f Cross-well tomographic inversion constrained by log data and a priori geological knowledge: a bayesian approach

The limited coverage inherently present in cross-well seismic surveys leads to a nonunique solution to the inverse problem. Besides, the nonlinearity resulting from the use of curved rays increases the dimension of the solution space. Therefore it is desirable and sometimes compulsory to stabilize the inversion through the use of a priori information. The goal is to enable a log-constrained tomographic inversion in areas where logs are available, and more generally to allow local velocity bounds as well as intercell velocity dependences to be fed into the inversion in order to get geologically meaningful roodels while stabilizing the inversion .