A tunneling approach to constrained full waveform inversion

Scott Keating, Kristopher A. H. Innanen

Prior information can be a powerful tool in seismic inversion that can substantially improve on the results that seismic data alone can provide. Knowledge about clustering of rock physics properties may be especially significant, but this type of information is problematic in full waveform inversion due to the local optimization methods which are typically used. Here, we propose a regularization tunneling strategy for full waveform inversion, in which global regularization information is partially accounted for. By introducing the potential for elements of the subsurface model to tunnel between clusters, this approach is able to overcome obstacles associated with local minima in regularization terms from a priori data. We test a tunneling inversion approach on a simple synthetic problem, and find that when information about rock physics clustering is available, the proposed technique has the potential to better use this information than a conventional inversion strategy.