G021 Subsalt Velocity Analysis Using One-Way Wave Equation Based Poststack Modeling

Transcription

1 G021 Subsalt Velocity Analysis Using One-Way Wave Equation Based Poststack Modeling B. Wang* (CGG Americas Inc.), F. Qin (CGG Americas Inc.), F. Audebert (CGG Americas Inc.) & V. Dirks (CGG Americas Inc.) SUMMARY Subsalt velocity analysis using prestack wave equation migration scans through perturbed velocity model is an accurate but expensive approach. To reduce turnaround time and computation cost, we investigated an alternative approach for subsalt velocity analysis. In this approach, we separate the prestack focusing step from the poststack modeling step. We first use the current best velocity model to perform one prestack migration to produce a subsalt image. We then use the subsalt image as a reflectivity model to perform upgoing one-way exploding reflector modeling to produce demigrated zero-offset seismic data. Using this demigrated data as input, we perform a scan of poststack wave equation migrations through perturbed subsalt velocity model. We demonstrate the necessity of demigrating only to the base of salt in order to avoid significant image degradation and we show the feasibility of this methodology using the 2D Sigsbee data set.

2 Introduction Migration scan techniques have been developed within the framework of full prestack migration for some time (Yilmaz and Chambers, 1984; Audebert et al., 1996). Although the most recent attempts of using the scan technique for subsalt velocity updating (Wang et al., 2004) are promising, the cost of generating migration scans is very high, as it requires multiple prestack migration runs. To address the cost issue, in this paper, we present a low-cost alternative to generate migration scans. We separate the prestack focusing step from the poststack modeling step. We first use the current best velocity model to perform one prestack migration run and thus produce a stacked subsalt image. We then use the subsalt image as a reflectivity model to perform upgoing one-way exploding reflector modeling to produce demigrated zero-offset seismic data. This is followed by a set of poststack wave equation migrations corresponding to variations around the best velocity model and using the zero-offset seismic as the input. Migration velocity analysis explores sensitivity of migration image quality on velocity errors. Velocity errors manifest themselves in two aspects: 1) focusing and defocusing; 2) structure integrity of the image. For prestack migration, a defocused image reveals itself as significant residual moveout in Common Image Point gathers; therefore prestack focusing amounts to collapsing the reflection energy from non-zero offsets to zero offset. For poststack migration, focusing takes place if it collapses diffraction energy to diffractors across CDPs and Lines. The demigration and remigration approach we present here is more applicable for deep subsalt areas with subsalt folding structures, such as Alaminous Canyon, Gulf of Mexico. In those subsalt areas, the available range of reflection angles is very limited, residual moveout effects are not significant issues. The poststack migration scan provides information such as whether the structure (anticline or syncline) is under or over migrated and whether the structure makes good geological sense. Problems of directly using post stack migration scan for subsalt velocity analysis We postulate that wave equation poststack migration may not be directly applied for subsalt migration perturbation scans. To illustrate our point, we use the 2D Sigsbee data set as an example throughout this paper. Figure 1 shows the true zero-offset seismic data recorded at the surface. Figure 3 is the poststack wave equation migration of the zero-offset data of Figure 1. Although salt boundary and pure sediment reflection events are well imaged, the subsalt image is poor, and severely affected by coherent noise and migration artifacts; therefore it could not be used for subsalt velocity analysis. Why do we have such a poor poststack migration image in subsalt area, even though the input data is a synthetic data without any random noise? First, the forward modeling operator and the migration operator are not consistent. The true zero-offset data (Figure 1) are generated by twoway acoustic modeling, while poststack migration is based on the simple one-way exploding reflector operator. Thus, any complex events modeled by the two-way acoustic modeling operator but not modeled by the one-way exploding reflector operator become coherent noise after poststack migration. Second, there are strong migration artifacts, such as the typical swing noise due to amplitude imbalance. Third, as illustrated later in this paper, due to the geometric complexity of salt body and the high velocity contrast between salt body and surrounding sediment, severe illumination problems will affect a single fold dataset. Wave equation demigration/remigration to and from the free surface Naturally, we always use prestack migration to produce a good enough subsalt image. Figure 4 is the image after shot profile wave equation migration (note that to speed up this feasibility study, we used a simple 2D prestack migration based on Fourier Finite Difference scheme). The prestack migration image gives a much better subsalt picture, since it is a stack of all the individual partial images. Just like the poststack migration image, each partial image contains coherent noise and severe illumination artifacts. However, the migration artifacts and coherent

3 noise are not identical across different partial images, so they will tend to average out through the process of stacking, and only the true subsalt reflection events are constructively stacked. Prestack migration produces a far better reflectivity image that will serve as a much better starting point for demigration. The wave equation demigration of the image shown in Figure 4, followed by poststack wave equation migration, produces the image shown in Figure 5. Compared to the original poststack image in Figure 3, the subsalt image is much better. This is mainly due to the fact that both forward modeling (demigration) and poststack migration are now based on the same one-way exploding reflector operator. However, compared to the prestack image in Figure 4, the demigration/remigration loop has still significantly degraded the image quality. To better understand the process of de- and remigration we created an illumination plot (Figure 2). An exploding reflector is simulated at 8 km depth, and the illumination plot is the square root of the local propagation energy of the upgoing wavefield. When the dominantly vertically upgoing energy reaches the base of salt, because of the high velocity contrast between subsalt sediment and salt, most of the energy is critically reflected at the steep base of salt. Subsequentially the surface receivers within the limited aperture will not record this energy; hence shadow zones will be generated after remigration. Comparing Figures 2 and 5, we can identify these areas below the steeply dipping base of salt as the areas with most degraded image quality. Wave equation demigration/remigration to and from the base of salt In order to avoid the shadowing effects described above, we can choose the base of salt as the redatuming surface. Instead of performing demigration all the way up to the surface, we output the demigrated zero-offset seismic data at the base of salt horizon. Poststack migrating such zerooffset section results in the image shown in Figure 6. Comparing these new results (Figure 6) with those achieved with the poststack remigration from the surface (Figure 5) shows that the image quality of the new results is comparable to the initial prestack image in Figure 4. These results motivate the use of poststack migration in the quiet subsalt area: the post stack migration of the demigrated data gives an output equivalent to the one from the full prestack migration. This reminds the classic equivalence of NMO-Stack + poststack migration, in simple velocity medium, with prestack migration, but with the important difference here that the initial stack has been in fact produced by full prestack migration followed by poststack demigration. Subsalt scans using post stack migration below the base of salt Subsalt velocity analysis starts when the velocity model building is finished down to the base of salt. In our proposed methodology, the prestack migration and zero-offset demigration loop serves to produce the best possible stack section, taking advantage of prestack migration using the current best velocity model to unravel propagation effects in the complex overburden effects. It is well known that in many areas, such as the Gulf of Mexico, high velocity structures create lens effects that reduce the range of reflection angles. For a limited range of reflection angles, prestack residual moveout focusing and defocusing effects are less important. Optimal imaging therefore has to be based on structural image integrity. Structures such as a syncline or an anticline, faults, horizon terminations for example will provide the required velocity information. To illustrate this point, we use the Sigsbee velocity model, but replace the subsalt reflectivity model with a simple folded structure. The reflectivity model is first demigrated to the base of salt, then remigrated with three different velocity models, 100%, 85%, and 115% of the true velocity field (Figures 7A, 7B, and 7C). One can see significant over and under migration effects on the subsalt structure, that very well resemble the image features observed on real datasets. This test demonstrates the structural sensitivity of imaging to changes in the velocity field even though the range of reflection angles is very limited.

4 Conclusions We investigated an alternative methodology for subsalt velocity analysis, using poststack wave equation migration rather than its prestack version, to perform subsalt migration scans. In this methodology, the prestack migration followed by zero-offset demigration is used to produce the best-stacked data, a process that also attenuates coherent noise and migration artifacts. We have demonstrated that poststack wave equation migration cannot replace prestack migration for the propagation throughout the complex salt body. Be it at the migration or demigration stage, the propagation of a single fold dataset through complex salt bodies produces artifacts through evanescence and irregularities of illumination. As a consequence the zero-offset demigration and remigration have to be performed to and from the base of salt datum, to avoid shadowing effects. As an alternative to the full-blown subsalt migration scans using prestack migration, this methodology is a magnitude faster and cheaper. Acknowledgements We would like to thank the SMAART consortium for being able to use the Sigsbee2A data for this study. We thank Robert Soubaras, Chris Milne, Dan Wheaton and Stefan Kaculini for helpful discussions. Figure 1. The true zero-offset seismic data recorded on the surface Figure 2. The illumination plot simulating upgoing wave from exploding reflector at 8 km depth. The white color indicates weak

5 References Audebert, F., Diet, J.P., Zhang, X., 1996, CRP-scans from 3D pre-stack depth migration: a powerful combination of CRP-gathers and velocity scan: 66 th SEG Expanded Abstracts: Wang, B., Dirks, V., Guillaume, P., Audebert, F., Hou, A., and Epili, D., 2004, 3D sub-salt tomography based on wave equation migration perturbation scans, 74 th Ann. Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, Yilmaz, O., and Chambers, R., 1984, Migration velocity analysis by wave field extrapolation: Geophysics, 49, Figure 3. The post stack wave equation migration result using the true zero-offset data. Figure 4. The shot profile prestack wave equation migration image. Figure 5. The post stack remigration image using the demigrated seismic data at the surface. Figure 6. The post stack remigration image using the demigrated seismic data at the base of salt datum. Figure 7. Subsalt post stack migration scan using as input the demigrated seismic data at the base of salt and migrated with three different velocity models: (A) 100%, (B) 85%, and (C) 115% of the true velocity model.