f Optimization of Anisotropic Velocity Fields

Standard inversion techniques, such as least-squares Gaussian methods, tend to smooth the resulting images compared to the actual geological media. Here we report a novel use of the null space, defined by singular value decomposition (SVD), for resolving share discontinuities in the anisotropic vetocity field, and for allowing more realistic vetocity models. Since any linear combination of the null space model eigenvectors will have no effect on the modelled traveltimes, we search over the magnitudes of these eigenvectors for an optimum solution. By restricting our search to the null space, we achieve a more efficient optimization route than searching the full model spare.