SG-DeepONet: Source-generalized deep operator learning for full waveform inversion

TL;DR

This paper introduces SG-DeepONet, a novel deep learning framework for full waveform inversion that effectively handles diverse source conditions, supported by a new challenging seismic dataset SVFWI.

Contribution

The paper presents SG-DeepONet, a source-generalized deep operator learning model, and a new seismic dataset SVFWI to improve FWI robustness to source variability.

Findings

SG-DeepONet outperforms existing methods in accuracy and robustness.

SVFWI provides a comprehensive benchmark for source-variable seismic data.

The framework effectively integrates source parameters for high-fidelity velocity reconstruction.

Abstract

Full waveform inversion (FWI) aims to reconstruct subsurface velocity models from observed seismic wavefields and has recently benefited from advances in deep learning (DL). The performance of DL-based FWI critically depends on the diversity of training data, yet existing datasets such as OpenFWI rely on fixed or weakly varying source conditions, limiting their ability to represent realistic seismic scenarios and hindering source generalization. To address this issue, we construct a new source-variable seismic dataset, termed SVFWI, by systematically varying the frequencies and horizontal locations of multiple surface sources. SVFWI is further divided into three subsets that respectively model frequency variations, location variations, and their combined effects, providing a challenging benchmark in data-driven FWI. We further propose SG-DeepONet, a novel DeepONet-based encoder-decoder framework tailored for FWI. The branch network extracts multi-scale time-frequency features from seismic observations, the trunk network explicitly embeds source physical parameters, and an interactive decoding network enables effective nonlinear fusion and high-fidelity velocity reconstruction. Extensive experiments on SVFWI demonstrate that SG-DeepONet achieves superior inversion accuracy and robustness under varying source conditions compared with existing DL-based FWI methods.