Uses of wide-azimuth and variable-depth streamers for sub-basalt seismic imaging

Transcription

1 Uses of wide-azimuth and variable-depth streamers for sub-basalt seismic imaging To evaluate the optimal technique for imaging beneath a complex basalt layer, Robert Dowle, 1* Fabrice Mandroux, 1 Robert Soubaras, 1 and Gareth Williams 2 compare dense wide-azimuth and broadband towed streamer acquisition solutions when deployed in an experimental survey in the Faroe-Shetland Basin. A n evaluation of dense wide-azimuth and variable-depth streamer acquisition techniques for sub-basalt imaging was carried out over the Faroe- Shetland Basin, northwest of Scotland in August The aim of this experimental survey was to enhance the low-frequency signal and to improve the attenuation of multiples and scattered energy, thereby obtaining better imaging beneath the complex basalt layer. In this area extensive sequences of basalt dominate the northwestern flank of the UK continental shelf. The vertical and lateral inhomogeneity of these basalt flows results in loss of bandwidth as well as loss of signal. All but the lowest-frequency seismic energy penetrating the basalt becomes incoherent either by attenuation or scattering. Scattering in particular poses a difficult problem due to the rugose top and base basalt and internal heterogeneity. Various techniques have been employed in an attempt to overcome these problems, the most promising of which are dense multi- or wide-azimuth and broadband acquisition. The better areal sampling of wide-azimuth data is known to improve signal and attenuation of multiples, as the azimuthal variation and high fold allow true 3D processing algorithms to be used, and is therefore expected to be beneficial sub-basalt. The low frequencies recorded by broadband acquisition will be affected less by scattering and attenuation and provide a better image below the basalt. A dense, wide-azimuth experimental seismic survey performed in the Altantic Margin of Northwest Europe to improve attenuation of multiples and scattered energy and hence obtain better imaging beneath the complex basalt layer (Williams et al., 2011). In addition, the same line was recorded using a broadband variable-depth streamer technique, designed to enhance the recording of low and high frequencies. A conventional 2D line was also acquired along the same track so as to create a direct comparison between conventional, wide-azimuth, and broadband variable-depth streamer acquisition. Geological challenge Basalt is present in numerous sedimentary basins, including some known prospective areas; therefore exploration for hydrocarbons in both sub-basalt and intra-basalt environments has been attempted in a wide variety of locations around the world, such as the Atlantic Margin and offshore India. Although there has been some success, progress has generally been limited by the poor quality of seismic images in many of these situations. It has been difficult to obtain even structural images of sufficient quality and reliability to identify drillable prospects and amplitude analysis is usually even more problematic. The major difficulties are caused by strong multiples and strong absorption within the basalt. The impedance of basalt usually gives rise to a strong reflection coefficient at top basalt and hence strong multiples. Seabed multiples and interbed multiples between top basalt and the seabed may compound the complexity of the multiple energy. The basalt layer may contain multiple flows with erosion and sedimentation occurring between the flows. This can lead to strong absorption within the basalt and also significant scattered energy with short spatial wavelengths. The combination of strong, complex multiples, scattered energy, and absorption of the primary gives rise to poor signal-to-noise ratios beneath and within the basalt. It is commonly accepted that only very low frequencies, perhaps below 20 Hz, can be used for sub-basalt seismic imaging and, even then, multiples and scattered energy are problematic. Wide-Azimuth Solution Seismic imaging beneath structurally complex salt bodies in the Gulf of Mexico has been enhanced by the use of wide-azimuth recording geometries. This has not only allowed enhanced illumination of sub-salt layers, particularly around the edges of discrete salt bodies, but also provided an uplift in multiple suppression. However, scattered energy is not usually generated within salt 1 CGGVeritas. 2 Formerly CGGVeritas. * Corresponding author, robert.dowle@cggveritas.com 2011 EAGE 89

2 first break volume 29, December 2011 bodies to the same degree that we find within multiple basalt flows. Multi-azimuth techniques have been used offshore Egypt to improve sub-messinian imaging and to improve signal-to-noise ratios below a rugged surface. Furthermore, Keggin et al. (2002) showed that a dense, wide-azimuth geometry significantly attenuated multiple diffracted energy from a rugose seabed offshore Norway: a dense cross-line sampling was necessary because of the short wavelength nature of the point diffractors causing the multiple energy. This may be analogous to the scattered energy causing problems for sub-basalt signal-to-noise. A combination of dense sampling and azimuth diversity should provide a means of suppressing both the multiples of dipping events and the short wavelength scattered energy typical of basalt areas. Broadband solution Three octaves of signal are generally considered necessary for adequate seismic resolution; therefore, in order to achieve sufficient bandwidth in areas where high frequencies are limited, it is necessary to extend the low frequencies. Lower frequencies are affected less by attenuation, and provide greater accuracy and stability for seismic inversion. Conventional marine streamer acquisition lacks sufficient signal-to-noise ratio in the 2 7 Hz bandwidth due to streamer depth, streamer tow noise, source array configuration, source depth, and source bubble. BroadSeis is a variable-depth streamer broadband solution that uses variation of the receiver depths to produce receiver ghost notch diversity, allowing the streamer to be towed deeper to improve the low-frequency signal-to-noise ratio Figure 1 Map showing the location of the survey. The output line straddles the UK-Faroes border in the W. Shetland region of the N. Atlantic margin. At the UK end in the SE, there is no basalt present but there is an increasing thickness of basalt towards the NW EAGE

3 Figure 2 Basic shooting geometry illustrating the particular case of a 1km cross-line offset. We have a single, eight-streamer recording vessel and a single-source vessel with the target CMP line between them. The target streamer is marked by the triangle and is 500 m from the CMP line in the case of no feathering. The source is also 500 m from the CMP line and hence the total cross-line offset is 1 km. The cable length is 8 km and the source has been positioned mid-way along the cable so that in-line we have a split spread +/- 4 km offset geometry. Figure 3 The predominant currents in the area are from the SW so the two boats were positioned such that the source vessel was always on the NE side of the streamers. This allowed her to speed up and move quickly away if feathering suddenly increased even though in theory the cables should pass under the source as they are 5 m deeper. without compromising the high frequencies. This ghost notch diversity is exploited by proprietary deghosting and imaging techniques (Soubaras 2010), to produce a wavelet with both a high signal-to-noise ratio and maximum bandwidth. With this technique, signals are being routinely recovered down to 2.5 Hz, providing three octaves of data below 20 Hz. This solution capitalizes on the extremely low-noise characteristics and precise low-frequency response of solid streamers (Dowle, 2006). Solid streamers are quieter and can be towed deeper (up to 50 m) than other streamers, so that better low-frequency signal-to-noise ratios are achieved. The new generation of solid streamers has lower instrument low-cut filters, which allow recording down to 2 Hz, providing a full extra octave of data at low frequencies over some other systems. According to conventional wisdom, seismic sources do not produce sufficient low frequency to warrant recording below 5 Hz, but even using a conventional source we routinely observe coherent signal as low as 2.5 Hz. Survey design The wide-azimuth survey was designed using a 2D wideazimuth geometry following the method described by Keggin et al. (2002). Unlike the Gulf of Mexico wide-azimuth surveys, the primary concern was the ability to suppress both multiples and scattered energy; improved illumination was a secondary concern since the basalt flows are extensive and do not allow illumination around the edges of discrete bodies. Thus, the aim of the test geometry was not to provide sufficient data to achieve a 3D migration, but to test the ability of a wide-azimuth stack and other pre-stack de-multiple techniques to suppress the multiples and scattered energy. A single CMP line crossing the UK-Faroes border of the Atlantic Margin was chosen from existing 3D data. The line was chosen to intersect a Faroes Island well and to run from an area with no basalt in the southeast into an area towards the northwest with increasing thickness of basalt (see Figure 1), to test if horizons identified away from the basalt could be tracked underneath it and, if so, whether they could be tracked further and with more reliability than with conventional narrow-azimuth data. The data was acquired by a single recording vessel with eight streamers separated by 100 m and a separate source vessel firing a single source every 25 m, as shown in Figure 2. The streamer vessel always sailed along one side of the target CMP line and the source vessel on the opposite side, thus providing off-end shooting in the cross-line direction. The purpose of the multi-streamer configuration was to compensate for feathering rather than to record more than one CMP line; 3D binning allows us to use data from any streamer that has the required cross-line offset to hit the CMP line. The cable length used was 8 km. However, since the top basalt is reasonably shallow at approximately 2 s two-way time, the source vessel was moved back to be positioned at the mid-point of the cable. Given the 8 km streamers, this provided an in-line split spread configuration of +/- 4 km offset. Offsets greater than this would typically be muted to a reasonable depth below the basalt in this area due to normal moveout stretch. The cross-line offset spacing was 100 m which is much denser than that typically used for Gulf of Mexico wideazimuth recording. This was chosen since many previous studies in this area have shown that inline offset sampling 2011 EAGE 91

4 first break volume 29, December 2011 Figure 4 Offset distribution of the individual traces within a typical CMP. The different colours refer to different passes of the vessels. The effect of moving the source vessel closer to the streamer head at the shorter crossline offsets is clearly visible (towards the bottom of the figure) with the in-line coverage being much closer to off-end than split spread. of 100 m within CMP gathers is desirable for effective multiple suppression; wider spacing leads to aliasing that has not adequately been recovered with interpolation. In addition, the scattered energy is thought to have short spatial wavelengths and therefore needs dense sampling. This design provides near symmetrical sampling of in-line and cross-line offsets within the CMPs. Decimation tests during processing can be used to quantify the impact of the cross-line sampling on the quality of the final image and assess which parameters represent the optimum trade-off between acquisition cost and data quality. A total of 11 passes of the two-boat configuration were made with the shortest cross-line offset being 100 m. An Figure 5 Narrow- and wide-azimuth brute stack comparisons with no demultiple. The suppression of the water bottom multiple by the wide-azimuth stack is striking. additional central 2D line was acquired by the streamer vessel sailing along the CMP line using her own source. Thus, cross-line offsets from 0 m to 1100 m were recorded with 100 m increment. Other survey parameters such as source depth (10 m), cable depth (15 m) and 4260 in 3 source were typical of previous surveys in the area. Figure 6 Typical wide-azimuth NMO-corrected gathers showing flat primaries and strong multiples with residual moveout. The gathers have been sorted by increasing azimuth within increasing offset ranges and so show cyclical jitter on the multiple moveout which increases with offset. This means that when we stack the data the multiples are attenuated more effectively EAGE

5 Figure 7 Narrow- and wide-azimuth comparisons after 2D prestack time migration. The NAZ result has been achieved by processing the full WAZ dataset and then simply limiting the crossline offsets. Therefore, it still artificially benefits from the 3D radon demultiple which would not really be possible on NAZ data. Both sections would benefit from SRME, but insufficient data has been acquired for the application of 3D SRME and applying 2D SRME to the WAZ data would be incorrect. Figure 8 A zoom from the centre of the line, showing the deeper section. Not only do we see an improved S/N ratio, but the fault structure is much easier to identify. There is also greater continuity on some of the events EAGE 93

6 first break volume 29, December 2011 Figure 9 A zoom around the basalt, showing greater clarity for interpretation at and below the base basalt. The central 2D line was also acquired using a variabledepth streamer with receiver depths varying from 7.5 to 50 m. The combination of Sercel Sentinel solid streamers and Nautilus controllers is ideal for variable-depth acquisition, since the control of solid streamers is very robust and sta ble, even at depths of 50 m. The streamers retain their low-noise characteristics at all depths. Data acquisition In this area, it is known that currents can vary rapidly and are predominantly from the southwest towards the northeast. Therefore, the source vessel was always positioned on the northeast side of the line so that she could move away quickly if the cable feathering increased quickly towards her (see Figure 3). The first pass of the vessels targeted a crossline offset of 500 m to allow the vessel crews to gain experience of steering with the designed configuration including a safe separation between the source and the mid-points of the cables. The crossline offset was then reduced on successive passes by 100 m. During the 300 m cross-line offset pass, a rapid change in feathering meant that the source vessel had to steer away. Therefore, for the 200 m and 100 m passes, the source vessel was moved closer to the head of the streamer in order to reduce the risk from large feathering. For these passes, some split-spread in-line coverage was achieved but it was not symmetrical. The longer cross-line offset passes Figure 10 Amplitude spectrum comparison for the conventional and BroadSeis broadband data EAGE

7 Figure 11 Conventional and BroadSeis comparison, showing considerable improvement in signalto-noise ratio at depth on the broadband data, allowing deep sub-basalt stratigraphy and volcanic intrusions to be interpreted with more confidence. were recorded subsequently and with fully symmetrical split spread positioning. The offset distribution of individual traces in a typical CMP is shown in Figure 4; the reduced split spread coverage of the in-line offset distribution can be clearly seen at the shorter cross-line offsets. For the longer cross-line offsets, the receiver vessel was positioned so that one of her central streamers would provide the designated cross-line offset to the CMP line where there was no feathering (see Figure 2). Should feathering occur in either direction, one of her other cables would then be correctly positioned to provide the necessary coverage for binning. However, at shorter offsets the outer cables had to be targeted. For example, at the shortest cross-line offset of 100 m, the source was positioned 50 m away from the CMP line and the outer cable steered to be 50 m the other side. The multiplicity of cables then only allowed for feathering in one direction (towards the source vessel) to be binned successfully. Analysis of wide-azimuth results The comparison between the 2D narrow-azimuth and the wide-azimuth brute stacks (Figure 5) shows a striking suppression of the water-bottom multiple using the wideazimuth acquisition technique. This can be understood by considering the multiple of a dipping event. During processing, we typically choose velocities and imaging tools to flatten the primary of that event within a CMP gather. However, the multiple will usually have residual moveout and this moveout will depend, amongst other things, on the dip of the event. Therefore, the residual moveout also depends on the shot-receiver azimuth, i.e., 2011 EAGE 95

8 first break volume 29, December 2011 Figure 12 Zooming in around the top and base basalt shows the clearer definition of this layer. The Kettla Tuff is also clearly identifiable on the BroadSeis section. Figure 13 Zooming in still further shows clearer definition of intra and base basalt including probable multiple basalt flows pinching out. whether the trace is shot in the dip direction, the strike direction, or another direction. Thus, azimuthal diversity helps to suppress multiples during the stack process by introducing an azimuth-dependent residual moveout. This effect can be seen clearly in the gathers in Figure 6. After full processing through 3D radon demultiple and prestack time migration (Figure 7) there is a less marked improvement. The narrow-azimuth stack was created by selecting a restricted crossline offset range (0 200 m) to stack, after the data had been processed through 3D radon demultiple, a process that gives superior multiple attenuation results, but which cannot be applied to real narrow-azimuth data. This means that the narrow-azimuth demultiple result is better than would normally be the case. In addition, insufficient data were acquired in this experiment to apply true 3D SRME to the data, a process that would normally be expected to provide better multiple attenuation and to benefit the wide-azimuth data more than the narrow-azimuth. In spite of this artificial reduction in the improvement to be gained, the wide-azimuth data clearly shows better multiple and noise attenuation than the narrow-azimuth data, due to the powerful effect of azimuthal diversity. A zoom over the deeper data (Figure 8) not only shows the improved signal-to-noise ratio, but also easier identification of the fault and greater continuity of the events. A zoom around the basalt (Figure 9) shows greater clarity for interpretation at and below this layer. There are other possible causes of the improvement in the stack. Firstly, the wide-azimuth stack is much higher fold than the 2D stack. Secondly, a wide-azimuth geometry on a regular grid in x and y introduces offset weighting, EAGE

9 Figure 14 Initial test results for a prototype broadband source and source deghosting solution to complement the variable-depth streamer acquisition and receiver deghosting solution. The removal of the source ghost provides increased clarity and bandwidth. since there are preferentially more traces with a mid offset. The first effect was evaluated by producing a stack containing a full range of azimuths but a fold equal to that of the 2D. The second simulation involved applying an offset weighting to the 2D stack to match that of the wide-azimuth stack. These simulations were performed on data that had been through the complete de-multiple sequence. In both cases, the benefits of azimuthal diversity could not be matched, indicating that this is the dominant effect. Analysis of broadband results The bandwidth recovered using variable-depth streamer acquisition and a proprietary deghosting technique was Hz, compared to Hz (see Figure 10) achieved with the conventional deep-tow acquisition typical for this area. A comparison of the conventional and variable-depth streamer broadband data is shown in Figure 11. The definition of the fault in the centre of the section is much clearer on the broadband section than on either the conventional or the wide-azimuth data. There is considerable improvement in signal-to-noise ratio at depth on the broadband data, allowing deep sub-basalt stratigraphy and volcanic intrusions to be interpreted with more confidence. The probable top and base of the Paleocene flood basalt is indicated by the blue lines. Below this major formation, a continuous and distinctive low-frequency event corresponding to the Kettla Tuff can be identified all the way through the broadband seismic section (pink dotted line). This reflector is hardly visible on the conventional data, especially in the western (left) part of the seismic section where the overlying flood basalt thickens, emphasizing the benefit of low frequencies for sub-basalt imaging. Zooming in around the top and base basalt shows the clearer definition of this layer (Figure 12) on the broadband section, where the Kettla Tuff is also clearly identifiable. Zooming in further around top and base basalt (Figure 13) shows probable multiple basalt flows pinching out EAGE 97

10 first break volume 29, December 2011 Broadband Evolution BroadSeis broadband marine streamer acquisition has matured over the 18 months since its launch and is continuously evolving with current development focused on two key areas: dedicated processing algorithms and source deghosting. Whilst the key processing step for BroadSeis is the proprietary deghosting approach (fully 3D and true-amplitude), it also benefits from the adaptation of conventional processing algorithms in order to make full use of the extreme bandwidth now available, for example, surface related multiple elimination and shallow water demultiple (Sablon, 2011 and Lin, 2011). Many algorithms are also being developed, capitalizing on the additional information provided, leading to continuous improvement throughout the processing sequence for variable-depth streamer data and resulting in even better broadband seismic images and attributes. Having developed a solution to the receiver ghost problem, we are now tackling the source ghost. A broadband source and source-deghosting algorithm, which complements BroadSeis variable-depth streamer acquisition and receiver-deghosting, has been tested on a different line in the West Shetland area (see Figure 14). This enables us to remove the source ghost as well as the receiver ghost, thus eliminating constraints on the highest frequencies that we can record. The results from this initial test using a prototype source show outstanding bandwidth and clarity (see Figures 15 and 16). The upper limit of 200 Hz was due to the anti-alias filters that were applied for recording at 2 ms sample interval and is not the limit of the source or receivers. At the low-frequency end of the spectrum we expect only an incremental uplift from the removal of the source ghost as we already have a steep gradient in the amplitude spectrum from 0 to 2.5 Hz after the removal of the receiver ghost (see Figure 15). However, the removal of the source ghost from the data provides a clearer wavelet for interpretation and in sub-basalt areas any improvement in low frequencies is to be welcomed. Figure 15 Amplitude spectra from the initial test results for a prototype broadband source and source deghosting combination to complement the variable-depth streamer acquisition and receiver deghosting solution. The upper limit of 200 Hz recorded here was due to the anti-alias filter applied for 2 msec sampling. Figure 16 Zooming in around the basalt shows the exceptional clarity and sharp images obtained from a fully deghosted wavelet EAGE

11 Conclusions Both wide-azimuth and broadband seismic were individually shown to improve sub-basalt imaging. Dense wide-azimuth seismic was shown to have benefits for multiple and noise suppression thanks to improved spatial sampling and azimuthal diversity. The broadband data showed improved resolution above and within the basalt, with an improved signal-to-noise ratio and continuity of events below the basalt. Both these techniques help to reveal sub-basalt structures in their separate ways. The BroadSeis proprietary receiver deghosting is fully 3D and so is perfectly suited to wide-azimuth acquisition, as is the prototype source deghosting solution. The combination of wide-azimuth and broadband acquisition using variabledepth streamers is expected to provide an optimal solution to sub-basalt imaging. An acquisition using this combination has recently been completed, processing is underway, and the results are eagerly awaited. Acknowledgements The authors would like to thank CGGVeritas for permis sion to show this data, Gregor Duval for providing geologi cal input and the CGGVeritas processing crews: Tony Wallbank et al. in Crawley, UK and Ronan Sablon et al. in Massy, France for their help in the preparation of this article. The authors would particularly like to thank Jo Firth for her help in writing this article. References Dowle, R. [2006] Solid streamer noise reduction principles: 75 th SEG Annual Meeting. Expanded Abstracts 25, Keggin J., Widmaier, M., Hegna, S. and Kjos E. [2006] Attenuation of multiple diffractions by multi-azimuth streamer acquisition. 64 th EAGE Conference and Exhibition, Extended Abstracts, F39. Lin, D. et al. [2011] Optimising the processing flow for variable depth streamer data: First Break 29(9), Sablon, R. et al. [2011] Multiple attenuation for variable-depth streamer data: from deep to shallow water: 81 st SEG Annual Meeting. Expanded Abstracts, Soubaras, R. [2010] Deghosting by joint deconvolution of a migration and a mirror migration: 80 th SEG Annual Meeting. Expanded Abstracts, Soubaras, R. and Lafet, Y. [2011] Variable-depth Streamer Acquisition Broadband Data for Imaging and Inversion. 81 st SEG Annual Meeting. Expanded Abstracts, Williams, R. G. et al. [2011] Sub-basalt Multiple and Noise Suppression Using a Dense, Marine Wide-azimuth Recording Geometry. 73 rd EAGE Conference and Exhibition. Extended Abstracts, G41. Call for papers deadline 15 January 2012 Responsibly Securing Natural Resources 74 th EAGE Conference & Exhibition incorporating SPE EUROPEC June 2012 Bella Center Copenhagen 2011 EAGE 99