Almost everyone has heard, perhaps more often than they would like, their mother’s admonishment, “You can’t have your cake and eat it too.” Parents used this pronouncement liberally, applying it to almost every situation where a choice was involved. As children agonized over what was, in their small world, a crisis, the adults seemed to think that this sage advice would speed up the decision process. In fact, it did the opposite.

Finally, as we grew older, our minds came to grasp the idea that perhaps our parents were correct in this regard — there was simply no method by which we could have something both ways. Such would seem to be the case with deepwater marine seismic.

This is an example of the sparse over/under acquisition geometry used in the validation test. NOTE: Geometry may vary for different acquisition designs. (Images courtesy of WesternGeco)

Either an operator could design an acquisition project to image the shallow sediments overlying the salt or focus the acquisition on the deep subsalt sediments surrounding the target reservoir. One seismic design favored the high frequencies that provide sharp images of shallow sediments while the other favored the low frequencies that image very deep sediments. But operators could not have them both because of interference from the so-called free-surface ghost effect. Caused by seismic energy reflections off the sea surface interface, the ghost effect creates interference that affects both low- and high-frequency signals, depending upon the tow depth of the receivers.

Current exploration and development activity in many deepwater plays creates a dual dilemma. Asset teams want to minimize uncertainties of the overburden so they can penetrate the salt efficiently and safely, but they also want to image the deep-seated sub-basalt reservoir clearly and accurately so wells can be steered to intersect the most prolific reservoir volumes.

Vertical incidence ghost responses are combined to illustrate how simultaneous low-frequency data provide an average 10 dB of signal uplift (red-shaded area), resulting in a greatly improved stacked image below 20 Hz from the deep array.

Either objective can be targeted by adjusting the depth at which the seismic streamers are towed: towing a shallow spread favors high-frequency signals, while towing a deep spread favors the deep-reading low-frequency signals. But geoscientists and drilling engineers wanted to have their cake and eat it too — they wanted both signals processed in a single dataset so rock properties could be inverted to properly model and extrapolate well log data over long distances.

A potential solution emerges

Attempting to solve the problem by making dual passes, firstly with a shallow spread and subsequently with a deep spread, was cost-prohibitive and posed data integration challenges. However, some years ago, a dual-streamer, over/under technique was introduced that involved towing simultaneously two vertically parallel streamers separated by a specific distance which was determined during the acquisition design. The dual-streamer technique was relatively costly and not particularly efficient; there did not seem to be an economic solution that could be optimized to achieve both objectives.

Partial solution leads to discovery

More seismic, less model — The structure seen in the 2 Hz data is difficult to include in a low-frequency model from velocities or well log data.

The over/under concept was viable and effective, but it had drawbacks. Consequently, more effort was invested to improve the acquisition process with the goal of achieving the ability to efficiently deliver 3-D seismic with enhanced frequency bandwidth, providing both high resolution and deep penetration. The result was the development and successful testing of the DISCover Deep Interpolated Streamer Coverage technique.

The technique is enabled by a shallow-tow Q-Marine point-receiver marine seismic system spread augmented by a smaller number of deep-tow cables. It was discovered that there was no need to run an equal number of deep cables as is done in traditional over/under acquisition; only a small number were required. The resulting sparse deep cables serve to optimize the system’s signal-to-noise (S/N) ratio over the entire bandwidth when their data is combined with that of the shallow spread. The new technique decouples the high- and low-frequency design criteria, allowing both frequency bands to be recorded and combined in a way that optimizes S/N through the entire frequency band.

Single-sensor shot records from the shallow and deep tow depths are compared following Digital Group Forming noise attenuation to illustrate reduced noise and stronger reflection signals from the deeper sensor.

The key lies in the fact that since the deep-tow streamers are only used to record the low frequencies that image the deep strata, only a few are required, providing improved cost efficiency. To illustrate how it works, images from a validation test conducted offshore Northwest Australia are presented. It should be noted that the spread layout is specific to that particular job, and the depths and streamer spacing intervals are part of that design (Figure 1). The DISCover technique works for any design, which may dictate other streamer tow depths and intervals. When the data from the lower streamers are combined with that of the shallow tow spread, a large area of valuable low-frequency data is conserved that would have been lost if the shallow spread acquisition had been run alone (Figure 2). Data below 20 Hz contributed an average of 10 dB uplift to the signal strength. When the stacked images are compared, the deep streamer data exhibits strikingly better quality. When the shot records for both shallow-tow and deep-tow streamers are compared, the deep shot data is clearly improved down to 6.0 sec. At the same time, the improved S/N from the suppression of shallow ghosting results in better quality shallow sediment imaging as well.

Figure 3 shows an example of the extremely low frequencies which can be obtained using the DISCover technique. The data have been filtered with a 2 Hz high-cut filter. Coherent energy following the gross structure is still visible. There is significantly higher amplitude and significantly more coherence on the DISCover data compared to the shallow-only data.

Knowledgeable processing is key

The interpolated result is achieved during data processing. The actual workflow varies depending upon the spread design, but for the Australia example we first treat the dataset as a sparse set of over/under pairs. Then the workflow proceeds as follows:

• Combine the over/under pairs up to a given frequency, typically around 20 Hz, where most of the deep streamer uplift is obtained.

• Re-datum the low-frequency enhanced and deghosted data to the shallow depth datum.

• Interpolate and regularize the low-frequency data to the positions defined by the shallow spread.

• Merge the low-frequency data with the high-frequency data from the shallow spread.

• From this point, follow any standard 3-D processing procedure.

Alternatively, if the designed spread geometry does not place a deep streamer directly below a shallow one, the third step may be performed first. Either way, spread geometry control is essential so that interpolation distances do not become overextended and trace binning does not exceed acquisition specifications. Appropriate streamer position control can be achieved using steerable-streamer technology. This technology can control both lateral and vertical positioning.

The final 3-D images from the DISCover technique (right) are compared with those acquired conventionally. Processing was the same for both examples, yet resolution is clearly superior on the right image despite poor weather conditions during acquisition.

In the Australia example, the dataset was acquired immediately following a cyclone, and both weather and sea state were sub-optimal. Experienced geoscientists and operations personnel were of the opinion that environmental conditions were beyond the capabilities of a conventional seismic array because of severe noise interference. Nevertheless, when compared with conventional tow data from the area, the DISCover data exhibited increased shallow resolution, better shallow fault imaging, and a more elegant structural picture. The DISCover data had a higher S/N with better continuity and interpretability. Downtime was minimal despite the poor weather.

An important processing point is that although the data from the sparse deep streamers are generally the source of the low-frequency deep-reading seismic images, the shallow-tow streamer acquires a certain amount of low-frequency data as well. DISCover data benefits from further enhanced S/N improvement as the data from both sets of streamers are combined and merged. The process optimizes the low-frequency response of both streamers. By estimating the noise level on each streamer array, the data from each streamer can be weighted to optimize the output S/N using a least squares procedure. The resulting improvement is significant (Figure 4).

Results achieved

The data were processed using a conventional 3-D processing sequence that included a prestack Kirchhoff time migration. The shallow-only data were also processed using the same technique to produce a comparison. The low-frequency (deep) image is considerably clearer on the combined data image, providing much more valuable insight at reservoir depths. While at first glance the shallow image appears well-defined in the shallow-only example, in fact, details like the shallow faulting are more clearly defined on the combined image. This provides drillers with more useful information as they drill through the overburden layers.

The broadband result achieved with the sparse over/under acquisition technique provides results believed to be unobtainable using a simultaneous acquisition scheme. Through the use of this procedure, valuable information can be acquired economically from both shallow and deep sediments at once, with spectacular results (Figure 5).