Robust Physics-Guided Diffusion for Full-Waveform Inversion

TL;DR

This paper introduces a robust physics-guided diffusion method for full-waveform inversion that enhances reconstruction quality by integrating a Wasserstein-2 based data consistency and a preconditioned reverse-diffusion scheme, outperforming standard methods.

Contribution

It proposes a novel diffusion framework combining Wasserstein-2 based likelihood guidance with a preconditioned reverse-diffusion scheme for improved robustness and stability in waveform inversion.

Findings

Enhanced reconstruction quality over deterministic baselines

Improved robustness to amplitude imbalance and phase misalignment

More stable and effective data-guidance compared to standard diffusion methods

Abstract

We develop a robust physics-guided diffusion framework for full-waveform inversion that combines a score-based generative prior with likelihood guidance computed through wave-equation simulations. We adopt a transport-based data-consistency potential (Wasserstein-2), incorporating wavefield enhancement via bounded weighting and observation-dependent normalization, thereby improving robustness to amplitude imbalance and time/phase misalignment. On the inference side, we introduce a preconditioned guided reverse-diffusion scheme that adapts the guidance strength and spatial scaling throughout the reverse-time dynamics, yielding a more stable and effective data-consistency guidance step than standard diffusion posterior sampling (DPS). Numerical experiments on OpenFWI datasets demonstrate improved reconstruction quality over deterministic optimization baselines and standard DPS under comparable computational budgets.