Answers are grounded in this paper and cite their sources.
This study investigates model order reduction (MOR) in time-lapse full-waveform inversion (FWI) using radial basis function (RBF). It introduces a dual-parameterization approach that combines a reduced-order basis space for time-lapse changes with a finite-difference (FD) grid for the baseline model, facilitating wave propagation. The RBF approach allows for representing time-lapse changes on a nonuniform grid, offering variable resolution. This method is validated through three synthetic examples: a CO2 plume monitoring scenario, a reflection-based time-lapse FWI, and an extension to elastic FWI for near-surface monitoring using surface waves. Results demonstrate effective MOR without compromising FWI accuracy, provide implicit regularization for various geological structures, and highlight the method’s potential to enhance computational efficiency in advanced inversion techniques such as Newton-like methods and Hamiltonian Monte Carlo for Bayesian inference.
No takes yet. Share an insight, caveat, or question.
Li et al. (2024) studied this question.