|IMPROVED SUB-BASALT STRUCTURAL IMAGING IN THE FAROE-SHETLAND BASIN USING FULL-SEQUENCE MIGRATION MULTI-VELOCITY ANALYSIS|
In January 2010 TGS completed the reprocessing of two regional long offset 2D multi-client seismic surveys in the Faroe-Shetland Basin (FSB). This reprocessing forms part of TGS’ ongoing Atlantic Margin Revival (AMR) project that extends from the FSB area to the western
Figure 1: Location map of the FSB 1999 and FSB 2000 surveys in the far northwestern UKCS crossing into Faroese territory.
Here we describe three seismic data processing approaches in the time domain. These in combination provide a dramatic improvement in the interpretability of the sub-basalt image on the reprocessed datasets which were acquired with a source and streamer configuration typical of most surveys conducted in the West of Shetland region at that time. These are (1) the enhancement of recorded low frequencies through spectral manipulation, (2) noise attenuation in several domains to maximise the signal-to-noise ratio and (3) an interpretation-lead method we term ‘Full Sequence Migration Multi-velocity Analysis’, which adopts a strategy analogous to that used in pre-stack depth migration for updating velocity models in areas of complex structure. This third approach, which is described in the most detail, is seen to be the key to improving the structural interpretability of concealed Mesozoic and Palaeozoic structures and is demonstrated with examples over the highly prospective Corona Ridge (figure 2).
Figure 2: Tectonic elements map of the
|PROBLEMS OF SUB-BASALT ACQUISITION AND PROCESSING|
The vertical and lateral inhomogeneity of basalt flows in the FSB area 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 anelastic attenuation which converts it to heat, or by elastic back-scattering. Scattering in particular poses a difficult problem due to the rugose top and base basalt and internal heterogeneity. Both give rise to complex multi-phase arrivals, significant multi-pathing and non-hyperbolic behaviour.
It is now accepted that the key to improved sub-basalt imaging is to generate and retain as much low frequency energy as possible (e.g. Ziolkowski et al., 2001). As a consequence, recent acquisition in the FSB area has seen the towing of cables and sources at increasing depths to concentrate more energy into the low frequency end of the source amplitude spectrum through constructive interference of the free surface ghost to image sub-basalt targets (e.g. White et al., 2002). Whilst accepting a deep towed streamer and source are beneficial, Gallagher and Dromgoole (2007) conclude that the sub-basalt image is primarily dependent on the processing sequence. From the reprocessing of vintage data with a shallow towed configuration similar to TGS’s FSB 1999 and 2000 surveys a significant improvement to the sub-basalt image is made through the processing of low frequencies only, iterative velocity analysis and cascaded demultiple schemes
TGS approach to sub-basalt imaging with vintage datasets
Enhancement of low frequency energy
Energy passing through the basalt is strongly attenuated at frequencies above ~ 30 Hz. At an early stage of processing we manipulate the source wavelet to recover as much energy as possible in the low frequency end of the amplitude spectrum. After applying operators to debubble and zero phase the shot gathers, a wavelet is statistically derived representing the averaged source amplitude spectrum for the FSB 1999 and 2000 datasets. The low frequency components of the wavelet are edited to generate a zero phase equivalent ‘target wavelet’ and matching operator. An unsurprising consequence of applying the matching operator is the boosting of swell and other low frequency acquisition noise alongside low frequency, low amplitude primary signal. Rather than being counter-productive, the noise becomes easier to discriminate and remove using frequency-amplitude threshold based techniques.
Multi-domain noise attenuation
Every opportunity is taken to de-noise the data in several domains. Multiple passes of noise attenuation were performed in the shot, receiver, common mid-point (CMP), common offset and common channel domains to enhance weak low frequency sub-basalt primary returns and minimise both coherent and incoherent noise. The noise attenuation methods include several iterations of a technique which decomposes the data into frequency bands and identifies and attenuates anomalous amplitudes within those bands based on time-variant thresholds. This is followed by F-X deconvolution and time/space variant dip filtering guided by primary stacking velocities. Unlike deep towed source and streamer configurations where high frequencies are irrecoverably lost due to destructive interference of the source-receiver ghost, the shallower towed configuration of the FSB surveys produces a receiver ghost notch centred at ~ 82 Hz. By minimising the effects of scattering through this approach, the high frequency content of the overlying Tertiary section is not affected.
Full sequence migration multi-velocity analysis
After spectral manipulation, noise attenuation, SRME and predictive deconvolution in the tau-p domain to remove intra-basalt multiples, we apply a time domain approach analogous to interpretation-driven subsalt Wave Equation Migration (WEM) scans in depth (Wang et al., 2006). At greater travel times and beneath the top basalt reflector, interpretation of seismic velocity, based upon semblance maxima and gather flattening is problematic due to relatively weak primary returns and structural complexity. For this reason, an easier velocity analysis may be made by interpretation of images derived from stacking velocities based on a percentage of an approximate input function, commonly referred to as ‘multi-velocity’ stacks.
Increased computing speed now allows the quick, full pre-stack Kirchhoff time migration of data to produce a suite of migrated images using migration velocities initially derived from selected percentages of the best set of stacking velocities. We term this ‘Full Sequence Migration Multi-velocity Analysis’ as these images have an almost complete pre-stack sequence applied and post-stack signal-to-noise enhancement to improve their interpretability. Velocity-based demultiple is performed immediately before migration to effectively remove residual multiple for each migrated velocity panel using the corresponding percentage velocity function. We generate a suite of up to 15 pre-stack migrated panels using Radon demultiple and migration velocities scaled typically within the 60-140% range, which are interpreted using the multi-velocity picking tool within TGS’ PRIMA software. Picks on the migrated panels are related to gathers and semblances interactively. The data are then re-migrated with the updated migration velocity field and further iterations are performed if necessary.
Figure 3 shows selected panels from Full Sequence Migration Multi-velocity Analysis from a reprocessed FSB 1999 line transecting the Corona Ridge. Highlighted are examples of events which give a preferred image, corresponding to the analysis picks made. The left side of the image (northwest) demonstrates thickening of the Palaeogene flood basalts compared to the right side (southeast) showing extensive piles of sills intruded into the Flett sub-basin (Figure 2). Figure 4 shows the composite image obtained from the update after re-migration and stacking over the initial (100%) reference velocity function. The final reprocessed time image (Figure 5(b)) transecting many of the major intra-basinal highs and sub-basins in the Faroe-Shetland Basin shows a dramatic improvement over the original version (Figure 5(a)), particularly with respect to previously poorly imaged Mesozoic and Palaeozoic structure, but without compromising the frequency content of the post-basalt Tertiary section.
Figure 3: Zoom of selected Full Sequence Migration Multi-velocity Analysis panels across the northern Corona Ridge with the percentage velocity variation relative to the initial 100% function annotated. Examples of preferred images are ringed.
Figure 4: 2D Kirchhoff pre-stack time migration image from the initial (100%) reference function (left) and updated image (right) after re-migrating and stacking with the derived Full Sequence Migration Multi-velocity Analysis function.
We demonstrate a significant improvement in image quality through the reprocessing of our FSB surveys using the three key approaches outlined. Careful attention to the preservation and enhancement of low frequency signal and the attenuation of noise in multiple domains are seen to be crucial to improving signal-to-noise beneath the basalt. We show there is no need to compromise the frequency content of the overlying Tertiary section when reprocessing shallow towed datasets for sub-basalt targets. The third approach, Full Sequence Migration Velocity Analysis, is seen to be the key to improving the structural interpretability of concealed Mesozoic and Palaeozoic structures during the velocity analysis stage. This procedure can now be run within the time frame of a quick turnaround large scale production project. It permits the early involvement of interpreters, to whom the choice of a range of alternative structural images may be presented, each with the processing and signal-to-noise characteristics of a final section, and a ‘composite’ image built up from these. The final time domain migration velocity field should provide an excellent reference model for initial passes of pre-stack depth migration and tomography-based velocity updates.
Figure 5: Comparison of the final time migrated section across the Faroe-Shetland Basin; (a) original and (b) reprocessed FSB 1999 time image. Annotated above the reprocessed image are features which can be related to the major intra-basinal highs and basins shown in Figure 2: Fug R = Fugloy Ridge; Ste B = Steinvør Basin; EFH = East Faroe High; Cor B/Ssl B = Corona/Sissal Basin; Cor R = Corona Ridge; Fltt SB = Flett Sub-Basin; Fltt R = Flett Ridge; Fou SB = Foula Sub-Basin; RR = Rona Ridge; WSB = West Shetland Basin.
The authors would like to thank TGS for permission to show the data examples, David J. Moy for the provision of the structural element map, and the hard work of all in the TGS Bedford office involved in the reprocessing of the FSB datasets.
Gallagher, J.W. and Dromgoole, D., 2007. Exploring below the basalt, offshore Faroes: a case history of sub-basalt imaging. Petroleum Geoscience, 13, 213-225.
Moy, D.J. 2010. The Architecture, Growth and Tectono-Stratigraphic Significance of Rift-Oblique Lineaments on the
Wang, B., Dirks, V., Guillaume, P., Audebert, F., Hou, A., and Epili, D., 2004. 3D sub-salt tomography based on wave equation migration perturbation scans. Extended Abstracts, SEG Annual International Meeting.
White, R.S., Christie, P.A.F., Kusznir, N.J., Roberts, A., Davies, A., Hurst, N., Lunnon, Z., Parkin, C.J., Roberts, A.W., Smith, L.K., Spitzer, R., Surenda, A., and Tymms, V., 2002. iSIMM pushes frontiers of marine seismic acquisition. First Break, 20 782-786.
Ziolkowski, A., Hanssen, P., Galiff, R. and Li, Xiang-Yang, 2001. The Use of Low Frequencies for Sub-Basalt Imaging. Extended Abstracts, SEG Annual International Meeting.