Deep neural Helmholtz operators for 3-D elastic wave propagation and inversion
Geophysical Journal International · 07.10.2024 · 239 (3) · 1469-1484
DOI: 10.1093/gji/ggae342 ↗Kurz gefasst
Fachartikel in der Zeitschrift Geophysical Journal International, online erschienen am 18.09.2024. Verfasst von Caifeng Zou und 3 weiteren Autor:innen. Die Arbeit wurde bisher 44-mal zitiert (Crossref) und nennt 83 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: David and Lucile Packard Foundation.
Abstract (Originalsprache)
Numerical simulations of seismic wave propagation in heterogeneous 3-D media are central to investigating subsurface structures and understanding earthquake processes, yet are computationally expensive for large problems. This is particularly problematic for full-waveform inversion (FWI), which typically involves numerous runs of the forward process. In machine learning there has been considerable recent work in the area of operator learning, with a new class of models called neural operators allowing for data-driven solutions to partial differential equations. Recent work in seismology has shown that when neural operators are adequately trained, they can significantly shorten the compute time for wave propagation. However, the memory required for the 3-D time domain equations may be prohibitive. In this study, we show that these limitations can be overcome by solving the wave equations in the frequency domain, also known as the Helmholtz equations, since the solutions for a set of frequencies can be determined in parallel. The 3-D Helmholtz neural operator is 40 times more memory-efficient than an equivalent time-domain version. We use a Helmholtz neural operator for 2-D and 3-D elastic wave modelling, achieving two orders of magnitude acceleration compared to a baseline spectral element method. The neural operator accurately generalizes to variable velocity structures and can be evaluated on denser input meshes than used in the training simulations. We also show that when solving for wavefields strictly at the free surface, the accuracy can be significantly improved via a graph neural operator layer. In leveraging automatic differentiation, the proposed method can serve as an alternative to the adjoint-state approach for 3-D FWI, reducing the computation time by a factor of 350.
Förderung
- David and Lucile Packard Foundation
Zitieren
APA
Zou, C., Azizzadenesheli, K., Ross, Z. E., & Clayton, R. W. (2024). Deep neural Helmholtz operators for 3-D elastic wave propagation and inversion. Geophysical Journal International, 239(3), 1469-1484. https://doi.org/10.1093/gji/ggae342
Vancouver
Zou C, Azizzadenesheli K, Ross ZE, Clayton RW. Deep neural Helmholtz operators for 3-D elastic wave propagation and inversion. Geophysical Journal International. 2024;239(3):1469-1484. doi:10.1093/gji/ggae342
BibTeX
@article{Zou2024,
author = {Zou, Caifeng and Azizzadenesheli, Kamyar and Ross, Zachary E and Clayton, Robert W},
title = {Deep neural Helmholtz operators for 3-D elastic wave propagation and inversion},
journal = {Geophysical Journal International},
year = {2024},
volume = {239},
number = {3},
pages = {1469--1484},
doi = {10.1093/gji/ggae342}
}
EUVIMED