Exorcising the ghosts
D.I. Hill, J. Bacon, T. Brice, L. Combee, C. Koeninger, M. Leathard, S. McHugo - WesternGeco
An over/under, towed-streamer configuration is a method of acquiring seismic data where cables are towed in pairs at two different depths, with one cable positioned above the other. The depths of these paired cables typically are significantly deeper than for a conventional towed streamer. It is possible to acquire data with paired sources at two differing depths in conjunction with the paired cables. The paired sources are also deeper than those used in a conventional towed-streamer configuration.
Better data
The seismic data recorded by the over/under towed-streamer configuration are combined in data processing into a single dataset that has the high-frequency characteristics of conventional data recorded at a shallow towing depth as well as the low-frequency characteristics of conventional data recorded at a deeper towing depth. This combination process is commonly referred to as “de-ghosting.”
Recent commercial applications of the over/under seismic data acquisition technique have been made possible by the development of the steerable cable. The removal of the source and cable ghosts via over/under de-ghosting extends the usable bandwidth of the data at both the low and high ends of the spectrum. Field applications demonstrate that the increased low-frequency content gives deeper penetration to improve imaging beneath highly absorptive overburdens such as basalt.
In a conventional towed-streamer marine acquisition configuration, shallow sources and shallow cables increase the high-frequency content of the seismic data needed for resolution, but attenuate the low frequencies needed for deep structural imaging and seismic inversion.
Shallow towing also makes data more susceptible to environmental noise. Deep sources and deep cables enhance the low frequencies and attenuate the high frequencies. The resulting recorded data also have a higher signal-to-ambient-noise ratio because of the more benign towing environment.
A conventional towed-streamer survey attempts to balance these conflicting aspects to arrive at a tow depth for the sources and cables that optimizes the bandwidth and signal-to-noise ratio of the data for a specific target depth or two-way travel time, often at the expense of other shallower or deeper objectives.
Over/under de-ghosted data is better than conventional data because it yields the following:
- A broader signal bandwidth, where the low-frequency content gives deeper penetration, resulting in improved imaging beneath highly absorptive overburdens such as basalt or salt
- A bandwidth extension to lower frequencies making seismic inversion less dependent upon model-based methods
- A simpler signal wavelet in conjunction with the bandwidth extension to higher frequencies that enhances resolving power to allow a more detailed stratigraphic interpretation
- A simpler signal wavelet in conjunction with the bandwidth extension to lower frequencies that gives better steep-dip imaging
- Higher signal-to-ambient-noise ratio as a consequence of the deeper towed-cable pairs.
Over/under acquisition
A look at over/under acquisition geometry shows paired sources at two different depths and paired cables at two different depths. The wavefield radiates from the seismic source and travels not only downward, but upward. The upward-propagating wavefield is reflected from the sea surface with an opposite polarity and interferes destructively at some frequency with the downward-propagating wavefield. This interference results in a source-ghost notch in the amplitude spectra of the downward propagating wavefield. An equivalent mechanism at the receiver results in the addition of a receiver-ghost notch. These ghost notches are key factors in restricting the bandwidth of recorded seismic data. The basic principle of over/under de-ghosting is to combine recorded datasets to produce a dataset with a flat, ghost-free spectrum. Various methods have been proposed to achieve de-ghosting.
The amplitude spectra of vertically propagating ghost responses for depths of 11 m (36 ft) and 27.5 m (90 ft) shows that ghost notches occur at frequencies that are integer multiples of 68.2 Hz and 27.2 Hz, respectively. The amplitude spectrum of the de-ghosted data using the Posthumus (1993) technique is a weighted value of these two spectra. Data recorded at a depth of 11 m (36 ft) would be used to infill the notches in the spectrum of the data recorded at 27.5 m (90 ft). Data recorded at a depth of 27.5 m (90 ft) would be used to infill the notches in the spectrum of the data recorded at 11 m (36 ft).
A typical 2D over/under acquisition geometry comprises four cables: the over/under cable pair and one cable towed on each side of the over/under pair to provide the acoustic positioning network necessary to locate and control the over/under cables. De-ghosting requires that the wavefield recorded by the over cable be a time-delayed version of the vector wavefield recorded by the under cable.
Operationally, this translates into keeping the over and under cables at constant depths with a constant vertical separation and no lateral separation. Consequently, the main requirement of an over/under cable acquisition system is that the cables be aligned vertically above one another and that their alignment be maintained in the vertical plane throughout the survey.
Achieving this can be difficult when current strength and direction vary as a function of depth below the sea surface. Recent successful applications of the over/under technique have only been made possible by the development of the steerable cable. The control systems associated with this cable are capable of keeping the over/under cables in within the very small tolerance required for the technique to work correctly.
North Atlantic trial
In the fall of 2005, the seismic survey vessel,Western Pride, completed a regional 2D over/under survey for Chevron in the North Atlantic west of the Shetland Islands. The survey objective was to improve seismic images of the subsurface beneath stacked basalt flows, which are a characteristic of this region. Stacked basalt flows typically act as a low-pass high-cut filter on seismic waves. Successful sub-basalt imaging relies on technologies that extend the useable seismic bandwidth toward very low frequencies.
Key de-ghosting steps in an over/under data processing comprise:
- Over source, over/under cable combination
- . Under source, over/under cable combination
- Over/under combination of over and under source datasets (The data from the previous two steps are combined)
When the over/under combinations are complete, both the source and receiver ghosts have been removed. The data processing sequence then follows a more conventional route, but in such a way as to maximize the benefits of the enhanced bandwidth delivered in the over/under combination steps.
Datasets were processed from the Kirchhoff prestack time-migrated image from the over source at a depth of 12 m (39 ft) and the over cable at a depth of 20 m (66 ft), equivalent to a conventional deep-towed-streamer acquisition configuration and for the corresponding Kirchhoff prestack time migrated image from the fully de-ghosted data from the over/under source and cable combination.
Both datasets were processed using identical parameters; they were migrated with the same velocity field, stacked with the same velocity field, and have the same post-migration processing.
A comparison of the results shows that the combined over/under dataset is far richer in low frequencies than the deep-towed-streamer acquisition configuration. The quality of the image beneath the basalt (at 4.0 second two-way travel-time) on the data from the over/under combination is far superior to that from the conventional deep-towed-streamer acquisition configuration.
The enhanced low-frequency content also makes seismic inversion less dependent upon model-based methods.
The image shows a small portion of an inversion to relative acoustic impedance of both the over source over cable data, and the fully combined and de-ghosted data. Two wedge models are also shown. The geologic model common to both is a sandstone wedge encased in shale.
Synthetic seismic data generated from Hz to Nyquist were used for the seismic modeling, and for the wedge on the right, a bandwidth from 5 Hz to Nyquist. Both seismic models then were inverted to relative acoustic impedance. The relative acoustic impedance results appear as the wedge models. The benefits of the additional low frequencies can be seen clearly when the wedge model relative acoustic impedance results are compared. The relative acoustic impedance wedge on the right identifies the acoustic impedance contrasts at the top and bottom of the wedge, but fails to reflect accurately the thickness of the sandstone wedge. The relative acoustic impedance wedge on the left not only identifies the acoustic impedance contrasts at the top and bottom of the wedge, but also depicts the sandstone wedge as a single, solid geobody.
A striking similarity emerges when the inversion to relative acoustic impedance of the real data is compared to the wedge models. The inversion to relative acoustic impedance of the de-ghosted data can be interpreted to have a wedge-shaped feature. The inversion to relative acoustic impedance of the over source and over cable data cannot.
A relative acoustic impedance inversion of over/under data results in a more reliable estimate of geobody volume than with conventional data. This could impact exploration prospect economics and field development planning.
Test offshore India
The seismic vesselWestern Pride carried out an over-under experiment off the west coast of India for ONGC in May 2006. The area was one with multiple stacked basalt flows of varying thicknesses. The primary test objective was to enhance imaging of Mesozoic marine clastics beneath the basalt. The test applied the same over/under de-ghosting data processing sequence used in the previous case history.
As in the first case, the quality of the image beneath the basalt on the de-ghosted data is superior to anything produced from data gathered by a conventional deep-towed-streamer.
The directional de-ghosting of data recorded using both over/under sources and over/under cables removes the ghosts notched associated with source and receiver depths. This removes the band-pass filtering effects at the low and high frequencies caused by destructive interference. The bandwidth of over/under data now becomes governed by the bandwidth of the source itself, absorption at the high-frequency end of the spectrum, and the instrument filters at the low-frequency end of the spectrum. The case histories demonstrate that the technology delivers benefits and improvements to the quality of seismic data.
Acknowledgements
The authors thank Kevin Davies and Gary Hampson of Chevron as well as Chevron Corp. for permission to publish the data examples in the first case history. Thank you also to ONGC for permission to publish the data examples in the second case history.
References available.