2011 SEG SEG San Antonio 2011 Annual Meeting 3938

Transcription

1 Depth imaging Coil data: Multi azimuthal tomography earth model building and depth imaging the full azimuth Tulip coil project Michele Buia 1, Peter Brown 2, Bakhrudin Mansyur 2, Michelle Tham 3, Suyang Chen 3, Swee Leng Ng 3, Olga Zdraveva 3 and Martin Bayly 3 1 Eni E&P, 2 Eni Indonesia, 3 WesternGeco Summary The Tulip 3D is a full azimuth single vessel Coil shooting survey in Indonesia. Coil geometry acquisition records a dataset with a wide range of source to detector azimuths. Amongst the numerous benefits of extended azimuth sampling methods is the advantage of illumination diversity. This has a desirable impact with tomographic velocity model building for depth imaging. The broad range of source to detector azimuth information allows an additional constraint on the travel time tomography by providing a network of intersecting rays not available with narrow azimuth geometries. Introduction The Tulip survey is located over the continental slope east of Kalimantan island, offshore Indonesia in a water depth of m. The overburden geology in the survey area is extremely complex due to rugose water bottom, very bright Bottom Simulating Reflectors (BSR), very heavy faulting, sub-surface channels, and shallow overburden patchy gas pockets throughout the survey area. The target reflectors have very low acoustic impedance contrast and are heavily contaminated by surface and diffracted multiple energies. The consequence of these complexities results in overall poor seismic response, very low amplitude or near invisible target reflections, very low signal-to-noise ratio (S/N). With conventional narrow azimuth marine acquisition, this results in poor imaging and poor illumination of the reservoirs. A different approach is required not only in terms of acquisition method, but also a tailored processing workflow to improve the imaging and illumination of the reservoirs. illumination improvement, increase in signal-to-noise and event continuity, reducing multiples and enhancing amplitude fidelity, but also improving the tomographic velocity model updating and imaging stages. However, velocity model building with full multi-azimuth data not only requires the use of a tomographic solver method that can fully comprehend the ray-path direction between the source and receiver, but also correctly account for the source to receiver azimuth contribution of traces in common image point gathers (CIPs) when performing residual moveout (RMO) analysis (Woodward et al. 2008). This paper will discuss the multi azimuthal tomography velocity model building workflow and depth imaging of the Tulip survey. In addition a comparison will be made of individual azimuth sectored migration volumes between using a straight sum compared relative to optimum weighted summation. Multi Azimuthal Anisotropic Tomographic Velocity Model Building and Depth Imaging Workflow The Coil configuration and geometry used in the survey recorded a very high density (offset, fold, and azimuth) rich azimuth dataset. This additional azimuth information allows for a improved constraint for the full multi azimuth tomography to produce a detailed velocity model with higher degree of accuracy. Although the Coil recording produces a high density of irregular fold, offset and azimuth distribution, after careful spatial regularization in earlier processing (Buia et al, 2010), we were able to sector the full azimuth dataset into 29 offset groups, and three 60 degree wide azimuth groups for migration velocity analysis and azimuthal tomography velocity model building. Figure 1 shows the multi azimuth workflow. After an extensive survey design and feasibility study effort, Eni had chosen a Coil shooting geometry (French, Cole, 1984; Tozzi et al., 2009) to acquire an extended azimuth solution to more densely and diversely illuminate the reservoir for an appraisal campaign. During the data processing phase of the project, opportunities to best utilize the available azimuth sampling were considered, multi azimuthal earth model building proved to be one technology which can be successfully applied on a rich azimuth survey such as this Coil project. It is well known that the benefits of an extended azimuth (multi, wide, rich, full, etc) datasets include not only Figure 1. Multi azimuth tomographic workflow. SEG San Antonio 2011 Annual Meeting 3938

2 All three azimuth sectors were iteratively prestack depth migrated for common image point (CIP) gather production for the model building loops. These gathers were produced on a 50mx50m grid to provide a high resolution tomography. The gathers were individually analyzed to measure multi-valued residual moveout (RMO) for each azimuth sector. All three sets of RMO picks were provided to a single all azimuths Grid Tomography to update the model. The assumption being that velocity change seen on individual CIP locations at this early stage with initial earth models is most probable to be unresolved overburden heterogeneity rather than true azimuth anisotropy. The grid tomography used a damped least-squares solver at progressively finer scale lengths, with a preconditioningsmoothing operator being applied at each scale in order to improve the convergence of the solution. Extensive autopicking of gathers was used, and both smooth and complex moveout were used in the tomographic back-projection, depending on the scale length of the solution. This process was iterated until a satisfactory velocity model was achieved. Figure 2 shows the illustration of the Grid Tomography workflow. slow and some too fast) at the first iteration to a normal distribution around zero by the third iteration, for all azimuth sectors. This indicates the single earth model being produced is reducing the residual moveout for all azimuths; indicating that the bulk of the residual moveout seen at earlier iteration is due to previously undefined overburden heterogeneity (lateral velocity change) and not change of velocity with direction due to potential HTI anisotropy. As noted by Whitfield et al. (2009), overburden lateral velocity heterogeneity needs to be reduced first before indications of Azimuthal Anisotropy can be investigated. In this case, observation of azimuth and offset organized butterfly gather QC displays indicated that any azimuth velocity change appears to be minimal once the earth model was refined in this manner. Figure 4 shows the initial model and fifth iteration model overlaying onto the seismic data, this displays velocity details corresponding to the geology and reflection boundaries. Figure 3a. RMO Gamma maps and histograms QC of 1st iteration - window at target depth. Figure 2. Illustration of the Grid Tomography workflow. The initial velocity model for grid tomography was derived from edited time domain interval velocities with strong smoothing. The first two iterations were isotropic with tomography resolved features to a scale length of 5000m horizontally and 1000m vertically. Further iterations used generalized Tilted Transverse Isotropy (TTI) modeling approach (Zdraveva, Cogan 2011), with limited well control or other anisotropy information directly available, the Thomsen ε parameter was derived from a regional δ function and anellipticity (η) from the seismic data, the TTI tilt angles were assuming to be perpendicular to the seismic image reflectivity dip. The following three tomography iterations converged to resolving features of a scale length of 1700m in the horizontal direction and 200m in the vertical direction. Each model loop was checked with quality control (QC) plots of the RMO gamma attributes as histograms and maps. Figure 3a and 3b show the RMO gamma QC maps over a window containing the main target depth for 1st iteration and 3rd iteration. Notice how the histograms have converged from being bi-modal (some too Figure 3b. RMO Gamma maps and histograms QC of 3rd iteration - window at target depth. Figure 4. Initial (left) and after 3 iterations of TTI modeling (right). SEG San Antonio 2011 Annual Meeting 3939

3 After the three iterations of TTI multi azimuthal tomography, the single earth velocity model was considered to be sufficiently accurate to migrate the entire survey to meet the initial phase of the appraisal campaign timeline. The depth migration of the data using this velocity modeling technique resulted in imaging improvements at the target level and also at deeper depths. Figure 5 shows a comparison stack of the prestack time imaging (stretched to depth) compared to the improvement of depth imaging using the multi azimuthal velocity model building technique. the single available VSP derived interval velocity measurement (Figure 7). This confirmed the interval velocity gradient changes with depth computed by the tomography. Also of note were the slow velocities indicated below the seabed, these are likely to be due to free, patchy gas. This not only produces very slow P wave velocities but contributes to the strong absorption effects present in the dataset. Furthermore, although the tomography did not use interpreted horizons as constraints, the resulting model matches the geologic boundaries; for example a slow velocity zone from the model correlating to an overpressure zone encountered during the first exploration well. Figure 5. Comparison stacks of time imaging and depth imaging - the depth imaged data shows improved data continuity, visibility and sharpness of the key dipping events. Following the initial depth migration of the entire survey, a subset volume of data was extracted for a more detailed work program. Two additional iterations of TTI tomography were performed on this subset volume, resolving to a scale length of 1000m horizontally and 200m vertically. This further improved the earth model and the details can be seen on Figure 6. Figure 6. Detailed velocity model after two further iterations of TTI grid tomography updates. During the entire earth model building process, the developing model was constantly monitored against geological knowledge and expectations; in order to give a physical meaning to the velocity anomalies observed. A good match was observed between the velocity model and Figure 7. TTI azimuthal tomography model and well VSP Azimuth Sector Migrations and Optimum Weighted Stack Although the original project timeline only allowed for the initial migration of the data as an all azimuth migration, there was an opportunity later to perform azimuth sectored migrations with the coil dataset. By deferring the summation of the different azimuths from the migration step to a separate sum later enables analysis of event illumination in differing directions. The individual azimuth sector migrations show depth and space variant quality differences of stacking response, signal to noise level and data continuity depending on the azimuth sector. Different azimuth sectors illuminate different areas of the zone of interest with varying signal to noise ratios. Ideally, the optimal dataset will be a combination of the best stacking response and highest signal to noise data regions from each azimuth sector thereby producing an optimized final stack volume. To analyze the benefit of this approach, a comparison stack was produced by straight summing the three azimuth migrations, another stack was then produced using an optimum weighted stack technique. This method derives azimuth, depth and space variant weights based on comparative data quality so that the best data zones from SEG San Antonio 2011 Annual Meeting 3940

4 each azimuth sector are given more weight in the stacking process. As a consequence, maximum coherent/illuminated events are favoured on the resultant weighted stack output. Figure 8 and 9 show the comparison of the two methods and the advantage of the optimum weighted stack method is clearly visible. Conclusions The success and value of the multi azimuthal TTI anisotropic earth model building using grid tomography and following depth imaging workflow has been demonstrated. The final velocity model matches the VSP data very well and the resulting images in this challenging prospect are demonstrably better in terms of data continuity, visibility and sharpness of the dipping events compared to traditional methods. The azimuth sectored migrations and optimum weighted stack also showed another advantage of rich azimuthal diversity datasets. Acknowledgments The authors would like to thank Eni E&P Management, Eni Indonesia, MIGAS, BPMIGAS, and WesternGeco for permission to publish this paper. Figure 8. Azimuth sectored migrations. Figure 9. Straight stack vs. optimum weighted stack. SEG San Antonio 2011 Annual Meeting 3941

5 EDITED REFERENCES Note: This reference list is a copy-edited version of the reference list submitted by the author. Reference lists for the 2011 SEG Technical Program Expanded Abstracts have been copy edited so that references provided with the online metadata for each paper will achieve a high degree of linking to cited sources that appear on the Web. REFERENCES Buia, M., R. Vercesi, M. Tham, S. L. Ng, A. T. Waluyo, and S. Chen, 2010, 3D coil shooting on Tulip Field: Data processing review and final imaging results: 80th Annual International Meeting, SEG, Expanded Abstracts, Cole, R.A., and W. S. French, 1984, Three-dimensional marine seismic data acquisition using controlledstreamer feathering: 54th Annual International Meeting, SEG, Expanded Abstracts, French, W., 1984, Circular seismic acquisition system: U. S. Patent 4,486,863. Tozzi, E., M. Buia, and L. Mapelli, 2009, Circular shooting on Tulip Field Full azimuth streamer acquisition and illumination QC: 71st Annual International Conference and Exhibition, EAGE, Extended Abstracts, V010. Woodward, M., D. Nichols, O. Zdraveva, P. Whitfield, and T. Johns, 2008, A decade of tomography: Geophysics, 73, no. 5, VE5 VE11. Whitfield, P. J., R. T. Woods, T. Manning, and D. Baptiste, 2009, Multiazimuth QC for velocity model building A Nile Delta case study: 71st Annual International Conference and Exhibition, EAGE, Extended Abstracts, V015. Zdraveva, O., and M. Cogan, 2011, Building anisotropic models for large-scale TTI imaging in areas with limited well control: Kwanza Basin case study: 73rd Annual International Conference and Exhibition, EAGE, Extended Abstracts, C043. SEG San Antonio 2011 Annual Meeting 3942