79th EAGE Conference & Exhibition 2017 Paris, France, 12-15 June 2017

Risk reduction on the Ivory prospect via geologically constrained non-parametric inversion and Bayesian uncertainty estimation on a fault-bounded reservoir James Raffle1, Thomas Earnshaw1; Juergen Fruehn1; Stuart Greenwood1; Jeet Singh1; Claudia Hagen1, Ian Jones1, Rodrigo Felicio1, Doug Sassen1, Zhijiang Luo1, Svein Idar Forsund2, Mark Ackers2, Lars Aamodt2, 1: ION; 2: Centrica Norway

79th EAGE Conference & Exhibition 2017 Paris, France, 12-15 June 2017

Introduction In 2014 Centrica Energy operated the Ivory Exploration well (6707/10-3 S) in 1420m of water in the Norwegian Sea within the PL528B license. Gas was found in turbidite sandstones of the uppermost Kvitnos formation (informal ‘Delfin Member’). The Ivory structure is a fault bounded, three-way dip closure situated on the Nyk High. The well is located about 20km north east of the Aasta Hansteen gas development, which is expected to begin producing in 2018. The Nyk High is located in the north-eastern part of the Vøring Basin, which has been tectonically active in several phases, from Carboniferous to Late Pliocene time, with the main tectonic phases in Late Paleozoic, Late Mid-Jurassic-Early Creataceous and Late Cretaceous-Early Tertiary times. The tectonics of the Late Cretaceous and Tertiary periods were controlled by the relative movements along plate boundaries, with the last rift phase ending with continental breakup at ~54Ma followed by seafloor spreading (Brekke, 2000). During the last phase of intra-continental rifting and separation, uplift, erosion and increased clastic input in the Vøring Basin occurred. The present configuration of the Nyk High mostly dates from the Late Cretaceous to Earliest Tertiary and probably involved both extension and compressional/transpressional reactivation (Blystad et al., 1995). Prior to the formation of synclines and highs the area probably constituted a single broad Early Cretaceous Basin where deposition of Kvitnos and Nise formation turbiditic sandstones took place (Brekke, 2000). The main challenge in mapping the extent of the Ivory discovery has been seismic imaging at the crest of structures bound by major faults (e.g. fault shadow effects), together with depth conversion uncertainty and a poor well to seismic tie. To address the uncertainties associated with these issues, a new Pre-Stack Depth Migration (preSDM) was run using a dataset re-processed with the latest demultiple and deghosting technology. From this final migrated volume, structural uncertainty was then estimated using a Bayesian statistical analysis of the tomographic resolution matrices in conjunction with prior uncertainty estimates. Input Data Quality and Data Processing The seismic data covered an area of 1000sq.km, and were acquired in 2010 using a conventional towed source and streamer configuration which led to a good initial data quality. Pre-migration processing utilised several applications of both random and linear noise attenuation in preparation for deghosting with particular attention focused on low frequency noise, 0-6 Hz. The deep water environment allowed for a cascaded 3D and 2D SRME demultiple approach to target free surface multiples which dominated the target depth and deeper section. Deghosting was applied to suppress side-lobes and boost low frequency content within the data along with increasing the useable bandwidth from 5-95 Hz to 2.5-110 Hz (O’Driscoll et al., 2013, Zhou et al., 2012). Figure 1 compares a stack before and after de-ghosting, whilst figure 2 compares spectra of the corresponding migrated data, and also after application of attenuation (Q) compensation. Vintage processing showed that interbed multiples were prevalent throughout the data and caused difficulty at the interpretation stage. Therefore, IME was applied in a single iteration and was successful at removing interbed multiples generated by the near top Palaeocene horizon. 4D regularisation was also utilised as it provided beneficial noise reduction and near trace reconstruction over conventional 3D regularisation. Velocity model building

Six tomographic model building updates using Kirchhoff and Beam TTI migration were used in the project (Jones 2010, 2015). A non-parametric generalised move-out picker (GMO) is used to resolve fine velocity details in the subsurface. The GMO picker tracks wavelets above a user-defined semblance value (Fruehn et al., 2014, Luo et al., 2014) to feed into the tomographic update of velocity and epsilon. The main challenge was to adequately pre-condition the data to prevent the GMO picker from picking unwanted noise and residual multiple as well as preventing it from picking across Class 2P AVO anomalies that can cause serious inversion problems. A total of 200 internal tomographic iterations were run with a cell size of 150x150x25m to capture the fine velocity details in the region

79th EAGE Conference & Exhibition 2017 Paris, France, 12-15 June 2017

above the Top Nise Fm surface and a cell size of 250x250x50m was used for the deeper section. A tomographic update, constrained by depth-of-burial guided ties to the well sonic, resulted in good imaging of the strata leading up to the main fault (Figure 3), and significantly improved well-tie. Migration algorithm tests showed that Beam migration delivered a much clearer picture of the deep structure, as it was less affected by cross-cutting noise (Figure 4). Figures 5 and 6 compare depth slices and cross line vertical sections through the migrated volumes from vintage processing with the new processing, which show fine geological structural detail with enhanced continuity and better fault structure resolution.

Figure 1: Inline stack section before (left) and after (right) deghosting. Note the reduction in side lobes has resulted in a sharper wavelet and enhanced the low frequency content of the data. This resulted in improved fault definition on the final migrated data.

Figure 2: Amplitude spectra of corresponding migrated data before and after de-ghosting and application of amplitude Q-compensation

Figure 3. Left: Initial tomographic updated velocity. Right: result after constraining the velocities to match the well sonic and to follow the stratigraphic trends.

1km

79th EAGE Conference & Exhibition 2017 Paris, France, 12-15 June 2017

Figure 4. Kirchhoff migration (left) and beam migration (right) images. The beam image (right) has less noise contamination in the deep section, resulting in a more plausible geological structure. Beam migration was used for velocity model building in the last two iterations.

Figure 5. Vintage PreSDM depth slice (left) shows less continuity, poor resolution, and limited structural information. The recent processing (right) shows fine detailed geological structure of the bounding faults and better seismic resolution.

Figure 6. Vintage preSDM stack (left) shows less continuity and poor structural resolution. The re-processed preSDM stack (right) shows more fine detailed geological structure and enhanced resolution. Both images are Kirchhoff migrations.

1km

1k

m

1km

79th EAGE Conference & Exhibition 2017 Paris, France, 12-15 June 2017

Uncertainty analysis The final stage of the project involved uncertainty analysis using Bayesian estimation based on eigenvector decomposition of the tomographic resolution matrices (Osypov et al. 2013). Figure 7 shows the corridor of 200 model realizations of the main fault plane, with an instance of the standard deviations of a point termination of the Kvitnos horizon against the fault.

Figure 7. Left: image showing fault plane and uncertainty corridors based on 200 model realizations. Red: fault average position, Green: 2*Sigma confidence interval, Blue: outer bounds. Positioning uncertainty of a single point for the 200 model realizations Conclusions Improved detailed pre-processing imaging, in conjunction with analysis of structural uncertainty in the final preSDM image, has helped to de-risk future development of the Ivory gas discovery. The final preSDM has given the asset team a higher resolution and clearer seismic image, as well as proper depth calibration. The fault shadow effects have been reduced, and no sag along the bounding fault can be observed in the Ivory discovery on the reprocessed data. Improved imaging of both small and large scale faults has been achieved through an increased understanding of the velocity field and broad-band processing. The Ivory well shows an excellent well to seismic tie and the container can now be mapped with higher confidence. The area outside the Ivory structure, also covered with reprocessed data, shows improved imaging at both shallow and deeper levels, potentially leading to interesting exploration opportunities in the future. Acknowledgements We are grateful to Centrica, PL528 partners and ION for permission to publish these results.

References Fruehn, J., K., Greenwood, S., Valler, V., and Sekulic, D., [2014] Resolving small-scale near-seabed velocity

anomalies using non-parametric autopicking and hybrid tomography, CSEG Recorder, 39, no.10. Jones, I.F., [2015], Estimating subsurface parameter fields for seismic migration: velocity model building, in:

Encyclopaedia of Exploration Geophysics. SEG, pp. U1-1-U1-24. Editors: Vladimir Grechka and Kees Wapenaar. http://library.seg.org/doi/pdf/10.1190/1.9781560803027.entry3

Jones, I.F., [2010], An introduction to velocity model building, EAGE, ISBN 978-90-73781-84-9, 296 pages Luo Z., Brittan J., Fan D., Mecham B., Farmer P. and Martin, G., [2014]. Imaging complexity in the earth —

Case studies with optimized ray tomography. The Leading Edge, 33, no.9, 1016-1022. O’Driscoll, R., D. King, A. Tatarata, and Y. Montico, [2013], Broad-bandwidth data processing of conventional

marine streamer data: An offshore West Africa field case study: Presented at the 83rd Annual International Meeting, SEG.

Osypov, K., Yang, Y., Fournier, A., Ivanova, N., Bachrach, R., Yarman, C.E., You, Y., Nichols, D., Woodward, M., [2013] Model-uncertainty quantification in seismic tomography: method and Applications. Geophysical Prospecting, 61, 1114–1134

Rutherford, R. and Robert, H., [1989], Amplitude-versus-offset variations in gas sands, GEOPHYSICS. 54 no. 6, 680-688.

Zhou, Z. Z., M. Cvetkovic, B. Xu, and P. Fontana, [2012], Analysis of a broadband processing technology applicable to conventional streamer data : First Break, 30, no. 10, 77–82.