Estimation of velocity macro-models using stochastic full-waveform inversion

We present a stochastic full-waveform inversion that makes use of a genetic algorithm to estimate acoustic 2D macro-models of the subsurface. We test this method on the Marmousi model. The proposed method is computationally expensive but yields a final (long wavelength) model that is well-suited to play the role of the starting model for a local full-waveform inversion.