3-D tomographic updating with automatic volume-based picking Dimitri Bevc*, Moritz Fliedner, Joel VanderKwaak, 3DGeo Development Inc. Summary Whether refining seismic images to evaluate opportunities in mature areas and exploit the maximum resource, or exploring in frontier areas, determining an accurate velocity model within the turnaround time constraints of reservoir management and exploration timeframes is critical. Speed, robustness, and accuracy are equally important. Seismic imaging has made great strides in recent years with the advent of so-called wave-equation migration imaging methods. Given the correct acoustic propagation velocity for seismic waves in the Earth s subsurface, these waveequation methods yield the highest resolution and most accurate images of the earth. However, the process of determining the correct acoustic propagation velocity can be an elusive, time consuming, and costly procedure. We describe an approach both to shorten the process and to make the process less biased and more accurate. The process is shortened by automating the labor-intensive portion of the workflow, and made less biased and more robust and accurate by using much more data than is commonly used in manual picking approaches. Introduction We describe the implementation of an automatic method of signal detection that eliminates the need for manual reflector picking by scanning the seismic data volume with prediction-error filters and automatically selecting back projection points based on dip coherency and semblance strength. This approach can save months of human time on a typical 3-D seismic imaging project and, thereby, shortening seismic imaging project turnaround time while exploiting the full redundancy of the recorded data. The automation also reduces human bias and manual picking error, while retaining the option to control quality and steer the solution. Tomography For migration velocity analysis, residual moveout in the 3D angle-domain common-image gathers (ADCIGs) is parameterized as a residual slowness squared (or residual velocity), either by fitting a reflection event in a gather by semblance analysis, or by picking the relative depth error between normal and oblique incidence traces and applying the appropriate residual moveout formula for ADCIGs, which depends on the relative velocity error, depth, opening (reflection) angle, azimuth (angle of ray plane with acquisition direction), and local geological (reflector) dip (Biondi and Tisserant, 2004). The velocity model is then updated by raytracing tomography applying a global (Ji, 1995) or a local (Stork, 1992) approach to calculating the residual traveltime. The main difference between the two approaches is the use of the normal ray (zero offset) reflection to determine traveltime errors for the tomographic inversion. In the global approach, we compare the actual (raytraced) total normal ray traveltime with the expected, calculated from the residual slowness at the reflection point, i.e. a single parameter (residual slowness or velocity) describes the moveout behavior of the prestack image at each analysis point. The main advantage is the possibility to perform a less costly inversion based on normal rays alone (rather than tracing and inverting ray fans); applying one, consistent correction at each point makes the inversion also more robust. The local approach is based on converting the depth error between normal and oblique incidence rays in the prestack image into a relative time error based on the local velocity. Since each trace in the prestack image is analyzed separately, this approach has potentially a higher resolution (more than one independent parameter per backprojection point), but may for the same reason be less robust. Inversion of the normal rays alone is not possible with this approach (the normal ray residual is by definition zero). Backprojection points for the tomographic inversion can be chosen in two ways: (1) picking geological horizons, or (2) selecting individual reflection points based on local dip (i.e. reflector) coherence and semblance strength (i.e. reflection event coherence). Automatic picking of reflectors and reflection moveout is performed based on a horizon flattening algorithm (Lomask et al., 2005). The first approach has the advantage of producing geologically reasonable velocity models without further steering filters applied to the tomographic inversion, but it requires the manual input of an interpreter. To automate this process, an automatic horizon picker is required. The second approach requires an automated process to select backprojection points (Fliedner et al, 2003; Clapp, 2001; Clapp et al, 1998). Both approaches start with the calculation of a dip field from the stacked migrated image. Several methods have been tested and the approach that gives the best results (Fomel, 2000) selected: it achieves a sharp delineation of reflectors, as well as a smooth dip field (by applying increasing smoothing filters). Automatic Picking Procedure Selecting backprojection points independent of manually picked horizons involves calculating the best single dip in a window and the coherency of the dip by iterative 3330
Automatic Tomography application of plane-wave destruction filters. Points that satisfy specified levels of dip coherence, amplitude, semblance strength, and distance from other points and the edges of the image are selected as backprojection points. This method allows for an even distribution of backprojection points in the absence of strong geological boundaries (reflectors) that define the velocity model. We implement a plane wave destructor prediction error filter (Claerbout, 1992) to run on 4-D and 5-D migrated data volumes from 3-D prestack data. Rather than using picked reflectors as the basis for back projection locations, points are selected according to reliability factors. This technique first calculates dip and coherency of the migrated image at each image location, providing an initial dip and coherence estimate in a region around each model location. From this result of the previous step, we calculate the best single dip within the region, and the coherence of that dip. This is then used as an initial dip estimate for the nonlinear, space varying dip estimation procedure such as the one described in Fomel (2000). Data Example To illustrate our approach, we present the back-projection point selection process as follows with an accompanying set of data images from the preliminary migration of a 2-D seismic data set (Figure 1): 1. Calculate dip and dip coherency at every model location (Figure 2). 2. Iteratively select a set of preliminary backprojection points that meet specified threshold criteria taking into account amplitude, coherency (Figure 3), spacing, etc. Amplitude and coherence criteria are relaxed at each iteration to select the best points in each region. 3. Calculate semblance (Figure 4). 4. Filter the preliminary point set, retaining points with maximum semblance above a specified threshold (Figure 5). Output from this is the coordinates, dips, coherency, and semblance of each remaining back-projection point. 5. The final point set is used as input into the tomographic inversion of the set of spatiallydisconnected points (Figure 6). This automatic migration velocity analysis method has been demonstrated on a wide ranging suite of 2-D and 3-D data sets: simple synthetics with analytically known kinematics, complex synthetic, and real data sets in 2D and 3D. Results from these tests will be presented. Conclusions We present a method to eliminate, or at least significantly reduce, the need for manual horizon picking. The method calculates a dip field and coherency from a migrated image by using a plane-wave estimator. The dip estimate is then refined and back projection points are automatically selected based on dip coherency and semblance strength. The selected points are used in a regularized migration reflection tomography inversion. The tomographic operator simultaneously accounts for velocity correction and reflector movement. The model can be preconditioned with a steering filter, which tends to create velocity variations consistent with geologic dip. The resulting tomographic update produces robust images with significantly reduced turnaround time for the entire velocity model building process. References Biondi, B. and T. Tisserant, 2004, 3D angle-domain common-image gathers for migration velocity analysis, Geophysical Prospecting, 575 591. Claerbout, J. F., 1992, Earth Soundings Analysis: Processing versus Inversion: Blackwell Scientific Publications. Clapp, R. G., 2001, Geologically constrained migration velocity analysis: Ph.D. thesis, Stanford University. Clapp, R.G., B.L. Biondo, S.B. Fomel, and J.F. Claerbout, 1998, Regularizing velocity estimation using geologic dip information: 68 th Ann. Internat. Mtg. Soc. of Expl. Geophys., 1851-1854. Fliedner, M., Bevc D., and Clapp R., 2003, Depth imaging velocity estimation by layer-stripping Dix update and dip-constrained tomography in a compressional tectonic regime73rd Ann. Internat. Mtg: Soc. of Expl. Geophys., Dallas. Fomel, S., 2000, Applications of plane-wave destructor filters: SEP-105, 1-26. Ji, J., 1995, Sequential seismic inversion using plane-wave synthesis, Ph.D. thesis, Stanford. Lomask, J., A. Guitton, S. Fomel, and J. Claerbout, 2005, Update on flattening without picking, SEP Report 120, 137 158. Stork, C., 1992, Reflection tomography in the postmigrated domain, Geophysics, 680 692. 3331
Figure 1. Starting model stacked image. Figure 2. Smoothed Dip Field calculated from starting seismic image of Figure 1. Figure 3. Coherency field used for point selection and tomographic inversion. 3332
Automatic Tomography Figure 4. Peak semblance field - prestack information for tomographic inversion. Figure 5. Automatically selected backprojection points. Different coherency and semblance criteria can be used to generate more or less points. Figure 6. Raytraced trajectories for backprojection in tomographic inversion. 3333
EDITED REFERENCES Note: This reference list is a copy-edited version of the reference list submitted by the author. Reference lists for the 2006 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 Biondi, B., and T. Tisserant, 2004, 3D angle-domain common-image gathers for migration velocity analysis: Geophysical Prospecting, 52, 575 591. Claerbout, J. F., 1992, Earth Soundings Analysis: Processing versus Inversion: Blackwell Scientific Publishing Company, Inc. Clapp, R. G., 2001, Geologically constrained migration velocity analysis: Ph.D. thesis, Stanford University. Clapp, R.G., B. L. Biondo, S. B. Fomel, and J. F. Claerbout, 1998, Regularizing velocity estimation using geologic dip information: 68th Annual International Meeting, SEG, Expanded Abstracts, 1851 1854. Fliedner, M., D. Bevc, and R. G. Clapp, 2003, Depth imaging velocity estimation by layer-stripping Dix update and dip-constrained tomography in a compressional tectonic regime: 73rd Annual International Meeting, SEG, Expanded Abstracts, 2191 2194. Fomel, S., 2000, Applications of plane-wave destructor filters: SEP, report 105, 1-26. Ji, J., 1995, Sequential seismic inversion using plane-wave synthesis: Ph.D. thesis, Stanford University. Lomask, J., A. Guitton, S. Fomel, and J. Claerbout, 2005, Update on flattening without picking, SEP, report 120, 137 158. Stork, C., 1992, Reflection tomography in the postmigrated domain: Geophysics, 57, 680 692. 3334