Approximate Langevin Monte Carlo with Adaptation for Bayesian Full-Waveform Inversion

Abstract
In this work, we present a proof of concept for Bayesian full-waveform inversion (FWI) in 2-D. This is based on approximate Langevin Monte Carlo sampling with a gradient-based adaptation of the posterior distribution. We apply our method to the Marmousi model, and it reliably recovers important aspects of the posterior, including the statistical moments, and 1-D and 2-D marginals. Depending on the variations of seismic velocities, the posterior can be significantly non-Gaussian, which directly suggest that using a Hessian approximation for uncertainty quantification in FWI may not be sufficient.

Citation
Izzatullah, M., Van Leeuwen, T., & Peter, D. (2021). Approximate Langevin Monte Carlo with Adaptation for Bayesian Full-Waveform Inversion. 82nd EAGE Annual Conference & Exhibition. doi:10.3997/2214-4609.202112443

Acknowledgements
The first author would like to thank Tristan van Leeuwen at Utrecht University for visiting his research lab, which led to this work, and his continuous support. The research visits and the work reported here were supported by funding from King Abdullah University of Science and Technology (KAUST).

Publisher
European Association of Geoscientists & Engineers

Conference/Event Name
82nd EAGE Annual Conference & Exhibition

DOI
10.3997/2214-4609.202112443

Additional Links
https://www.earthdoc.org/content/papers/10.3997/2214-4609.202112443

Permanent link to this record