Obtaining Geologically Conformable Tomographic Models Through Anisotropic-Diffusion Preconditioning.

Tomography driven model building workflow is currently pushing for increasingly higher resolution in the subsurface model. One key challenge in doing this is significant under-illumination in the model space due to fine inversion grids used. The poor illumination boosts data errors and introduces inversion artifacts into the solution, thus seriously degrading the resolution and accuracy of inverted models. To mitigate this issue, we propose to incorporate anisotropic diffusion smoothing operators into the conjugate gradient algorithm to precondition tomography. Synthetic and real examples demonstrate that this preconditioning can produce more detailed and structurally conformable velocity models than the conventional regularization-based tomography. Thus, we believe that the new approach is a valuable addition to the imaging workflow to derive high resolution velocity models for seismic imaging.