arXiv:2603.00377v2 Announce Type: replace Abstract: Full Waveform Inversion (FWI) is a highly nonlinear and ill-posed problem that aims to recover subsurface velocity maps from surface-recorded seismic waveforms data. Existing data-driven FWI typically uses small models, as available datasets have limited volume, geological diversity, and spatial extent, leading to substantial concerns about overfitting. Although they perform well on synthetic datasets, current methods fail to generalize to more realistic geological structures. In this work, we show that a model trained entirely on simulated a
The paper addresses a current limitation in data-driven Full Waveform Inversion concerning generalization, suggesting a path forward by leveraging large-scale simulated data, which is becoming more feasible now. This particular paper is a 'v2', indicating refinement and continued work in this area.
Improving FWI generalization has direct implications for sectors relying on accurate subsurface mapping, such as energy exploration and geological surveying, by enhancing efficiency and reducing the need for costly physical data acquisition. This is an application of AI that moves beyond 'small models' into 'large model era'.
The ability to train FWI models entirely on simulated data, without requiring extensive real-world datasets for generalization, changes the paradigm for subsurface imaging. It potentially lowers the barrier to entry and increases the accuracy of predictions in complex geological structures.
- · Oil & Gas exploration companies
- · Geophysical services providers
- · AI model developers
- · Computational simulation software companies
- · Companies reliant on traditional, data-intensive FWI methods
- · Organizations with limited access to large-scale geological datasets
More accurate and efficient subsurface resource discovery and mapping becomes possible.
Reduced operational costs and environmental impact in sectors like energy exploration due to fewer physical surveys and more precise drilling.
New geological insights could lead to discovery of previously uneconomical or inaccessible resource deposits, impacting global resource markets and geopolitics.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG