First Break tomography with TV regularization and structural constraints

Common way to counteract the well-known ill-posedness of travel time tomography is Tikhonov regularization, which allows to exploit prior information coming from geological assumptions or interpretation of intermediate results. However, in order to preserve edges and discontinuities, some further techniques have been proposed: among these, Total Variation (TV) has shown its effectiveness even against the sharpest models. This work aims at investigating the combined use of TV and apriori structural information, applied to a First Break tomography problem. More specifically, from migrated images we can obtain priors about the main dip fields and the discontinuity location. We propose to exploit this information in the regularization by means of the structure oriented filter, comparing the performances of the methodology with common quadratic regularization. Results confirm our strategy as a viable and accurate solution for the velocity model building problem.