Complex structural imaging of transition zones in Bohai Bay, China, by OBC technology

Complex structural imaging of transition zones in Bohai Bay, China, by OBC technology Xiaogui Miao 1, Wei Yan 1, Yongxia Liu 1, Joe Zhou 2, and Kunlun Yang 2 Abstract A series of ocean-bottom cable (OBC) surveys has been conducted in the Bohai Sea in China in recent years to overcome difficulties experienced with streamer surveys in shallow water, such as strong currents, missing near offsets, and obstacles. The main challenges in OBC data imaging include steeply dipping structures, serious multiples in the shallowwater environment, large lateral velocity variations, fault shadow effects, and low signal-to-noise ratio (S/N). To obtain optimal images, advanced processing technologies have been developed and applied to OBC data which involve effective PZ summation and shallow-water demultiple, a high-fidelity beam migration in the wide-azimuth domain, and accurate velocity-model building in 3D tilted-transverse-isotropy (TTI) media. The PZ summation and shallow-water demultiple methods aim to effectively eliminate shallow-water ghosts to achieve broadband seismic data. Furthermore, high-fidelity controlled-beam migration (CBM) and TTI velocity-model updates greatly enhance steep dip imaging, improve S/N, and reduce turnaround time. Through the combination of these technologies, OBC data processing provides high-quality images with well-defined steeply dipping structures to reduce exploration risk in the Bohai area. strike-slip faults and some complex, steeply dipping structures and fractures. As a seaward extension of the Liaohe, Dagang, and Shengli oil fields, the Bohai Sea sedimentary basin is in the Neozoic sedimentary center of the North China Basin. The oilbearing rock system covers more than 24,000 km 2 and has great exploration potential (Fan and Song, 2000). Streamer surveys encounter problems when conducted in such a shallow-water transition zone, such as strong currents, congested areas (e.g., production rigs), gas clouds, and missing near offsets. Ocean-bottom-cable (OBC) acquisition (Figure 2) provides an alternative and has been developed to overcome all the above problems by laying cables on the seabed (Soudani et al., 2006). It is used extensively in transition zones and is generally superior to towed-streamer acquisition, partly because of the use of dual-component sensors for deghosting or the use of fourcomponent (4C) sensors for converted-wave imaging. Furthermore, OBC acquisition can easily record wide-azimuth seismic data to handle complex dipping structures. Introduction The Bohai Sea consists of three bays: Liaodong, Bohai, and Laizhou Bay (Figure 1). It is in the shallow-water transition zone of the Bohai Straits with an average water depth of 18 m and a maximum depth of 70 m. The Tan-Lu fault system passes through the field, resulting in several major secondary Figure 1. The Bohai Sea. Figure 2. Seabed acquisition with (a) Sercel SeaRay OBC cable and (b) SeaRay flatpack on the sea bottom. 1 CGG (Beijing). 2 CGG (Singapore). http://dx.doi.org/10.1190/tle33111256.1 1256 THE LEADING EDGE November 2014 Special Section: China

However, to obtain a clear and reliable seismic image, several key steps should be considered during data processing. First, the Bohai seabed consists mainly of silts and fine sands which cause serious surface-related multiples. PZ (hydrophone-geophone) summation to cancel receiver side ghosts and shallowwater demultiple to eliminate other surface-related multiples are important. Second, imaging quality and accurate positioning of the strike-slip fault systems are key factors for further exploration (Zhou et al., 2011). Wide-azimuth imaging, using advanced depth-migration technologies, is required to image complex, steeply dipping structures with large lateral velocity variations. Accurate velocity-model building with tilted-transverse-isotropy (TTI) anisotropy and high-fidelity controlled-beam migration (CBM) are applied to improve the structural image and the signal-to-noise ratio (S/N). OBC equipment and acquisition Sercel s SeaRay redeployable OBC cable is used widely in Bohai exploration projects because of its reliability and flexibility. The system can record large spreads over wide blocks (n 1000 km 2 ) with high productivity (as many as 8500 pops per day) (Figure 2a). It offers capabilities of high channel count and long offset. Low power consumption and high data-transmission rates allow for deployment of cable lengths of as much as 35 km down in water depths of as much as 500 m. The system supports as many as 6000 4C receivers deployed in multiple lines, each with power and telemetry redundancy. The redeployable OBC cable is made up of sections, each based on a continuous lightweight cable that links 10 receiver points. The sensor package is a flatpack (Figure 2a) that provides optimum coupling with the seabed and avoids rolling. The omnidirectional 3C digital sensor unit (AQDSU) provides superior vector fidelity, as confirmed by first-break XY hodograms and spectral matching (shots at 45 ). Microelectromechanical-system (MEMS) accelerometers can record wide broadband signals (linear 0- to 800-Hz responses) with low distortion ( 90 db). In 3C configurations, they are insensitive to any inclination of the sensor package on the seafloor. After tilt correction, data are recorded as if the Z component were perfectly vertical and the XY perfectly horizontal. For 3D operations, there are two acquisition patterns patch and swath. Patch acquisition is based on the roll by one full template, both inline and crossline. Operations are therefore often discontinuous. Shooting is orthogonal and is extended by half of the maximum offset inline and crossline to keep the fold of coverage constant. Azimuthal distribution is wide, but near offsets are missing, which might be a problem for data processing (e.g., PZ summation). Azimuth and offset-preserving data interpolation should be applied to facilitate wide-azimuth processing. For more continuous operations, swath acquisition is based on the roll by one or more cables, mostly in the crossline direction. Dense shooting is done parallel to receiver lines, often in flip-flop mode. This results in a narrow-azimuthal distribution, which might be a problem for subsurface illumination. To avoid narrow-azimuth coverage, more OBC cables need to be laid out with more overlap, which is, of course, more expensive. Figure 3. PZ summation to remove receiver ghost. Ghosts have opposite polarities in P and Z components, respectively. PZ summation In shallow-water cases, the notch frequency generated by receiver ghosts usually appears in the dominant frequency band. PZ summation is intended to remove receiver ghosts by summing P and Z data. Ideally, the receiver ghosts recorded in P and Z components have different polarities. Through summing, the ghosts are canceled out (Figure 3). The key step in PZ summation is data calibration. In shallow-water environments, the ghost arrives shortly after the primary, with the traveltime difference between them usually less than one wavelength, which imposes serious challenges for PZ calibration. A cross-ghosting method has been developed (Soubaras, 1996) to overcome this problem. It uses the geophone s ghost function to crosscorrelate with hydrophone data, and vice versa, which can be applied in a deeper window free of various noise types. The method then takes the least-squares inversion to compute an inverse calibration operator for each receiver. Traditionally, we calibrate the geophone wavefield to the hydrophone wavefield because the hydrophone records the data with a higher S/N, whereas geophone data are noisier. However, the resolution of the resulting data is usually similar to that of the hydrophone, which is usually lower than that of the geophone. If the hydrophone is calibrated to the geophone data when the geophone data are not noisy but after particularly effective noise attenuation, the resolution of the results can improve significantly. We therefore developed a method to calibrate the hydrophone to the geophone wavefield and thus achieve better ghost attenuation and higher resolution after PZ summation in the real data. Figure 4 shows a detailed comparison of PZ summation. Figures 4a and 4c show the stacks of P and Z, respectively; Figure 4b shows the PZ summed result. It is easy to see that the P stack has high S/N, whereas the Z stack is of higher resolution but is noisier. After PZ summation, the high-resolution events (marked as green circles) from the Z component are added, 1258 THE LEADING EDGE November 2014 Special Section: China

whereas the good S/N of the P component is retained in the summed result. The average water depth in this case is shallow, about 14 m, which makes the multiple periods about 19 ms. The OBC ghosts become a problem because the notches fall within the seismic bandwidth. For this example, the notch frequency is about 54 Hz. Figure 4d shows the amplitude spectra of the hydrophone (blue line) and geophone (black line) and the upgoing wave (red line). From the amplitude spectra, it is noticed that the hydrophone data have the notch at about 48 Hz, and the geophone data have the notches at about 24 Hz and 78 Hz. Figure 4. Comparison of PZ summation. Stacks of (a) P and (c) Z, respectively; (b) PZ summation stack. (d) Amplitude spectra of the hydrophone (P) (blue line), geophone (Z) (black line), and Phase shift of approximately 180 is between the geophones and hydrophones. upgoing wave after PZ summation (red line). In this case, the frequency notch is at about 54 Hz. For the P component, represented by the blue curve, the notches are at 48 Hz and so on. For the Z After summation, the spectrum of the data indicated by the black curve, the notches are at 24 Hz, 78 Hz, and so on. After summation, upgoing wave no longer has the first the upgoing wave (in red) eliminates the notches between 30 and 80 Hz. notch between 30 and 80 Hz (see the red line in the amplitude spectra in Figure 4d), which means the receiver-side ghosts are effectively removed. Residual multiples, including source-side ghosts and peg-legs, need further demultiple processing. Shallow-water demultiple PZ summation can attenuate receiver-side ghosts; therefore, the output of PZ summation should be free of such ghosts but would retain peg-legs and other multiples, including source-side multiples. The difficulty in applying surface-related multiple elimination (SRME) on OBC surveys is caused by the asymmetric raypaths of OBC data, which result in a lack of sources at receiver positions and/or a lack of receivers at source positions. Thus the raypaths of surface multiples cannot be constructed from primaries as in streamer data, in which sources and receivers are both at the sea surface. However, the interferometry method, which crosscorrelates two recordings of wavefields at two receiver positions, can lead to the Green s function that would be observed at one of those receiver positions as if there were an impulsive source at the other location, but without any prerequisite knowledge of subsurface (Yang et al., 2014). Wapenaar and Fokkema (2006) describe the basic theory of seismic interferometry. The OBC survey, as illustrated in simple form in Figure 5, can be considered as a half plane enclosed by a free surface. The additional required raypaths or Green s functions to predict the source-side multiple, indicated in Figure 5a, and the receiver-side multiple, indicated in Figure 5b, can be achieved by intersource and interreceiver interferometry integrating the crosscorrelation of receiver points at the ocean bottom indicated by S i S j or shot points at the surface indicated by R i R j, which have been acquired in conventional OBC surveys. Note that in Figure 5, we plot only the raypath related to the water layer for simplicity. However, the result from Figure 5. (a) Interferometry obtaining the Green s function (blue arrows) to predict OBC shot-side surface-related multiples; (b) receiver-side illustration. interferometry is as if there were an impulsive source at every receiver point or a receiver at every source point. These raypaths are ideal components for applying SRME on OBC. Again, the reconstructed wavefield can include reflectors that go beyond the water bottom. With the extra data from interferometry, it is straightforward to predict OBC multiples using SRME-type methods with crossconvolution. However, instead of combining streamer data with OBC data, here we predict both source-side and 1260 THE LEADING EDGE November 2014 Special Section: China

receiver-side surface-related multiples with only the acquired OBC data. In summary, taking deghosted seismic data as input, on the source side, we compute intersource operators using an interferometry method and convolve the shot gathers and intersource operators to predict source-side surface multiples. We then perform the same steps on the receiver side to predict receiver-side multiples. We have successfully applied this method on an OBC survey acquired in the Bohai Sea. Water depth of the case survey varies from 21 to 30 m. The top part of Figure 6a shows a stack section that has had noise removal and PZ summation applied (but before any demultiple). The bottom part of Figure 6a shows the autocorrelation of the stack section, with the ringing around the zero lag indicating strong short-period multiples. Figure 6b shows a stack in the top part and an autocorrelation below it after multiple removals using our method (Yang et al., 2014). It can be observed that considerable peg-leg multiple energy has been attenuated, resulting in better-defined primary events because the interference from short-period multiples on the wavelet is much reduced after application of our interferometry demultiple method. High-fidelity beam migration Prestack Gaussian-beam migration has become a useful alternative to Kirchhoff and wave-equation migrations for accurate seismic imaging of structurally complex areas (Hill, 2001). It overcomes the limitations of Kirchhoff migration in imaging multipathing arrivals while retaining flexibility with input geometry and topographic variations. Such flexibility is especially important to OBC data processing in which irregular oceanbottom topographic variations might occur. This migration method can image complicated geologic structures, including steep dips, in areas where seismic velocity varies rapidly. Gaussian-beam migration (Zhu et al., 2006; Gray et al., 2009) begins by decomposing a common-shot, common-offset record into small patches using overlapping Gaussian windows, which form a function of partition of unity (POU). Local slant stacks of these windowed data patches are then generated. Using the beam summation and imaging condition, the image at subsurface x can be obtained. As a specialized version of beam migration, high-fidelity controlled-beam migration enhances signal-to-noise ratio and images steep dips (Zhou et al., 2011). This powerful imaging tool can deliver clear, easy-to-interpret structural images in complex areas (Notfors et al., 2006). CBM has been used widely for velocity-model building, structural imaging, and imaging of sparse and noisy data. As a local slant-stack migration, CBM operates in localized space and localized angle; hence, it can handle multipathing naturally (Zhou et al., 2011). TTI velocity-model building Sediment dip can be as much as 60 or higher in basins that feature strike-slip faults. It is quite obvious that a TTI velocity model is required because neither isotropic nor VTI models can describe the dip-dependent velocity and flatten the commonimage gathers (CIG) at the same time (Zhou et al., 2011). OBC surveys take advantage of wide-azimuth acquisition if enough Figure 6. (a) The top image shows a P-wave stack after noise and ghost removal but before demultiple, and the bottom shows its autocorrelation before demultiple. (b) The top image shows a demultiple stack using OBC-SWD, and the bottom image shows its autocorrelation after demultiple. cables are laid out, as mentioned earlier. TTI anisotropy needs to be handled properly in a wide-azimuth manner. TTI anisotropy was introduced after a few iterations of isotropic tomographic update. The smoothed isotropic velocity field was then calibrated at various well locations to provide the initial anisotropy parameter δ. Because of the relatively shallow penetration of the wells, the epsilon was estimated based on a scan of the ε/δ ratio. The vertical velocity was updated further by more iterations of tomographic update until the CIGs were flat. The dip and azimuth angle fields were repicked after each iteration, based on the updated depth volume (Zhou et al., 2011). Taking advantage of the wide-azimuth acquisition, we generated CIGs in two orthogonal directions during migration velocity analysis. Each CIG consisted of offsets along inline (right) and crossline (left) directions, so that the tomographic velocitymodel updating was in a true 3D manner. Figure 7 shows CIG comparisons before and after the TTI anisotropic update. In the CIGs with an isotropic model, events dipping in the crossline direction need slower velocities, whereas events dipping the inline direction need velocities faster than those in the crossline direction but still slower than the best velocity needed to flatten the gather. The best approximation here is to honor the layering of the geologic structure and introduce TTI anisotropy. After TTI anisotropic updating, the CIGs for the dipping events are much flatter (Zhou et al., 2011) (Figure 7b). 1262 THE LEADING EDGE November 2014 Special Section: China

Impact on interpretation The primary objective of the survey was to image the major fault systems and accurately position the boundary faults, which is important for subsequent interpretation to identify the associated oil and gas reservoirs. In Figure 8, we compare images of Kirchhoff PSTM and high-fidelity CBM (HFCBM) converted to time. It is obvious that HFCBM has enhanced the S/N of the image and steep dips. The major fault system, which is unclear in Kirchhoff migration, is well defined and easy to interpret based on client feedback. The secondary target of the processing was to image the top boundary of buried hills. It is difficult to identify this boundary on the Kirchhoff PSTM section. However, because HFCBM preserves the relative amplitudes, the top boundary can be interpreted much more easily in the HFCBM section (Figure 8b). Truncation of the fault structures is also critical to characterize shallow oil and gas targets (< 2500 m). It is very hard to interpret shallow faults around 2 s in the middle of the Kirchhoff section, whereas with the HFCBM result, image quality is improved greatly, and S/N is enhanced as well (Zhou et al., 2011). References Fan, Z., and C. Song, 2000, The Bohai sea oil exploration and environmental protection: SPE International Conference on Health, Safety and Environment in Oil and Gas Exploration and Production, SPE conference paper 61011-MS, http:// dx.doi.org/10.2118/61011-ms.28. Gray, S. H., Y. Xie, C. Notfors, T. Zhu, D. Wang, and C.- O. Ting, 2009, Taking apart beam migration: The Leading Edge, 28, no. 9, 1098 1108, http://dx.doi.org/10.1190/ 1.3236380. Hill, N. R., 2001, Prestack Gaussian-beam depth migration: Geophysics, 66, no. 4, 1240 1250, http://dx.doi.org/10.1190/ 1.1487071. Notfors, C. D. B., Y. Xie, and S. Gray, 2006, Gaussian beam migration A viable alternative to Kirchhoff?: 68th Conference and Exhibition, EAGE, Extended Abstracts, G046. Soubaras, R., 1996, Ocean bottom hydrophone and geophone processing: 66th Annual International Meeting, SEG, Expanded Abstracts, 24 27, http://dx.doi.org/10.1190/1.1826611. Soudani, M. T. A., J. L. Boelle, P. Hugonnet, and A. Grandi, 2006, 3D methodology for OBC pre-processing: 68th Conference and Exhibition, EAGE, Extended Abstracts, B044. Conclusions We have demonstrated the advantage of OBC data acquisition using SeaRay and advanced imaging technologies through the Bohai case study. Compared with a conventional raybased Kirchhoff migration technique, HFCBM provides superior structural imaging for the area with complex geology and steep faults and improves the signal-to-noise ratio in the data effectively. Combined with PZ summation, shallow-water demultiple, and TTI model building, this workflow provides a promising solution for resolving imaging issues in the complex fault system and in the middle to deep low-s/n zone in the Bohai area. Editor s note: This article was submitted for the China special section in August TLE. Figure 7. Common-image gathers of HFCBM migration. Each CIG consists of offsets along inline (right) and crossline (left) directions (a) before and (b) after TTI anisotropic update. Acknowledgments The authors would like to thank Bin Zhou, Zhiliang Wang, and Chao Wu of CNOOC Ltd. Tianjin for their help during processing and data interpretation; Denis Mougenot of Sercel and Sara Pink-Zerling of CGG for their help in preparing the article; and Botao Li of CGG Beijing for processing the data. We would also like to thank CNOOC Ltd. Tianjin and CGG for permission to publish this work. Corresponding author: Xiao-Gui. Miao@CGG.com Figure 8. Migration comparison of (a) old Kirchhoff PSTM section and (b) HFCBM PSDM section converted to time. 1264 THE LEADING EDGE November 2014 Special Section: China

Wapenaar, K., and J. Fokkema, 2006, Green s function representations for seismic interferometry: Geophysics, 71, no. 4, SI33 SI46, http://dx.doi. org/10.1190/1.2213955. Yang, K. L., L. B. Liu, B. Hung, J. Zhou, B. Zhou, Z. L. Wang, and X. D. Gong, 2014, OBC surface multiple prediction using seismic interferometry: 76th Conference and Exhibition, EAGE, Extended Abstracts, http://dx.doi.org/10.3997/2214-4609.20141032. Zhou, B., J. Zhou, Z. Wang, Y. Guo, Y. Xie, and G. Ye, 2011, Anisotropic depth imaging with high fidelity controlled beam migration: A case study in Bohai, offshore China: 81st Annual International Meeting, SEG, Expanded Abstracts, 217 221, http://dx.doi. org/10.1190/1.3627641. Zhu, T., S. Gray, and D. Wang, 2006, Prestack Gaussian-beam depth migration in anisotropic media: 76th Annual International Meeting, SEG, Expanded Abstracts, 2362 2366, http://dx.doi. org/10.1190/1.2370009. 1266 THE LEADING EDGE November 2014 Special Section: China