Abstract

Making full-waveform inversion (FWI) robust in a realistic scenario when low frequencies are missing, illumination is incomplete, and starting models are uncertain remains a major challenge in seismic imaging.
I develop a method that combines a computationally efficient one-way wave-equation engine with frequency-domain model extension, allowing the inversion to absorb phase mismatch while still converging toward physically meaningful velocity models. 
Across synthetic seismic and medical-ultrasound examples, I demonstrate how the proposed approach can be a viable alternative to conventional full-waveform inversion methods based on two-way wave equations.
Frequency-model extension provides robustness to cycle skipping, while constraints, regularization, multiparameter parameterizations and preconditioning lead to recovering stable, interpretable physical models. 
A 3D field-data application on a marine OBC survey demonstrates the approach works beyond controlled tests, producing approximately frequency-consistent models that improve image focusing.
Finally, I explore connections to renormalization-group ideas as a conceptual lens for interpreting model extension.