Migration velocity analysis for anisotropic models

Conclusion and discussion

We have presented a new methodology for performing image-domain migration velocity analysis in anisotropic media. Our method is a natural extension of isotropic MVA theory and retains the same properties as isotropic MVA. We demostrate our method on a 2-D synthetic data set. After inversion, we obtain better-focused subsurface-offset images and better-defined depths. By including the geological information and the wider-offset data, we should be able to eliminate the model error at depth.

Experience shows that the DSO operator has a layer-stripping effect during the iterations. One cause of this effect is the unbalanced amplitude for the reflectors in depth. Therefore, an illumination-corrected image is preferred to compensate for this effect. On the other hand, a residual-moveout-based objective function (Sava, 2004; Sava and Biondi, 2004b; Almomin, 2011; Sava and Biondi, 2004a; Zhang and Biondi, 2011) could avoid the problem.

Compared with ray-based image-space model-building methods, our wavefield-based image-space method is computationally more intensive. However, the wavefield method better approximates wave propagation in complex areas. We can also utilize the phase-encoded target-oriented image-space wavefield tomography (Guerra and Biondi, 2010; Guerra et al., 2009) technique to reduce the computational cost.

Finally, by introducing another parameter $\eta$ into the MVA inversion, we now have a larger model space and hence a larger null space with respect to the same data. Therefore, the surface reflection seismic data is inadequate for resolving a unique earth model. Other information, such as borehole measurement, geological interpretation (Bakulin et al., 2010), or rock-physics prior knowledge (Li et al., 2011b,a), is necessary to obtain a consistent, unique and reliable earth model.

Migration velocity analysis for anisotropic models