A047 Simultaneous-source Acquisition in the North Sea Prospect Evaluation

Transcription

1 A047 Simultaneous-source Acquisition in the North Sea Prospect Evaluation B. Szydlik* (WesternGeco), C.J. Beasley (WesternGeco) & I. Moore (WesternGeco) SUMMARY This paper outlines a method for quantitative evaluation of marine simultaneous-source separation processes. The method doesn t require acquiring the simultaneous-source data beforehand. Instead, it generates simulated simultaneous-source from available sequential-source data to emulate as close as possible desired simultaneous-source dataset. We demonstrate how this method enables quantitative assessment and show results for undershoot data from the North Sea.

2 Introduction Simultaneous-source methods are well established for land data, and have a proven history of providing substantial increases in acquisition efficiency (Bagaini, 2010). Although introduced more than a decade ago (Beasley et al., 1998), methods for marine acquisition have seen slower development. This is largely caused by the extra source constraints associated with marine acquisition and the cost of fielding extra source boats. Marine airgun source encoding is generally limited to timing dithers, and the sources must move continuously. Despite these limitations, it has been shown recently that commercial use of simultaneous-source acquisition for marine data is viable (Moore et al., 2012). Qualitative improvements in imaging and multiple attenuation have been demonstrated for marine wide-azimuth, subsalt acquisition (Beasley et al., 2012), but quantitative measures are needed to demonstrate, before acquisition, that the value of simultaneous-sources will outweigh any possible degradation associated with crosstalk between sources. We developed a technique to evaluate the suitability of simultaneous-source acquisition for a particular prospect area in one key aspect: the quality of source separation. Significantly, it can address the benefits of simultaneous-source acquisition due to improved efficiency, but not due to improved sampling and so it is a conservative measure. This paper outlines this evaluation method, demonstrates how it enables such an assessment and shows results for undershoot data from the North Sea. The evaluation method uses simulated simultaneous-source data. Evaluation method Typically, processing of simultaneous-source data requires separation of the sources at an early stage, and the quality of this separation process, along with the potential benefits of simultaneous-source acquisition, then determines the suitability of the method. In this paper we consider separation via sparse inversion, which is known to give near-perfect results on simple, synthetic data even if those data are aliased (Moore et al., 2008, Akerberg at al., 2008, Moore, 2010). For field data however, the complexity of the data in the separation domain can have a significant impact on the separation quality, and this needs to be evaluated on a prospect-specific basis. Ideally one would acquire simultaneous-source test data and compare with equivalent data acquired with ordinary, sequential-source shooting. However, acquiring test data is costly, and for the datasets to be equivalent, the environmental conditions must be the same and the source and receiver positions must be repeated accurately. In practice, this is rarely the case, making it difficult to determine whether differences are due to simultaneous sources, or to other issues such as differences in noise or a lack of repeatability. Simulation of simultaneous source data can be a better controlled alternative. We simulate simultaneous-source data by selecting line(s) from existing data that are close to the desired simultaneous-source acquisition. We select shots that emulate the simultaneous source acquisition, apply timing dithers (as would be used in acquisition) and sum them. After separation, any differences between the separated and original shots would generally indicate the quality of separation, however, ambient noise is not modeled in the separation process and can also be a factor. Qualitative and quantitative measures of the differences can be used to test various aspects of a proposed simultaneous-source acquisition for a specific geological setting and survey objective. For example, the impact of the shotpoint interval (SPI) can be evaluated (though this may require interpolation of the sequential-source datasets) as can the effect of the dithering scheme. Standard quantitative measurements used in 4D seismic analysis such as NRMS and predictability (Kragh at al., 2002) can be used to assess any observed differences. North Sea example The data used in this example comes from a two-boat undershoot portion of the Norne field survey that was acquired with a single source at 25-m SPI. The receiver positions of the undershoot data (S 2 )

3 essentially duplicate those of the close-pass single vessel primary line (S 1 ). Acquisition of these two sets of lines in a single pass would significantly reduce acquisition cost and HSE risk, but requires the use of simultaneous sources if the shotpoint interval and record length is to be maintained. We acquired a simultaneous source test dataset with even denser 12.5-m SPI to enable separation testing at both 12.5 and 25-m SPI, albeit 12.5-m SPI acquisition limits the trace length. The simultaneous-source data (Figure 1a) were separated using the algorithm presented by Moore et al., (2008) and the undershoot (S 2 ) stack (Figure 1c) was compared to that of the corresponding sequential-source data (Figure 1b). To avoid any issues due to different stack fold, we decimated the acquired simultaneous-source data to 25-m SPI before stacking. In general, this is a conservative approach as it does not take advantage of possible benefits of the improved sampling that can derive from simultaneous sources. We addressed issues relating to non-repeatability (source signature variations, tidal statics, etc.) as far as possible, though the positioning differences showed on Figure 1e, could not be handled at this early processing stage. The stack difference (Figure 1d) contains a significant amount of coherent energy. It is difficult to determine to what extent the difference is due to non-repeatability, rather than imperfect separation. To isolate the source separation effects as much as possible, a simulated simultaneous-source dataset was created from the sequential-source datasets using shot interpolation when necessary. Separation was performed at both 12.5-m and 25-m SPIs. In addition, demultiple was applied to evaluate the suitability of separated data for further processing. The results for S 2 (Figures 2b and 2c) were compared to demultiple sequential-source data (Figure 2a). The difference sections (Figures 2d and 2e) are now free of effects related to acquisition non-repeatability. Little signal is visible in either case, though the 25-m SPI difference section is noisier (20% NRMS for 25-m SPI and 14% NRMS for 12.5-m, measured in shallow time window). We see no obvious detrimental effects related to the application of demultiple processing to separated simultaneous-source data. Ideally the final evaluation would use fully processed products. The suitability of processed data for inversion is often a concern. Comparisons of sequential-source and simulated simultaneous-source CMP gathers (Figure 4) indicate that the difference is mainly incoherent noise and is attenuated by multichannel processes such as stacking, AVO and migration. Graphs of AVO (Figure 3e) for the strong event at 2.42 s suggest that AVO products are unlikely to be affected by the use of the simultaneous-source acquisition technique. Discussion and Conclusions The comparison between acquired simultaneous-source and sequential-source data is difficult because there can be significant differences in the datasets that are unrelated to the use of simultaneous-source. Simulation of simultaneous-source data allows the risks associated with simultaneous-source acquisition to be evaluated on a prospect-specific basis, using already existing sequential-source data. The acquired simultaneous-source data are not required for the evaluation method, but if available, they can be used in additional qualitative analysis particularly to evaluate the potential benefits of improved sampling. Our example demonstrates that simultaneous-source data can be processed without compromising either pre- or post-stack products for typical SPIs and, as expected, errors increase with larger SPIs Simultaneous-sources could significantly reduce cost and HSE exposure by reducing the number of boat passes made close to the platform. As stated above, simultaneous-source acquisition has been shown to be beneficial in several settings (single-boat, flip-flop and wide-azimuth surveys), however, we recommend that an evaluation should be part of the survey design process for each individual survey, especially in complex areas. Although useful, our approach comes with some caveats. Shot interpolation is often required to simulate the improved shot interval frequently obtained with simultaneous-source acquisition, but can itself generate separation errors not associated with simultaneous-source data. Moreover, the ambient noise level in simulated simultaneous-source data is typically higher than that in real simultaneous-

4 source data because two copies of the noise are summed in the simulation process. Further, such evaluations tend to be 2D and thus do not benefit from 3D migration. Finally, the important aspect to keep in mind is that simulation can t show the benefits of actually recording extra shots which can only be demonstrated before acquisition with data modelling. As these issues don t work in favour of simultaneous-source acquisition, we consider our method to be conservative. Acknowledgments The authors would like to acknowledge Statoil and the Norne license partners Eni Norge and Petoro for permission to acquire and present these data. a) b) c) d) e) 250m Figure 1 Stack sections for a) acquired simultaneous-source, b) sequential-source data, c) acquired simultaneous-source data separated at 12.5-m SPI, d) difference between 1b and 1c. e) A single CMP geometry (triangle=source, square=detector, blue=primary simultaneous-source, red= undershoot simultaneous-source, yellow=primary sequential-source, green=undershoot sequential-source) a) b) c) d) e) Figure 2 Stacks after multiple attenuation for a) Sequential-source data, b) simulated simultaneoussource data separated at 12.5m SPI, c) simulated simultaneous-source data separated at 25m SPI, d) difference between Figure 2a and 2b, and e) difference between Figure 2a and 2c.

5 a) b) c) d) e) Figure 3 CMP gathers for a) sequential-source data before demultiple, b) simulated simultaneoussource data before separation, c) sequential-source data after demultiple, d) simulated simultaneoussource data after separation at 12.5m SPI and demultiple, and e) difference between 3c and 3d. The red and yellow graphs indicate the AVO functions for the event at 2.42 s on 3c and 3d References Akerberg, P., Hampson, G., Rickett, J., Martin, H., and Cole, J. [2008] Simultaneous source separation by sparse Radon transform. 78 th Annual International Meeting, SEG, Extended Abstracts, Bagaini, C. [2010] Acquisition and processing of simultaneous vibroseis data. Geophysical Prospecting, 58, Beasley, C.J., Chambers, R.E. and Jiang, Z. [1998] A new look at simultaneous sources. 68 th Annual International Meeting, SEG, Expanded Abstracts, Beasley, C.J., Dragoset, W. and Salama, A. [2012] A 3D simultaneous source field test processed using alternating projections: a new active separation method. Geophysical Prospecting, 60 (in press). Kragh, E., and Christie, P. [2002] Seismic repeatability, normalized rms, and predictability. The Leading Edge, 21, Moore, I., Dragoset, W., Ommundsen, T., Wilson, D., Ward, C. and Eke, D. [2008] Simultaneous source separation using dithered sources. 78 th Annual International Meeting, SEG, Expanded Abstracts, Moore, I. [2010] Simultaneous sources Processing and applications. 72 nd EAGE Conference and Exhibition, Extended Abstracts, B001. Moore, I., Monk, D., Hansen, L. and Beasley, C. [2012] Simultaneous sources: The inaugural fullfield, marine seismic case history from Australia. 22 nd ASEG Conference and Exhibition, Expanded Abstracts.