum Exploratie BV The Hague, The Netherlands For help in preparation of this article, thanks to Ian Anstey, Robert Bloor, George Jamieson, Patrick Ng and Erick Zubay, Houston, Texas, USA; and Mark Egan, Gatwick, England. 1. Fold is the number of source-receiver pairs whose signals constitute a trace. 2 Throughout the last century, interpreters accepted seismic images processed and displayed in the time domain. In many of today’s active exploration areas, especially where structures are complex and seismic velocities vary abruptly because of faulting or salt intrusion, time-domain processing can give misleading results; only depth imaging can define the true position and correct geometry of subsurface features. In some cases, the difference between depth and time images can make or break a prospect: structures gain or lose closure, targets move by hundreds of feet or meters, and reserves can be added or lost. The difference can be an expensive dry hole instead of a discovery. This article explains how depth imaging has emerged as the technique of choice for processing seismic data to image complex subsurface features. Case studies show how oil and gas companies operating in the Gulf of Mexico, North Sea and onshore US are improving their drilling success rates with depth imaging. Events in Seismic History Over the course of the 20th Century, notable milestones marked advances in seismic prospecting methods. Although many new technologies have taken about 10 years to mature from first introduction to accepted practice, each one has ultimately created new exploration opportunities. Starting in the 1920s, single-fold analog traces were introduced to detect dipping subsurface layers (next page).1 In the 1930s, this innovative technique was the key to discoveries around salt domes, and became standard practice. The 1950s saw the arrival of multiple-fold seismic data achieved by common depth-point (CDP) stacking, which markedly improved signal-tonoise ratio. In the 1960s, digital data acquisition and processing were introduced, replacing earlier analog and optical methods. This created major improvements in the quality of seismic data and led to many new discoveries worldwide. Throughout the 1970s, digital data and twodimensional (2D) surveys became common. Together, these technologies opened up the North Sea and other challenging areas. Timebased processing was standard, but 2D poststack depth migration was introduced and tested. The first small three-dimensional (3D) surveys were acquired over developed fields to improve reservoir delineation. In the 1980s, 3D surveys gained wide acceptance in the industry and transformed the exploration business. Trace attributes and bright spots were used as seismic indicators of hydrocarbons. By the 1990s, seismic contractors routinely acquired 3D exploration data over vast portions of the world’s continental shelves. Three-dimensional poststack time migration evolved to become standard practice, reducing finding costs to their current levels; and 3D prestack depth migration was introduced for particular cases. Today, many operators won’t drill without 3D data over their prospects, and in the areas of highest risk, won’t drill without prestack depth imaging. Currently, depth imaging is creating exploration opportunities in regions that were considered too risky just a few years ago. This technique is helping explorationists generate new subsalt prospects in the deepwater Gulf of Mexico and discover new reserves in the North Sea that were unimaginable using conventional time-processed data. Oilfield Review 51381schD1R1 04/24/2002 09:06 PM Page 3 1920 Single-fold analog shooting 1930 1940 1950 Multiple-fold common depth-point stacking Discoveries around salt domes Common depth point (CDP) Offset 2 3 Digital data acquisition and processing 4 Hyperbolic curve Two-way time 1 + Stacking velocity 1960 2D surveys 1970 3D surveys 1980 Subsalt prestack depth migration 1990 Multiclient 3D prestack depth imaging 2000 > Chronology of selected advances in seismic methods. Spring 2002 3 51381schD1R1.p4.ps 04/24/2002 07:44 PM Page 4 Introduction to Imaging Imaging is the process that brings seismic reflections into focus at their proper positions. It consists of two main elements—stacking and migration. Stacking increases signal-to-noise ratio by summing records obtained from several seismic shots reflecting at the same point. The simplest case to illustrate is a flat layer of uniform velocity overlying the reflector. Traces from several source-receiver pairs, centered on the reflection point but separated by different distances, or offsets, are gathered together (below). The variation in arrival time with offset is called moveout. The shape of the arrival times plotted against offset defines a hyperbola. Before the gather can be stacked, the traces must be shifted to align arrivals. The offset-versus-time parameter that describes the shifts defines the stacking velocity of that layer. The result of stacking is a single trace—the enhanced version of a signal that would have been recorded for a normalincidence, or zero-offset, shot at the midpoint of the source-receiver pairs. The second ingredient in imaging—migration—uses a velocity model to redistribute reflected seismic energy from its assumed position at the midpoint to its true position (next page, top left). One of the several classes of migration may be chosen depending on the complexities of the target and overburden structures. Simple structures and smoothly varying velocities can be imaged with simple migration routines that may fail to work on complex structures with rapidly varying velocities.2 Migration is accomplished by various solutions to the wave equation that describe the propagation of elastic waves through rock. Migration algorithms often take the name of their inventor, such as Kirchhoff, or the type of mathematical solution, such as finite-difference.3 Each type of migration has advantages and drawbacks. Migration can be performed in two domains—time or depth—and either before or after stacking. Certain imaging problems can be solved with time migration, but the most complex problems need depth migration. In time migration, the velocity model, also called the velocity field, may vary smoothly (next page, top right). The velocity model has two-way traveltime as its vertical axis. Seismic velocity increases with traveltime, and horizontal variations are gradual. Since these constraints are valid in most sedimentary basins, time migration is often applicable, and is used in most parts of the world. In depth migration, the velocity model may have strong contrasts horizontally or vertically. Depth migration is chosen when steeply dipping faults, folds or intrusions juxtapose layers with vastly different elastic properties. Depth migration needs an accurate velocity model in depth and is a more labor-intensive operation. Migration applied after stacking—poststack—is much faster than migration before stacking, because stacking reduces by an order of magnitude the number of traces that must be processed. For poststack migration to be successful, the assumptions made in stacking must be well-founded: the amplitude of the stacked trace must represent that of the normal-incidence trace and reflected arrivals must be approximately hyperbolic. These suppositions are valid only when variations in lithology and fluid content over the span of the gathered traces can be ignored and when the structure is simple. Any other conditions call for prestack migration. Performed before stacking, prestack migration can handle the most complex structures and velocity fields. In the past, the main constraints on prestack migration were the computing power needed and the time and skill required to construct the velocity model within a reasonable turnaround time. Advances in computing technology have eased these constraints. Creation of the velocity model still remains a time-consuming process and depends on the local geology. In areas where the geology is layered, or well-defined fault blocks exist, velocitymodel building for depth migration proceeds on a Offset 4 Offset 3 Offset 2 Offset 1 Zero offset Two-way time 1 Offset 2 3 Offset 4 1 2 3 4 Hyperbolic curve With stacking velocity + + + = Common depth point (CDP) Corrected CDP gather Stacked CDP > Stacking traces from a common depth-point (CDP) gather. Traces from several source-receiver pairs at different offsets from the common depth point are collected to form a CDP gather (left). Gathered traces are displayed in coordinates of time versus offset (center), in which the shape of reflection arrivals from a flat reflector defines a hyperbola. The arrivals are shifted into alignment using a stacking velocity, or offset-versus-time relationship, and stacked (right), or summed, to create a single trace with higher signal-to-noise ratio than that of any of the original traces. 4 Oilfield Review 51381schD1R1.p5.ps 04/24/2002 07:44 PM Page 5 Migrated trace Source Midpoint trace Receiver Simple velocities + simple structure = poststack time migration Simple velocities + complex structure = prestack time migration Complex velocities + simple structure = poststack depth migration Complex velocities + complex structure = prestack depth migration Original data Salt Increasing velocity MIG > Migration of reflected seismic energy. For this simplified two-dimensional (2D) example, migration repositions the data trace from its recorded position at the source-receiver midpoint to its true position (MIG) using a velocity model. In 3D cases, reflections may be redistributed to and from positions outside the plane containing the sources and receivers. > Simple and complex velocity models and structures treated by four migration classes—time, depth, prestack and poststack. Poststack models are on the left and prestack models are on the right. Models appropriate for time-based migration are on the top, and depth-based models are on the bottom. For time migration, the velocity model may have smooth variations, but only with depth, and only monotonically—always increasing with depth, never decreasing. Depth migration is required for more complex velocity models, such as those with lateral variation or decreases of velocity with depth. Poststack migration works with models of low structural complexity. Prestack migration can handle even the most complex models. layer-by-layer basis. An initial model is constructed from the most suitable data available, then updated through several iterations of prestack depth migration for each layer. The initial velocity model can be constructed using all the available information, such as stacking velocities, time-interpreted horizons and velocities from borehole data. Stacking and borehole velocities can show representative velocity trends, which should be taken into account in the model. When the structure is not too complex, the entire velocity model can be updated and constructed layer by layer rather quickly. In more complex cases, the velocity analyst defines blocks or other volumes bounded by faults or intrusions, then builds the model for each block layer by layer. In areas where geology is more continuous, such as in the Gulf of Mexico, a continuous sediment-velocity model is defined using either tomography or local velocity updating. Once the sediment velocity is defined, salt bodies are inserted after their positions are determined using several iterations of depth migration. In areas where anisotropy is an important factor, significant differences may appear between borehole-based velocities, which typically represent velocities in the vertical direction, and stacking velocities, which represent horizontal velocities. These differences must be accounted for by introducing anisotropy into the velocity model. More discussion on depth migration in anisotropic velocity fields appears later in this article. Collaboration between operator and service company can facilitate successful velocity model building. Operating company interpreters often have better knowledge and expectations of the subsurface, and can help interpret layer boundaries and salt features for the velocity model. Service company staff, with their knowledge of processing, incorporate these interpretations to help create the model for depth migration. Spring 2002 Depth Imaging in the Gulf of Mexico The Gulf of Mexico has been the most publicized proving-ground for prestack depth-migration techniques.4 Salt bodies in various stages of intrusion and uplift have created complex structures that both motivate and challenge explorationists. Salt geometries can vary enormously, and are critical in terms of hydrocarbon migration and trapping. Salt massifs can appear to be rooted to a deeper salt layer or completely detached and floating. The high contrast in seismic velocity between the salt at 14,500 to 15,200 ft/sec [about 4500 m/s] and sediments, often at less than half that value, causes problems for time-migration approaches. 2. For more on prestack, poststack, time and depth migration: 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): 28–41. 3. Kirchhoff migration is based on Kirchhoff’s solution to the wave equation. 4. Huang S, Ghose S, Sengupta M and Moldoveanu N: “Improvements in 3-D AVO Analysis and Structural Imaging of Dipping Salt-Flank Events Using AmplitudePreserving Prestack Depth Migration,” The Leading Edge 20, no. 12 (December 2001): 1328, 1330, 1332, 1334. Donihoo K, Bernitsas N, Dai N, Martin G and Shope D: “Is Depth Imaging a Commodity? The Impact of New Imaging Technologies and Web-Based Collaboration,” The Leading Edge 20, no. 5 (May 2001): 486, 488, 490, 492, 494, 496, 543. Albertin U, Woodward M, Kapoor J, Chang W, Charles S, Nichols D, Kitchenside P and Mao W: “Depth Imaging Examples and Methodology in the Gulf of Mexico,” The Leading Edge 20, no. 5 (May 2001): 498, 500, 502, 504, 506, 508, 510, 512–513. 5 51381schD1R1 04/24/2002 09:20 PM Page 6 Technological advances have brought increases in production ever since hydrocarbons were discovered in the Gulf of Mexico (below). Early on, drilling technologies were key to exploration success. More recently, seismicimaging techniques have helped sustain the discovery rate. In the late 1980s, operators started testing 2D prestack depth migration as a way to improve images of layers truncating against the flanks of salt domes. In 1993, partners Phillips Petroleum, Anadarko and Amoco were the first to announce a subsalt discovery in the Gulf of Mexico with the Mahogany prospect. They attributed this success to prestack depth imaging.5 Today, companies continue to explore below salt, and many are also looking in deeper water. Several of the recent large discoveries in the Gulf of Mexico are in deep water—deeper than 5000 ft [1500 m] (bottom left). In these areas, the cost of drilling a well can exceed $50 million, but the rewards can be great. The Crazy Horse discovery by BP contains estimated reserves of 1 billion barrels of oil equivalent (BOE). BHP Billiton has reported 200 to 450 million BOE recoverable reserves at Mad Dog and 400 to 800 million BOE at Atlantis. Cumulative discoveries Barges 50 Jackups Semisubmersibles Production, billion BOE 40 Hydrocarbon indicators 3D seismic, deepwater 30 Deepwater, subsalt 20 10 0 1940 1950 1960 1970 1980 1990 2000 2010 Year > Impact of technological breakthroughs on Gulf of Mexico success rates. Starting with the early achievements in drilling capability, and continuing through modern seismic methods, each advancement has yielded an identifiable increase in production. Reducing risk is a key concern for deepwater operators, and prestack depth imaging is one of the technologies that help reduce risk. Depth imaging was performed to reduce drilling risk over many Gulf of Mexico discoveries such as Crazy Horse, Llano, Mad Dog, Atlantis and others. For BHP Billiton, prestack depth imaging is the critical technology for reducing risk and appraisal at Atlantis, Mad Dog and the rest of the Western Atwater Fold Belt trend that encompasses these discoveries. BP credits prestack depth-imaging breakthroughs with helping to describe the elements of the Crazy Horse prospect and to position the discovery well.6 Imaging a seismic volume containing a salt body is different from traditional processing, in which data tapes are sent off for processing and a finished product is returned to the interpreter for examination. Subsalt imaging requires several iterations of migration and interpretation (below). Many of these steps are based on proprietary processing techniques, allowing contractors to differentiate their results from those of other contractors. The first step after general prestack processing is to build the initial velocity model for the layers overlying the salt. In the Gulf of Mexico, sediments typically are sand-shale sequences without strong velocity contrasts between layers. The initial velocity model can often be derived from stacking velocities to produce a smooth interval-velocity field describing the sediments. Prestack processing Lake Charles Analyze velocities Houston Edit distorted velocities New Orleans Build depth-and spacevariant gradients 3D poststack migrate on fine grid B G D C A H E Discovery wells Older wells Salt A B C D E F G H Crazy Horse Mars Crazy Horse North Ursa Atlantis Mad Dog Mahogany Llano > Recent deepwater Gulf of Mexico discoveries, with many occurring near salt bodies. Large recent discoveries have estimated reserves in the hundreds of millions of barrels. Several of these have been discovered with the help of prestack depth imaging. 6 Update velocity model Analyze migration velocities Define salt geometry F GULF OF MEXICO 3D prestack migrate on fine grid 3D prestack migrate on sparse grid or depth window 3D prestack depth migrate entire volume for final image > Data-processing flow for subsalt prestack depth migration. The process is a complex interplay of several steps. Building the velocity model itself requires iterations in prestack depth migration to define the velocity of each layer and the geometric boundaries of each layer. Oilfield Review 51381schD1R1.p7.ps 04/24/2002 07:57 PM Page 7 Poststack Time Imaging Prestack Depth Imaging > Comparison of time migration and depth migration in the Green Canyon area of the Gulf of Mexico. The time migration (left) shows two salt bodies, each uplifting and doming the overlying sediments. The salt body on the left has a domed top and a flat base, and creates a shadow beneath. The one on the right seems to be in two pieces: a floating salt pillow has detached from the dome below. Prestack depth imaging (right) retains the general shape of the body on the left, although its base is now sloping. However, the depth imaging reveals layers beneath, which were shadowed in the time migration. The salt intrusion on the right has a completely different shape when depth migrated. Instead of rising in an anticlinal structure, sediments are truncated along the flanks of an hourglass-shaped salt body. The second step takes this velocity model and updates it. Velocity analysts have several ways of revising models, most belonging to a category of methods called tomographic inversion. Tomography uses traveltime information derived from seismic data to refine velocity models. Classical reflection tomography uses the difference between predicted and observed reflection traveltimes.7 Ray tracing predicts the arrival times of reflections on common depth-point gathers at control points. On each gather, the shape of the actual arrival time of the shallowest reflector is compared with the predicted arrival times, and the velocity that best flattens the actual arrival times is used to update the model. This step is time-consuming and requires experts in both processing and interpretation to create a model that satisfies the data at all control points. The next step applies depth migration using the updated velocity model. The migrated traces are gathered again and arrival flatness checked. If the preliminary time migration shows the top of Spring 2002 salt to be smooth, or structurally simple, the velocities of the overburden can be used in a poststack depth migration to obtain an image of the top of salt. If the top of salt is rough, or structurally complex, prestack depth migration should be applied. After the top of salt is imaged and interpreted, the velocity model is updated by filling the volume below the salt top with a uniform salt velocity. With this new velocity model, the volume is again prestack depth migrated, and the bottom of salt comes into focus. Applying the correct migration technique can bring surprising changes to the seismic image. Interpretation of one time-migrated section from the Green Canyon area of the Gulf of Mexico shows two anticlinal structures created by salt intrusion (above). The salt body on the left has a domed top and a flat base, and creates a shadow beneath, obscuring deeper reflections. The salt intrusion on the right appears to have pierced through the top of the anticline and left a dome of salt behind. With prestack depth imaging, the picture changes completely. The salt body on the left is still domed, but is thicker, with a sloping base. Layers can now be seen below the salt. The salt feature on the right looks entirely different. Instead of two disconnected salt bodies, the new image shows a single hourglass-shaped body with clearly delineated sides and base. Instead of rising in an anticlinal structure, sediments are truncated along the flanks of the salt hourglass. 5. Westcott ME, Leach MC, Wyatt KD, Valasek PA and Branham KL: “Mahogany: Seismic Technology Leading to the First Economic Subsalt Field,” Expanded Abstracts, 65th SEG International Meeting and Exposition, Houston, Texas, USA (October 8-13, 1995): 1161–1164. For more on subsalt exploration: Farmer P, Miller D, Pieprzak A, Rutledge J and Woods R: “Exploring the Subsalt,” Oilfield Review 8, no. 1 (Spring 1996): 50–64. 6. Pfau GE, Chen RL, Ray AK and Kapoor SJ: “Seeing Through the Fog: Improving the Seismic Image at Crazy Horse,” presented at the AAPG Annual Meeting, March 10–13, 2002, Houston, Texas, USA. Yielding CA, Yilmaz BY, Rainey DI, Pfau GE, Boyce RL, Wendt WA, Judson MH, Peacock SG, Duppenbecker SD, Ray AK and Hollingsworth R: “The History of a New Play: Crazy Horse Discovery, Deepwater Gulf of Mexico,” presented at the AAPG Annual Meeting, March 10–13, 2002, Houston, Texas, USA. 7. Other types of tomography can use refracted or transmitted waves. 7 51381schD1R1.p8.ps 04/24/2002 07:58 PM Page 8 Poststack Time Imaging Prestack Depth Imaging > Time and depth migrations of three large salt features. Poststack time migration (top) reveals the tops of the salt intrusions. However, this method leaves an unclear image of the salt bases; they might be interpreted at the lower limit of the zone that has little reflection or no character. Interpretation of the prestack depth image (bottom) suggests that the two left-most salt bodies are not floating, but connected to roots that extend to 40,000 ft [12,200 m]. 8 In another portion of the Green Canyon area, the tops of three large salt pillows are fairly clearly imaged by poststack time migration, but the bases are not (left). One reasonable interpretation would place the bases of the salt at the lower limit of the reflectionless area of the seismic image. There is some indication of layering between the salt bodies at great depth. Prestack depth migration reveals a surprisingly different image. The two large salt bodies on the left now are connected to roots that plunge to about 40,000 ft [12,200 m]. The root of the middle salt feature is about 3 miles [5 km] across. The vast volume between the two salt roots is filled with dipping sediments that are truncated against the roots. One of the achievements of the WesternGeco approach to prestack depth migration is the ability to image dips “beyond 90 degrees,” that is, layers that are overturned or below salt overhangs. Migration methods trace rays through the velocity model to a reflector, then follow the rays back to the surface. The rays bend at each interface according to the angle of incidence and the velocity contrast between layers. Usually it is sufficient to consider only those rays that bounce from the top side of a reflector. However, in some cases, reflections of interest can also occur from the bottom side—as in the case of reflections from underneath salt overhangs. Properly accounting for these reflections in migration requires the ray tracing to be done over long distances. By taking advantage of these rays, called turning rays, the undersides of salt overhangs can be imaged clearly. In another example from the Gulf of Mexico, poststack time migration is able to image the northern flank of a salt intrusion, but the southern side is lost in a shadow (next page, bottom). The time migration did not use turning rays. Prestack depth imaging, incorporating turning-ray energy as well as energy passing through the salt, illuminated the steeply dipping layers and the overhanging salt on the south flank of the intrusion. Imaging in the North Sea The Gulf of Mexico is not the only place where operators are using depth imaging. Many parts of the North Sea can claim structural complexity rivaling that of Gulf of Mexico salt intrusions. In addition to tectonically active salt, North Sea basins exhibit expanses of chalk and largescale faulting above and below the salt. The smoothly varying sand-shale sequences overlying the Gulf of Mexico salt bodies may seem simple by comparison. Oilfield Review 51381schD1R1.p9.ps 04/24/2002 07:45 PM Page 9 Wintershall Noordzee BV began exploring in Blocks K10 and K13 in the Broad Fourteens basin of the Dutch sector of the North Sea in 1968 (right). Since then, more than 30 wells have been drilled, delineating seven producing fields. As these fields enter their final years of production, new technology is being deployed to identify additional reserves and extend the producing life of this mature area.8 The area is structurally complex, with largescale normal faulting, overthrusts and salt intrusions. Large velocity contrasts around the salt domes and across major faults cause traditional seismic-imaging methods to produce poor pictures of structures and faults. Deep channels cut into the Tertiary sequence that overlies a thick chalk unit of variable thickness and velocity. The main reservoirs are even deeper—the Main Buntsandstein and Rotliegend sandstones. Highamplitude carbonate rafts can be mistakenly interpreted as Top Rotliegend reflectors, resulting in false targets. 6° 54° < The K10 and K13 blocks in the Broad Fourteens basin, southern North Sea. Wintershall Noordzee BV has achieved a clearer seismic picture of their reservoirs in this gas-producing region by applying prestack depth imaging. K10 K13 53° NORTH SEA Ijmuiden The Netherlands 52° 8. Dewey F, Whitfield P and King M: “Technology Offers New Insight in a Mature Area—A 3D PreSDM Case Study from the Dutch N Sea,” Transactions of the EAGE 63rd Conference and Technical Exhibition, Amsterdam, The Netherlands, June 11–15, 2001, paper A-04. 3D Poststack Time Imaging 3D Prestack Depth Imaging South North South North Turning-ray reflection Constant velocity Varying velocity Salt Spring 2002 > Imaging under a Gulf of Mexico salt overhang with time and depth migration. Poststack time migration (left) manages to image the north side of a salt diapir, but the southern side is lost in a shadow created by an overhang. By including turning rays (inset) and rays that pass through the salt, prestack depth migration (right) images the steeply dipping layers and the overhang on the south side of the intrusion. 9 51381schD1R1.p10.ps 04/24/2002 07:45 PM Page 10 An early depth-migration project in 1996 over a 50-km2 [19.3-sq. mile] portion of the two blocks showed some imaging improvements, but used simplistic velocity-modeling techniques, and so the results lacked the detail to improve fine-scale structural imaging. Enhanced imaging and increased resolution were required to improve understanding of the geological history of the area and identify remaining traps. In 1999, Wintershall and WesternGeco carried out a high-fidelity 3D prestack depth migration of both blocks. The new project incorporated data from three 3D surveys covering an area of 880 km2 [340 sq. miles]. The success of every depth-migration project depends on the accuracy of the velocity model. To create an accurate model, a meticulous approach was developed using a combination of state-of-the-art tools and more conventional techniques. Iterative layer-stripping formed the backbone of the analysis. For each layer, a combination of tomography and multivelocity depth scans was Depth Imaging 1996 used to derive the model. To verify the velocities of each layer, a dense grid of 3D prestack depth migrations was generated. The depth stacks were used to update the structural model, and the gathers were examined to monitor and update the velocities. This allowed both the structural and velocity variations to be continuously and consistently tracked and checked for each of the 11 layers in the model as it was built. The new images showed significant improvements over the previous time- and depthmigrated data sets, especially in the tectonically complex areas. For example, results from the 1996 project using a simple velocity model gave an unclear image of the top of the Rotliegend sandstone reservoir under a complex fault (below). The new depth migration with the detailed velocity model gave a much clearer image of this potential reservoir interval. In a second example, a feature that is difficult to interpret in the time-migrated section becomes identifiable as a “pop-up” of the Rotliegend formation in the properly depth-migrated image (next page, top). What appears to be an isolated discontinuous reflection in time migration can be seen in the depth-migrated section to be an abrupt pop-up with near-vertical sides. The complex structure overlying the pop-up, combined with its steeply dipping flanks, makes this problem difficult to solve with time migration, but completely tractable with depth migration. The project’s success relied on close cooperation between processing geophysicists, interpreters and researchers from both Wintershall and WesternGeco, along with the optimization of all available technologies. The additional effort put into deriving the detailed velocity model has shown the benefits of aiming for the 90% correct solution rather than making do with a 70% result, while satisfying tight operational time and cost constraints. A complete reinterpretation of the area is under way and will be combined with a basinmodeling study to improve definition of the producing fields and identify the presence of any untested reservoir compartments. Depth Imaging 1999 Rotliegend sandstone > Comparison of depth migrations with simple and complex velocity models. Depth migration for an earlier project used a simple velocity model, and produced an unclear image of the top of the Rotliegend sandstone under a complex fault (left). Depth migration with the newer, more detailed velocity model gives a much clearer image of the potential reservoir interval (right). 10 Oilfield Review 51381schD1R1.p11.ps 04/24/2002 07:46 PM Page 11 Time Imaging Depth Imaging > Complex Rotliegend structure revealed by depth imaging. A disrupted interval in the time-migrated section (left) is difficult to interpret. In the depth-migrated image (right), this becomes identifiable as a pop-up of the Rotliegend formation. Q4 Q5 in as Coastline Q8 sB Q7 en rte ou dF oa Br Building Reserves through Depth Imaging In another North Sea development, operators used depth imaging to improve delineation of reserves and increase reserve estimates. Clyde Petroleum and partners recently deployed state-of-the-art depth imaging in a renewed effort to explore, appraise and extend existing gas discoveries in Blocks Q4 and Q8 of the Dutch North Sea (right). The recently discovered Q4 gas fields lie in a complex inversion zone (once low-lying, now upthrust along reactivated faults) bounded by a series of major NW-SE-striking faults. The new fields are on a trend with two producing gas fields in the Q8 block. Before Clyde Petroleum began operating the block, seven dry wells had been drilled on shallower prospects. The tectonic history had produced highly deformed structures, and early conventional seismic processing gave suboptimal results. After the drilling of the first successful exploration well, a new program called for a comprehensive 3D prestack depth migration, followed by complete reinterpretation.9 54° No inversion Low reservoir risk Basin margin terrace Low reservoir risk Inverted terrace Moderate reservoir risk Deeply inverted terrace High reservoir risk Basin axis, maximum burial and inversion Very high reservoir risk Gas 53° NORTH SEA 52° Q4 Q5 Q7 Q8 Ijmuiden The Netherlands 9. Kemme M, Brown G, VanBuuren N and Greenwood M: “Depth Imaging Unfolds Complex Geology and Impacts Reserves—The Q4 Story,” Transactions of the EAGE 63rd Conference and Technical Exhibition, Amsterdam, The Netherlands, June 11–15, 2001, paper P071. > Gas reservoirs (red) operated by Clyde Petroleum Exploratie BV in the Q4 and Q8 blocks of the Dutch sector of the North Sea. Color-coding shows regions with different tectonic histories. Spring 2002 11 51381schD1R1.p12.ps 04/24/2002 07:46 PM Page 12 The complex history of normal, reverse and lateral movements had placed the basin fill on top of the reservoir block. Time-migrated images of these steeply dipping structures were limited in quality, and fault positioning was questionable. Image ray-tracing and boreholeseismic results indicated that lateral fault mispositioning could be as much as 300 m [1000 ft], depending on the overburden velocity model. The prestack depth-migration project was initiated to better understand the structural framework and correctly position faults, with the expectation that the results could have a strong impact on the size of the structure and the planning of the development wells. Four 3D data sets, approximately 400 km2 [154 sq. miles] of seismic data, were input to the prestack depth migration. Each data set was processed through a similar, conventional preprocessing flow with emphasis on noise reduction and multiple attenuation. Although the individual data sets had different orientations, no resampling was required. Phase matching and amplitude compensation were applied to each survey to match all surveys to a common base.10 Each data set was depth migrated individually and merged after migration, but before stack. Due to the complex nature of the geology, major velocity contrasts were expected. Therefore, the traditional top-down, layer-stripping approach to velocity modeling was not considered adequate. The structural model indicated that the 3D velocity model could be divided into five separate NW-SE-trending velocity blocks, with up to six velocity layers below the area-wide Tertiary layer on top (below). Within each block, velocity was determined layer by layer, but the inclination of the fault blocks dictated the order in which the velocity model should be built—from southwest to northeast. Typically, stacking velocities would be used to derive initial interval velocities for a particular layer. However, due to low confidence in the stacking velocities in this complex area, a model based on well data was used. A range of velocity maps based on the starting velocity was used to output a 3D prestack depth-migrated grid of inlines over the target area.11 A final velocity map for the target layer was then derived by interactively picking on the depth-migrated common image-point gathers.12 Finally, a 500-m [1640-ft] grid of 3D prestack depth-migrated in-lines and crosslines was generated. These lines were used In-line 3600 0 3800 4000 4200 4400 4600 4800 5000 5200 5400 5600 5800 6000 6200 SW 6400 NE 500 1000 1500 Zone 1 Zone 2 Zone 3 Zone 4 Zone 5 Depth, m 2000 2500 3000 3500 Tertiary 4000 Chalk Lower Cretaceous 4500 Upper Jurassic Lower Jurassic 5000 Triassic Permian to interpret the target horizon in depth, for inclusion in the velocity model. This procedure was iterated layer by layer within each fault block until the base horizon had been inserted into the velocity model. Then, the final velocity model was used to generate a full 3D prestack depth-migrated volume on a 25-m by 25-m [82-ft by 82-ft] grid. Residual moveout correction was performed, the data were stacked, and appropriate poststack processing was applied. The new depth data showed notable improvements over the time-migrated data, and increased the interpreter’s understanding of the structural model and confidence in the fault positioning (next page, top). The prestack depth migration enabled targeting of the second exploration well near a major fault without risk of encountering a reduced reservoir section, and revealed that the position of the fault was farther to the west, increasing the reservoir volume. This improved imaging also had a significant impact on the interpretation of the eastern bounding fault. Because of poor imaging of the traditionally migrated seismic data, this fault had been imaged as an easterly dipping normal fault. However, the superior resolution of the new images shows that the reservoir-bounding fault is actually a westerly dipping reverse fault, adding an extra fault block of gas-bearing reservoir. The updated structural interpretation resulted in an increase of almost 50% in gas initially in place (next page, bottom). Additionally, better seismic definition decreased uncertainty in the reserves estimate and allowed for detailed interpretation of faults within the reservoir, reducing the risk of leaving compartments undrained. The robust methodology followed throughout the project allowed the construction of an accurate velocity model for this complex area. The subsequent 3D prestack depth-migrated volume provided a significant improvement in the quality and confidence of the seismic image. As a result of the improved seismic quality, not only did the apparent volume of the structure increase significantly, but also the better data quality resulted in a much more detailed interpretation of intrareservoir faults. This allowed for more reliable planning for three to five future development wells. The Q4-A field came on-stream in December 2000, only 21⁄2 years after the first exploration well was drilled. Pre-Permian flood > Velocity model for the Q4 reservoir blocks. Steeply dipping faults laterally juxtapose contrasting velocities and place high-velocity layers on top of lower velocity layers. The red box delineates the area of interest. 12 Oilfield Review 51381schD1R1.p13.ps 04/24/2002 07:46 PM Page 13 Time Imaging 1996 Prestack Depth Imaging 1999 GWC > Comparison of interpreted time- and depth-migrated seismic lines over the reservoir in the Q4 block. Interpretation of the time-migrated image (left) shows a block of reservoir bounded on the west by a thrust fault (yellow), and on the east by an easterly dipping normal fault (black). Interpretation of the depth-migrated image (right) changes the picture. The new interpretation raises the thrust fault (white line), adding volume to the reservoir on the west. The normal fault on the east is no longer considered a bounding fault. The reevaluated reservoir boundary is a westerly dipping reverse fault (red), previously not recognized. The approximate gas-water contact (GWC) is indicated. Prestack Depth Migration on Land Many onshore prospects have the same imaging problems encountered offshore, but until recently, land-based seismic campaigns were less successful at imaging complex structures. However, depth-imaging projects on land are now showing the same level of improvement over time-based methods as their Gulf of Mexico and North Sea counterparts. Exploration in south Texas is notorious for complications caused by complex structures overlying potential reservoirs. Faults create shadows that obscure the shape and disposition of deeper layers. Imaging targets in “fault shadows” is a challenge with time-migration techniques, but depth migration gives much clearer pictures and more geologically reasonable features. 10. Phase and amplitude of seismic traces are affected by the timing and power characteristics of the acquisition source and by processing, which may vary from one survey to another. Combining data sets from different surveys requires phase and amplitude of all data sets to be matched. 11. An in-line is a seismic line within a 3D survey parallel to the direction of towed-streamer acquisition. A crossline is a seismic line perpendicular to the direction of survey acquisition. 12. A common image-point (CIP) gather is the set of all traces that reflect at the subsurface point being imaged. A CIP gather is created by searching over all possible rays in the acquisition geometry and collecting those that reflect at the point of interest. Spring 2002 0 feet 6560 0 meters 2000 < Increase in gas initially in place resulting from interpretation of depth-migrated seismic data. Interpretation of depth-migrated seismic data moved faults and added roughly 50% to the gas reserves in this reservoir. Old fault interpretations are shown in black; new fault interpretations are shown in blue. The increase in reservoir size is shown in pink. Q4-A Time-migrationbased field outline Gained area New prestack depth-migrationbased faults Old time-migrationbased faults Old time-migrationbased field outline Q4-B 13 51381schD1R1.p14.ps 04/24/2002 07:46 PM Page 14 Time Imaging Depth Imaging 10,000 2.2 Depth, ft Time, msec 2.4 12,000 2.6 14,000 2.8 > A time-migrated (left) and depth-migrated section (right) from south Texas. In the time-migrated image, velocity complexities cause a false anticline immediately to the left of the fault plane denoted by arrows. Also, the reflections on the left side of the fault appear to be broken and have less continuity than reflections on the right side of the fault. The depth-migrated section shows gently dipping and continuous structures in the fault shadow. The false structural high seen in the time-migrated data has become smoother, and reflection continuity is improved. One example of benefits gained through prestack depth migration is from a regional, 100-sq. mile [256-km2] WesternGeco multiclient survey in south Texas. A conventional timemigrated image across a large normal fault shows some of the typical problems seen in this area (above). A pronounced false anticline, or “pull-up” of the seismic reflections, appears beneath the fault on this section. Also, the reflections beneath the fault appear to be broken and have less continuity than reflections on the right side of the fault, particularly along the interpreted horizon. These imaging problems are caused by the juxtaposition of rocks of different velocities on opposite sides of the fault (next page, top left). The layers on the upthrown, or left, side of the fault, although older than the layers on the right, are overpressured, and so have lower seismic velocities. The lateral velocity contrasts cause seismic rays to bend as they cross the fault. Ray bending distorts the seismic image in the time domain. The depth-migrated section shows a different picture. The reflections in this section dip less steeply on the left side of the fault than do the corresponding reflections in the time-migrated section. The false structural high has diminished, and reflection continuity is improved. An interpretation of the depth-migrated section produces a different depth and shape to the layers beneath the fault, potentially yielding a different exploration target. 14 Depth migration has been successful in other parts of the world where land-seismic results are known to be problematic. WesternGeco has performed 3D depth-imaging projects in many of the world’s oil-producing countries, including Venezuela, Bolivia, Argentina, Germany, Russia, Kazakhstan, Egypt, Libya, Kuwait, the United Arab Emirates, Syria, China, Australia and Nigeria. Reaching Full Potential Today’s methods are more accurate than earlier ones, but the full potential of depth imaging has not yet been reached. The limitations to overcome center around creation of the velocity model, deciding what type of migration produces the best images, and the time required for completion of depth-imaging projects. Several factors can complicate the modelbuilding process. One that has received recent attention is anisotropy. Much of the subsurface is anisotropic in some physical property, such as elastic properties, permeability or electromagnetic properties.13 The simplest form of elastic anisotropy is called transverse isotropy (TI). This occurs when the seismic velocity has one value parallel to bedding and a different value perpendicular, or transverse, to bedding. In typical cases of TI anisotropy, velocity parallel to bedding is greater than transverse velocity. Usually, seismic data processing ignores anisotropy. However, the effects of strong anisotropy can produce a suboptimal data set if not taken into consideration. Ignoring anisotropy can result in vertically and horizontally mispositioned structures. The effects of anisotropy can be seen as a nonhyperbolic shape in the arrivals from a flat reflector (next page, top right). Traces from long offsets arrive earlier than predicted from a model with isotropic velocity because they have traveled longer in the faster, horizontal direction. Anisotropy can be incorporated into a prestack depth-migration velocity model, with conspicuous results (next page, bottom).14 Prestack depth imaging with an isotropic velocity model produces a fairly clear image of the sediment layers uptilted by a North Sea salt intrusion. However, the layers in the shadow of the salt overhang are not as clear as they could be, and the gently dipping layers on the lower flank of the salt show a mis-tie with formation depths measured in a well. Prestack depth imaging with a model that includes 10% anisotropy in the overburden produces a clearer image and one that ties with well data. Identifying which imaging problems require anisotropic velocity models and which ones are simply displaying velocity heterogeneity will become easier as more areas are tested. Processing experts debate which type of migration is best for imaging extremely complex volumes. Prestack Kirchhoff migration has been particularly effective in salt and subsalt imaging in the Gulf of Mexico, but sometimes it has difficulty imaging features under rugose salt bodies. Because this algorithm uses ray tracing, small errors in the shape or location of the salt interface can cause migration artifacts. Oilfield Review 51381schD1R1.p15.ps 04/24/2002 07:59 PM Page 15 10,736 10,000 Offset 12,000 Two-way time 12,464 Depth, ft Interval velocity, ft/sec 11,000 11,696 13,000 13,232 Anisotropic Isotropic > Comparison of arrival times versus offset for an isotropic and an anisotropic layer. If the layer were isotropic, the arrivals would define the red curve, and if the layer were anisotropic, the arrivals would define the black curve. 14,000 14,000 > Depth-migration velocity model for south Texas survey, showing the fault interpreted on seismic data. In such areas, finite-difference prestack migration can provide effective imaging. This approach uses wavefield extrapolation instead of ray tracing, and can produce better images.15 Efficiency gains and the use of larger computer systems have shortened project cycle times. But service companies continue to be pressured to depth-image larger areas and to do it quickly. Oil companies and contractors should share the responsibility to define realistic time frames. Depth migration brings a viable solution to complex imaging problems. After seeing the difference between depth-imaged data and conventional time-imaged sections, operators often change their interpretations and plans, whether for prospect exploration or reservoir develop- Isotropic Depth Imaging ment. Furthermore, seeing the difference in one seismic section forces the realization that all other data acquired in complex areas probably deserve a second look. Some operators now insist on depth imaging before drilling in deep water or other high-risk areas. Other operators are reluctant to apply depth imaging because of the costs involved in acquiring and processing target-specific data. To them, it appears that this technology is only for the super-major operators. However, it is possible to use depth imaging in a cost-effective manner on multiclient projects to improve understanding of regional geological petroleum systems. The WesternGeco approach to applying depth imaging on speculative regional-scale data sets is helping make the technology available to operating companies of all sizes. As more operators gain experience with the technique, the process will become more efficient. Experts predict that in the future, essentially all seismic data will be depth imaged. —LS 13. Armstrong P, Ireson D, Chmela B, Dodds K, Esmersoy C, Hornby B, Sayers C, Schoenberg M, Leaney S and Lynn H: “The Promise of Elastic Anisotropy,” Oilfield Review 6, no. 4 (October 1994): 36–47. 14. Bloor R, Whitfield P and Fisk K: “Anisotropic Prestack Depth Migration and Model Building,” Transactions of the EAGE 63rd Conference and Technical Exhibition, Amsterdam, The Netherlands, June 11–15, 2001, paper A-01. 15. Albertin U, Watts D, Chang W, Kapoor SJ, Stork C, Kitchenside P and Yingst D: “Improving Near-Salt-Flank Imaging with Shot-Profile Wavefield-Extrapolation Migration in the Gulf of Mexico,” to be presented at the EAGE 64th Conference and Technical Exhibition, Florence, Italy, May 27–30, 2002. Anisotropic Depth Imaging Well top Well top > Prestack depth imaging in the North Sea with an isotropic (left) and an anisotropic (right) velocity model. Including 10% anisotropy in the velocity of the overburden helps to produce a clearer image of the layers that are truncated against a salt intrusion and produces a better depth match to well data. Spring 2002 15