Application of blended sources offshore Abu Dhabi C.D.T. Walker 1*, G. Ajlani 1, M. Hall 2, S. Al Masaabi 3, A. Al Kobaisi 3, G. Casson 3 and H. Hagiwara 3 present applications of the pseudo-random shot-point interval method offshore Abu Dhabi. Introduction Simultaneous or blended sources were initially developed for onshore applications using vibratory sources with dramatic improvements in both data quality and operational performance (Foster et al., 2010; Al-Ghamdi et al., 2010; Al-Mahrooqi et al., 2012). Offshore, there have been a number of both towed streamer and ocean bottom surveys acquired (Moore et al., 2012; Moldovaneau et al., 2013; Walker et al., 2013; Abma, 2014) where both data quality through higher fold and improved survey efficiency have again been amply demonstrated. The improvement in survey efficiency arising from simultaneous sources is clearly illustrated in Figure 1 (Etgen et al., 2015) where the combination of improved seabed receiver deployment rates and multiple simultaneous sources reduces the time needed to acquire a 400 km 2 survey from 108 days to 21 days a staggering 80% reduction in survey duration. There have been a number of approaches applied in these marine cases to ease the task of separating the desired shots from the unwanted ones dithering the sources by small differences in firing time, distance separating the shooting vessels, firing each shot at pseudo-randomly distributed shot-point locations and operating the source vessels completely independently. In this article, we will focus on the application of the pseudo-random shot-point interval method offshore Abu Dhabi, where the very high amplitude levels of shot-generated low velocity Scholte wave noise (mud roll) represent a particular challenge given the long offsets cross-spread geometries needed to image the deep zone(s) of interest. Data quality offshore Abu Dhabi In areas with hard water bottom conditions, geophysical imaging becomes more difficult mainly due to increased reverberations and distorted velocities. This is dramatically observed in the Barents Sea where seafloor multiples echo as many as seven times before their energy drops below the noise floor obscure low amplitude target reflections and complicate accurate velocity analysis (Webb et al., 2010). Where the water depth is also shallow, the impact of source generated guided waves, comprised of near surface refractions and super critical multiples, on the far offsets can be even more of a challenge. This is especially evident if the velocities of these guided waves they are often dispersive are slow so that even with closely spaced receivers, they are spatially aliased (Berteussen and Sun, 2010). The source generated guided waves, as seen in Figure 2, arise because the combination of shallow water, typically less than 30 m, and the extremely high seabed reflection coefficient, usually of the order of 0.4 or more which coupled with an average seabed density of 2.1~2.2 kg/m 3, creates a very strong acoustic impedance contrast at the seafloor. Additionally, there are limestone stringers and near surface karsting in the area which are both highly spatially variable. This causes the amplitudes and phase velocities of these guided waves to be inconsistent and unpredictable, as well as changing their dis- Figure 1 400 km 2 Ocean Bottom Seismic (OBS) survey, 400 m x 400 m node grid, 50 m x 50 m shot grid. Figure 2 Shot gather offshore Abu Dhabi showing high levels of guided wave energy. 1 Seabed Geosolutions 2 GeoVectra Ltd 3 Abu Dhabi National Oil Company * Corresponding author, E-mail: cwalker@sbgs.com FIRST BREAK I VOLUME 35 I NOVEMBER 2017 59
persion characteristics. As a consequence, the ability to remove this energy is critical to imaging offshore Abu Dhabi, and the historical focus for survey design here has been to acquire surveys using orthogonal geometries with dense inline receiver spacing and high source effort. Pseudo-random shot point intervals From the various approaches that have been implemented for marine simultaneous source acquisition, Distance Separated Simultaneous Source (DS 3 ), Time Dithered Shooting, Pseudo-Random Shot Interval, Independent Simultaneous Source, one specific aspect is paramount the greater the randomness of the unwanted source(s) relative to the desired source, the greater the success in de-blending the resulting data (Abma et al., 2015). From a logistical perspective, the fewer the constraints on the relative locations of the shooting vessels, the more operationally robust the simultaneous source solution will be. Having to maintain distance separations, for instance, in producing areas with several different types of Simultaneous Operations (SIMOPS) can significantly impact both operational performance, as well as potentially giving rise to inconsistencies in both offset and azimuth distributions in the resulting data. The pseudo-random shot interval technique, patented by the Apache Corporation (US Patent number 9188693), imposes no separation requirements on the sources (Walker and Hays, 2015) and is both re-shoot and 4D friendly since the same pre-determined pre-plots can be used for any repeated shot lines if required. A number of OBS surveys have been acquired using this approach, most notably a survey by the Apache Corporation itself, where the method was employed to acquire more than 2000 km 2 of 3D data in a highly congested area of the Gulf of Mexico. By using two shooting vessels, each shooting simultaneously with a single source orthogonal to the 12 x 17.5 km receiver line spread, the survey duration was reduced by 45% (Walker et al., 2013). Test survey geometry A simultaneous source test was undertaken offshore Abu Dhabi in the autumn of 2015 utilizing the geometry shown in Figure 3 in water depths of approximately 25 m. Two shallow draft single source vessels, each equipped with an 1120 in³ source array, were shot across an ocean bottom cable (OBC) spread, configured to record continously in opposite directions. The source lines were 250 m apart, as were the receiver lines which had a 25 m receiver interval, reflecting the production survey geometry. The timing of the transits of the two shooting vessels across the receiver spread was arranged to have the minimum separation of the sources over the centre of the spread. Pseudo-random shot point intervals for each shooting vessel were pre-determined over a 12.5 m range between 18.75 m and 31.25 m to provide the equivalent shot density of the production survey shot point interval of 25 m. This pseudo-random shooting scheme resulted in 10 x 1.25 m equi-probability shot windows in which each source was fired to maximize mutual randomness between the two sources. It is important to emphasise that it requires a pseduo-random sequence to achieve this uniformity of shots being fired in each window since a purely random sequence would result in a Gaussian Figure 3 Blended source test shooting geometry. Figure 4 Hydrophone common receiver gather (CRG) illustrating the random timing of the second source shot energy relative to the desired source energy. Figure 5 Hydrophone common shot gather (CSG) showing spatial, temporal and frequency content overlap of data from the blended sources. distribution which would be biased towards the average shot interval of 25 m, thus degrading the mutual randomness between the two sets of shots. Initial results The resulting randomization in mutual shot timing is clearly evident in the common receiver gather (CRG) shown in Figure 4. This is not evident when the data are examined in the common shot domain, as shown in Figure 5. 60 FIRST BREAK I VOLUME 35 I NOVEMBER 2017
De-blending There have been many methods applied to de-blend simultaneous source data at the 2016 SEG Annual Meeting in Dallas there were 22 papers on de-blending so there is no silver bullet that necessarily works for every simultaneous source dataset. In this particular case, the very high amplitude guided wave noise arising from the shallow water/hard seabed conditions is spatially aliased, and it was thought that this might compromise the application of simultaneous sources in the area. Careful examination of the data, especially in the f-k domain as seen in Figure 6, and testing of a number of different Figure 6 f-k displays (left: blended, right: de-blended). Figure 7 Vertical geophone CSG (top: blended, bottom: de-blended). FIRST BREAK I VOLUME 35 I NOVEMBER 2017 61
de-blending methods indicated that an aggressive approach would be required, and it was decided to apply a combination rank reduction wavelet transform in the common receiver domain. This method has the advantage of allowing the removal of the (shot coherent) guided wave energy from the interfering shots, as well as the reflection and refraction energy, despite this guided wave energy being spatially aliased. The de-blending parameters were carefully selected and tested to achieve optimal attenuation of the interfering source energy while preserving both the reflection signal and the guided wave noise from the desired source. No temporal filtering was applied in the process but some high frequency incoherent noise sources were attenuated. The result of attenuating these interfering noise sources is an overall improvement in the signal-to-noise ratio. It is worth noting that we are focusing on the removal of the energy from other shots, not on suppressing the low velocity surface mud roll Scholte waves which can be clearly seen in the de-blended data in Figure 7. Of particular interest is that because the shots from each vessel are pseudo-randomly spaced, the subsequent shot from the same source vessel shown arriving at approximately 13 seconds TWT in the left hand display of Figure 7 has also been removed by the de-blending process. Record lengths longer than the shot interval can thus be generated. This could be used to decrease the shot interval and thereby increase the recorded trace density, improving both the resolution and signal-to-noise in the resulting seismic data. Robust AVO comparison Although the blended source test data only comprises two shot lines across an orthogonal OBC spread, it is possible to extract a subset of the data to examine AVO amplitudes. Data acquired using the same 1120 cu.in array fired conventionally i.e. unblended with a uniform 25 m shot point interval into the same receiver spread is used for the comparison shown in Figure 8. The robust AVO approach used, developed by Walden (1991), allows the underlying AVO in the data with high levels of noise and multiples to be mapped. The initial AVO results shown in Figure 8 are quite similar between the unblended conventional and the de-blended simultaneous acquisition. It should be noted that the data were offset limited to 900 m to avoid the very strong aliased guided wave energy, which is slightly different between the unblended and de-blended data owing to source positioning differences. Although these are quite small, less than 12.5 m or half the shot interval, the highly variable nature of this guided wave energy, noted earlier, results in observable differences in this energy between the two sets of data. The AVO intercept display from the de-blended data shows greater spatial coherency, most probably due to its improved signal-to-noise relative to the conventional unblended data that has no noise attenuation whatsoever applied. To examine this, the identical process used to de-blend the data acquired with Figure 8 Robust AVO intercept displays (top: unblended, bottom: de-blended). 62 FIRST BREAK I VOLUME 35 I NOVEMBER 2017
The AVO intercept comparison shows better agreement between the unblended and de-blended data after the incoherent noise reduction arising from de-blending has been applied to both data sets. The minor differences that can be seen are most probably a consequence of the small differences in shot locations between the unblended and de-blended shot lines. As it is never possible to exactly repeat shot locations, such differences will always be there. Having said that, these initial results exhibit similar trends, and there is no evidence that de-blending has contaminated the intrinsic AVO response in the data. Figure 9 Unblended geophone data (top: raw, bottom: after de-blending). simultaneous sources was applied to the unblended data, and the results are shown in Figure 9. The de-blending process has removed non-coherent noise from the data but left the high amplitude coherent noise untouched, as can be observed in both the x-t and f-k displays. The resulting AVO comparisons between the unblended and de-blended data, after the application of a combination rank reduction wavelet transform de-blending process in the common receiver domain, are shown in Figure 10. Conclusions Simultaneous source shooting offers the promise of significantly reducing the source time for a given ocean bottom survey without reducing the source effort. Pre-determined pseudo-random shot intervals provide a robust and repeatable way of randomizing both shots from different source vessels and also subsequent shots from the same vessel. Rank reduction wavelet transform noise suppression has been shown to be highly successful in both eliminating unwanted reflection signals, but also the spatially aliased guided wave energy that dominates the raw data in this area of offshore Abu Dhabi. These results the first time pseudo-random-based simultaneous sources have been tested in the area are very encouraging, and we believe that further studies examining the impact of post de-blending residual interference on high resolution reservoir characterization for instance will show that simultaneous source acquisition can allow the increased shot density necessary to deliver the higher quality data needed for enhanced appraisal and development. Acknowledgements We would like to thank ADNOC and Seabed Geosolutions (a Fugro-CGG joint venture) for allowing us to present this work. Figure 10 AVO intercept displays geophone data (left: unblended, centre: unblended with de-blending, right: de-blended). FIRST BREAK I VOLUME 35 I NOVEMBER 2017 63
References Foster, M.S., Howe, D., Allen, A., Buddery, D., Jack, I. and Choi, A. [2010]. Independent Simultaneous Sources Seismic Acquisition in Libya: Full Scale Implementation and New Developments. International Petroleum Exhibition and Conference, Extended Abstracts. Al-Ghamdi, T.M., Pecholcs, P.I. and Musser, J.A. [2010]. Cost-Effective, High Density, Wide Azimuth Seismic Sampling. 9th Middle East Geosciences Conference, GEO 2010. GeoArabia, 16 (4), 170. Mahrooqi, S., Rawahi, S., Yarubi, S., Abri, S., Yahyai, A., Hunt, K. and Clow, F. [2012]. DS 3 Recording for Faster Seismic Acquisition Rate without Data Compromise. 74 th EAGE Conference & Exhibition, Extended Abstracts. Moore, I., Monk, D., Hansen, L., and Beasley, C. [2012]. Simultaneous sources: The inaugural full-field, marine seismic case history from Australia. 22 nd International Geophysical Conference and Exhibition, Extended Abstracts. Moldoveanu, N., Quigley, J. and Beasley, C. [2013]. Simultaneous Shooting for Marine Acquisition Could We Catch Up with Land? 75 th EAGE Conference and Exhibition, Extended Abstracts. Walker, C., Hays, D. and Monk, D. [2013]. Blended Source Single Sided Full Azimuth Ocean Bottom Seismic Acquisition. SEG Annual Meeting, Expanded Abstracts. Abma, R. [2014]. Shot Scheduling in Simultaneous Shooting. SEG Annual Meeting, Expanded Abstracts. Ourabah, A., Keggin, J., Brooks, C., Ellis, D. and Etgen, J. [2015]. Seismic Acquisition, what really matters?. SEG Annual Meeting, Extended Abstracts. Webb, B., Sazykin, A., Grechishnikova, T., Barkov A. and Yakovlev, I. [2010]. 3D De-multiple over Shtokman A Case Study. 4 th EAGE International Conference and Exhibition on Geosciences. Extended Abstracts. Walker, C. and Hays, D. [2015]. Moving Nodes into Exploration. Oilfield Technology, 8 (4). Berteussen, K. and Sun, Y. [2010]. A critical look at the acquisition parameters for sea bottom seismic in the Arabic Gulf. SEG Annual Meeting, Expanded Abstracts. Abma, R., Howe, D., Foster, M, Ahmed, I., Tanis, M., Zhang, Q., Arogunmati, A. and Alexander, G. [2015]. Independent simultaneous source acquisition and processing. Geophysics 80 (6), WD37-WD44. Walden, A.T. [1991]. Making AVO sections more robust. Geophysical Prospecting, 39 (7), 915-942. ADVERTISEMENT 17 TH INTERNATIONAL CONFERENCE ON GROUND PENETRATING RADAR We are looking forward to meeting you in Switzerland. Please submit your papers before December 31, 2017. > www.gpr2018.hsr.ch 64 FIRST BREAK I VOLUME 35 I NOVEMBER 2017