We presented a new MVA method in conjunction with wavefield continuation migration. It is based on the computation of anisotropic velocity spectra from the RMO transformation in anisotropic ADCIGs.
We studied the estimation problem by formulating it mathematically as an inverse problem and showed that care must be taken when estimating the different anisotropic parameters: a small range of the aperture angle leads to poor estimation of the horizontal velocity. More importantly, we showed that when large aperture data is available, trying to estimate both parameters using a conventional inversion scheme leads to poor estimates of the NMO velocity. To solve this issue, we proposed to modify the cost function of the inversion problem, which in practice is equivalent to adding some weights in the computation of semblance cubes.
We eventually computed anisotropic velocity spectra from synthetic anisotropic ADCIGs. We showed that although we used a first-order approximation of the expression of RMO in anisotropic ADCIGs, picking the anisotropic migration velocities in semblance cubes improves the overall anisotropic parameter estimation.
Our future work will consist in the development of the two following estimation techniques:
Estimation from anisotropic velocity spectra: The computation of anisotropic velocity spectra and the iterative estimation procedure must be studied in more detail to improve the convergence of the estimation procedure. The use of weights in the computation of the semblance cubes might be the most effective approach.
Estimation from the inversion of RMO: Measuring RMO (using an automatic technique such as the one presented by (6)) and solving the associated inverse problem is potentially a successful way of estimating anisotropic parameters. The estimation problem would be reduced to a conventional inverse problem, making the different constraints such as spatial continuity much easier to handle.