Total variation regularization strategies in full waveform inversion for improving robustness to noise, limited data and poor initializations

We propose an extended full waveform inversion formulation that includes convex constraints on the model. In particular, we show how to simultaneously constrain the total variation of the slowness squared while enforcing bound constraints to keep it within a physically realistic range. Synthetic experiments show that including total variation regularization can improve the recovery of a high velocity perturbation to a smooth background model, removing artifacts caused by noise and limited data. Total variation-like constraints can make the inversion results significantly more robust to a poor initial model, leading to reasonable results in some cases where unconstrained variants of the method completely fail. Numerical results are presented for portions of the SEG/EAGE salt model and the 2004 BP velocity benchmark. ***Disclaimer.*** *This technical report is ongoing work (and posted as is except for the addition of another author) of the late John "Ernie" Esser (May 19, 1980 - March 8, 2015), who passed away under tragic circumstances. We will work hard to finalize and submit this work to the peer-review literature.* Felix J. Herrmann