f A Priori Model Estimation for FWI from Constrained Kinematic Inverse Problem

Exploration inverse methods such as Monte Carlo could provide the general solution for the fullwaveform inversion (FWI) inverse problem allowing incorporating complex a priori information and data with arbitrary noise distributions. Such an approach applied directly to the FWI problem is computationally not tractable. The alternative is to solve the highly non-linear part of the problem with a fast forward problem such as traveltime with a constrained random exploration. The obtained a priori models and uncertainties help greatly in solving the FWI. This approach was successfully applied to an offset vertical seismic profile (OVSP) case.