Exploration Technology

Wide- and multi-azimuth acquisition: Issues and answers

Gathering the data is only half the battle; processing and interpreting it is the other half; doing it cost-effectively is the third half.

Shuki Ronen and Phil Fontana, VeritasDGC, Houston

Three things have become increasingly clear in recent years regarding seismic acquisition. One is that increased data density, whether through sources or sensors, is good. The second is that, especially under complex overburden, some aspects of structure and rock properties appear different depending on shooting direction, and this is not predictable. Thus, wider and more varied azimuths are desirable. The third is that these two things are even better when combined and are not independent of advanced processing schemes, such as wave equation migration and surface-related multiple attenuation. And of course, oil companies want it all fast and at a lower cost.

GARDEN BANKS EXAMPLE

A clear demonstration of imaging targets under complex overburden with different shooting directions was presented by Bruce ver West in 2001.¹ In 2000, Veritas acquired 3D seismic data over a substantial area of 16 offshore blocks in the Garden Banks area of the Gulf of Mexico, over which the company already had non-exclusive (multi-client) data from 1996. There was nothing technically wrong with the 1996 survey and the 2000 survey used the same technology as the 1996 survey. Why would the company spend money to cover the same area again? The answer to this question is that the company wanted to test the hypothesis that wide-azimuth geometry would provide improved illumination under salt.

The 1996 survey was acquired in the east-west direction, and the 2000 survey was in the north-south direction. We reproduce here a few figures from the 2001 expanded abstract, Figs. 1 – 3. Ver West, et al., found that some sub-salt reflectors are imaged better with one direction, and other reflectors are imaged better with the orthogonal direction. The analysis showed that the differences in image quality came from different illumination of subsalt reflectors by the two orthogonal surveys. Remember the early days’ debates about dip versus strike? Now think subsalt, where the overburden has dips in all directions. Even simple structures under complex overburden may not be illuminated if we limit ourselves to narrow-azimuth geometry, Fig. 4.

Fig. 1. Improved subsalt imaging with Wave Equation Migration compared to Kirchhoff. Seismic waves scatter in the salt wedge, so, reflections from the same subsalt area with the same shot and receiver locations often have more than one arrival time. Kirchhoff methods depend on ray tracing and have more difficulties under salt than WEM. Yet, the WEM image is still poor subsalt due to compromised illumination.

Fig. 2. Some of the salt province in the deepwater Gulf of Mexico. There is no defined dip and strike direction to the lumps of salt which have flowed between other sediments.

Fig. 3. Acquisition in both directions is better than in any one direction; in fact, acquisition in all directions would be ideal.

Fig. 4. Conventional towed-streamer acquisition geometry. The source(s) at the head of the streamer spread produce a very well-sampled offset distribution, but in a very restricted azimuth zone.

The Garden Banks example was an early trial of what is now called multi-directional geometry. It can, of course, be extended to more than two orthogonal directions as in Fig. 5, which provides better illumination. However, it does not provide all the azimuths and, more importantly, it does not provide data that are well-suited for advanced processing methods, like Wave-Equation Migration (WEM) and Surface-Related Multiple Elimination (SRME).

Fig. 5. Multi-directional surveys use conventional narrow-azimuth towed-streamer geometries shot multiple times in multiple directions. The technique provides better sub-surface illumination than single-pass surveys, but does not provide well-sampled common shot or common receiver gathers, or a grid of coincident shot and receiver points for SRME.

THE PROCESSING/ ACQUISITION CONNECTION

SRME requires data sampled with a shot or a receiver in all locations between any other shot and receiver locations. This is for the purpose of predicting all surface-generated multiples that occur in various locations around the shot-receiver line. State-of-the-art towed-streamer geometry does not provide data suitable for SRME, and extensive interpolation techniques are used with considerable compromise to the output quality. The ideal geometry for SRME would be shots and receivers everywhere. Narrow-azimuth towed-streamer geometry does not provide the required data and neither does multi-direction geometry.

Wave equation migration (WEM) is a family of imaging methods that are based directly on the wave equation and do not assume that a reflection from any point has a single primary arrival. This is important for imaging targets under complex overburden, because the scattered waves often have a few paths and a few arrival times, even if we look at data from a single source and receiver location (a so-called single seismic trace) and are interested in imaging one reflection point at a time.

Other imaging methods – which we do not call WEM – are based on the wave equation only indirectly via ray tracing and the Kirchhoff approximation. Kirchhoff methods are very effective to image complex targets under simple overburden, but they fail to image many targets, simple or complex, under complex overburden such as salt. WEM methods that are based on wavefield extrapolation and ideally require the data to be sorted into common shot or common receiver gathers. An ideal input to WEM would be common shot or common receiver gathers, well-sampled in both offset and azimuth, Fig. 6.

Fig. 6. Common shot or common receiver gather, fully sampled in both offset and azimuth.

Sorting conventional streamer data into common receiver gathers is not an option, due to poor sampling of the shot points in the crossline direction. For WEM, streamer data are sorted into common shot gathers. However, these common shots are long, narrow and are not adequately sampled. A conventional state-of-the-art towed-streamer geometry would have 8 to 20 streamers, each 4- to 12-km long, with groups of hydrophones just 12-m long or single hydrophones at just 3-m spacing. The distance between streamers is generally between 50 to 100 m. But, at most, their spread is about 1-km wide. When such data are imaged with WEM, there are many edge effects, because everything is close to an edge.

One solution for these problems would be a mega-vessel capable of towing 30, 40, or 50 long streamers, with spread widths of 3, 4, or even 5 km. However, in terms of applied hydrodynamics and common sense economics, this is not a very practical approach. What we can do, however, is manipulate existing towed-streamer capabilities to simulate very wide receiver spreads.

WIDE-AZIMUTH TOWED STREAMER

Figure 7 illustrates a single shot into a grid of receivers producing an offset/ azimuth distribution that would be close to ideal for WEM imaging. We can approximate this type of sampling with towed streamers by decoupling the seismic source from the streamer vessel and sailing the streamer vessel multiple times to occupy all required receiver positions relative to the shot location. The idea of using multi-vessel and multi-pass towed streamers for wide azimuth has been proposed before, in 2002.^2,3

Fig. 7. Single shot recorded into a 50-m x 50-m receiver. Receiver grid dimensions are a function of required inline and crossline offset requirements.

Consider an example where the desired receiver grid is 4-km wide by 8-km long, with the shot located half-way along one edge of the spread. This acquisition design can be represented by a streamer vessel towing 40, 8-km-long streamers with 100-m separations between adjacent streamers, Fig. 8. At each shot, offsets will be recorded over 180° of azimuth in the direction of the streamer spread, Fig. 9a. By using a second source vessel on the other side of the spread, or reoccupying shot locations while sailing in the opposite direction, or using source/ receiver reciprocity, full 360° azimuth distributions may be obtained, Fig. 9b.

Fig. 8. Forty streamers, each 8,000 m long and 100 m apart.

Fig. 9a. Offset/ azimuth distribution for single shot into 40-streamer spread.

Fig. 9b. Possible complete offset/ azimuth distribution from 40-streamer spread.

Of course, as mentioned before, there are no seismic vessels that are capable of towing such a streamer spread; however, we can use currently available vessel capabilities to arrive at the same end. Consider, for example, splitting the 40-streamer spread into four sub-sets of 10 streamers each. There are many seismic vessels in the world fleet capable of towing 10 long streamers. Using these assets, the desired 4-km by 8-km receiver grid can be achieved in a couple of ways. One approach is four 10-streamer vessels arranged side-by-side. With this method, the data from the full 40-streamer spread is recorded simultaneously at each shot location.

Another, probably more pragmatic, method is to make four separate passes at each shot location with a single 10-streamer vessel. The first pass would place the streamers nearest the shot location. On subsequent passes, the streamer vessel is offset from the source vessel by successive increments of the 10-streamer width. Each pass provides a “segment” of the final desired offset and azimuth distribution, Fig. 10. The data can then be sorted and combined to reflect a common shot into the full 40-streamer spread.

Fig. 10. Offset/ azimuth distributions for each successive pass of the 10-streamer subset.

OBS NODES

Onshore, wide azimuth is much more straightforward. There, sources and sensors can be as dense as an oil company thinks best. With the advent of autonomous seafloor nodes, true multi-azimuth acquisition can occur in a manner analogous to land. Are such ocean-bottom seismic (OBS) nodes a viable option? Or would thousands of nodes, an additional vessel, remotely-operated vehicles (ROVs), or placement cranes make the cost prohibitive?

Any ocean-bottom seismic survey, whether with cables (OBC) or with stations (OBS) is more expensive than surface-towed streamers. However, when it comes to imaging under complex overburden, multi-direction towed-streamer acquisition is at least twice as expensive as conventional narrow azimuth. Wide-azimuth towed-streamer surveys, involving more than one vessel and repeating many lines with variable shot or receiver positions, are even more expensive than multi-directional. So, are OBS surveys competitive with towed streamers for comparable wide-azimuth sampling? Our analysis, which is summarized in Fig. 11, shows that OBS surveys are more economical over small areas, while streamers are more economical over large areas.

Fig. 11. Cost comparison of Wide-Azimuth Towed Streamers to ocean bottom stations. OBS are more economical over small areas. WATS are more economical over large areas.

This makes OBS surveys a viable technology for target-oriented surveys. In addition to the economy, OBS nodes provide a great advantage in areas with obstacles, such as drilling and production facilities. Last, but not least, nodes record multicomponent data, which provide converted waves and great advantages in multiple elimination or imaging with multiples.⁴ The added ability of ROVs to redeploy OBS in exactly the same locations, makes it a powerful tool for 4D seismic monitoring. 

To summarize, in our opinion, OBS are the best technology for smaller wide-azimuth reservoir-scale seismic surveys, while towed streamers are the best technology for large-area wide-azimuth exploration surveys.

WHAT ABOUT AZIMUTHAL ANISOTROPY?

Azimuthal anisotropy is both a challenge and an opportunity for wide-azimuth surveys. Obviously, in the presence of azimuthal anisotropy, it is more difficult to image wide-azimuth surveys than it is to image narrow azimuth surveys, because he velocity model has more parameters. The better the velocity model is, the more focused and accurate the image is that we obtain from each common shot or receiver gather. Even with narrow-azimuth geometry, the azimuthal anisotropy is sometimes strong enough to cause noticeable effects.

An example from offshore West Africa, presented at last month’s EAGE, showed a portion of a conventional 3D survey that was affected by a significant acquisition footprint due to azimuthal anisotropy.⁵ Correction for such effects, using either a residual moveout correction before migration, or incorporating the anisotropy within the migration step itself, can significantly improve the imaging of the data. As we progress from narrow to wide azimuths, the data are more sensitive to azimuthal anisotropy, but the problem is not new. We do have to improve our processing methods. It takes longer to process wide-azimuth data, but we can do it.

The positive side of this issue is that wide-azimuth geometry provides us the opportunity to analyze azimuthal anisotropy for the purpose of fracture characterization, well planning and for improving the quality of amplitude versus offset analysis (AVO). AVO analyses provide excellent lithology characterization. However, one pitfall is that azimuthal anisotropy may have a large effect on the AVO. With narrow-azimuth data, we cannot differentiate between a change in the direction, the amount of anisotropy, and a change in the lithology, such as sand to shale variations. Wide-azimuth data are more challenging to image, but the reward is not only improved imaging under complex overburden, but also improved AVO analysis under any overburden.

SUMMARY

Most, if not all, of the easy oil and gas may have been found, and a lot of it has been produced. Much of what is left is under complex overburden, such as the salt in the deepwater Gulf of Mexico. Processing methods started to adapt a decade ago with extensive use of Pre-Stack Depth Migration, Surface-Related Multiple Elimination, and more recently, Wave Equation Migration. Commercial acquisition methods, however, have not progressed until very recently.

We are now starting to acquire the wide-azimuth data that is needed to find and produce oil and gas under salt. Marine wide-azimuth data can be acquired in two ways, one is with towed streamers and the second is with ocean bottom seismic nodes. OBS are the best technology for smaller, wide-azimuth reservoir surveys, while towed streamers offer the best technology for large-area wide-azimuth exploration surveys. 

LITERATURE CITED

  ¹  Ver West, B., R. Hobbs and J. Young, “Multi-directional 3D acquisition and processing subsalt imaging,” presented at the EAGE in Amsterdam. Expanded abstract, 2001.
  ²  Paffenholtz, J., J. Keller, R. Ergas and B. McLain, “Surface marine all azimuth recording technique,” presented at the EAGE in Florence. Expanded Abstract, 2002.
  ³  D.V. Sukup, “Wide-azimuth marine acquisition by the helix method,” The Leading Edge, August 2002.
  ⁴  Ronen, S., L. Comeaux and X. Miao, “Imaging down-going waves from ocean-bottom stations,” presented at the SEG in Houston. Expanded Abstracts, pp. 963-966, 2005.
  ⁵  Wombell, R., “Characteristics of Azimuthal anisotropy in narrow-azimuth marine streamer data,” presented at the EAGE in Vienna. Expanded abstract, 2006.

THE AUTHORS
	Shuki Ronen is directing multi-component research in Veritas. He has contributed to multi-component seismic surveying mainly in the areas of vector fidelity, azimuthal anisotropy analysis, and imaging, including pioneering the use of seismic attributes for estimating reservoir properties. He earned a PhD from Stanford University in 1985. He then spent a year in Saxpy Computer engineering algorithms for seismic data processing on parallel computers, then a year at the Colorado School of Mines as a visiting professor. His industry experience includes various positions in Schlumberger and its subsidiaries before joining Veritas in 2001. In 2002 he received a special commendation award from the SEG.
	Phil Fontana manages the geoscience support group for Veritas’ Global Marine product line. He has over 25 years’ experience in marine seismic acquisition technology development, survey design, and technical QA/ QC of towed streamer and ocean bottom surveys. He holds a BS degree in geology and an MS degree in geophysics from the University of Connecticut.