Accelerating Numerical Wave Propagation by Wavefield Adapted Meshes, Part II: Full-Waveform Inversion

Abstract

We present a novel full-waveform inversion approach which can reduce the computational cost by up to an order of magnitude compared to conventional approaches, provided that variations in medium properties are sufficiently smooth. Our method is based on the usage of wavefield-adapted meshes which accelerate the forward and adjoint wavefield simulations. By adapting the mesh to the expected complexity and smoothness of the wavefield, the number of elements needed to discretise the wave equation can be greatly reduced. This leads to spectral-element meshes which are optimally tailored to source locations and medium complexity. We demonstrate a workflow which opens up the possibility to use these meshes in full-waveform inversion and show the computational advantages of the approach. We provide examples in 2-D and 3-D to illustrate the concept, describe how the new workflow deviates from the standard full-waveform inversion workflow, and explain the additional steps in detail.

WUS256: An Adjoint Waveform Tomography Model of the Crust and Upper Mantle of the Western United States for Improved Waveform Simulations

“…The greatly improved waveform fits provided by WUS256 demonstrates the promise of AWT to infer Earth structure from recorded seismograms. WUS256 could serve as a starting model for further improvements in seismic structure in the WUS or to represent structure in this part of the world for global models (e.g., Noe et al., 2021; Thrastarson et al., 2020; van Herwaarden et al., 2020).…”

Full Waveform Inversion Beneath the Central Andes: Insight Into the Dehydration of the Nazca Slab and Delamination of the Back‐Arc Lithosphere

“…Our waveform modeling and inversion are mainly based on the full waveform adjoint methodology (Fichtner et al., 2009; Tromp et al., 2004). Solutions of the visco‐elastic wave equation in a radially anisotropic earth media are obtained from Salvus (Afanasiev et al., 2019), which is a suite of highly parallelized software performing full waveform modeling and inversion, which makes use of graphics processing unit acceleration and offers wavefield adapted meshes (Thrastarson et al., 2020; van Driel et al., 2020). Compared to earlier works, we introduce some technical modifications of the inversion workflow and misfit functionals, with details presented below.…”

Adjoint tomography of the Hikurangi subduction zone and the North Island of New Zealand

“…Similarly, data storage of models, kernels, waveforms etc., necessary to solve the forward and inverse problems, push the limits of modern clusters, while complexity scales as tomography implements larger domains, higher resolutions, and increased data. Nevertheless, the trade-offs of full-waveform methods are attractive given the physical accuracy they provide, and much of the current research in the field is focused on optimizing the forward and inverse problems through novel approaches such as wavefield adapted meshes (Thrastarson et al, 2020), or dynamic batching (van Driel et al, 2020;van Herwaarden et al, 2020).…”

“…Rotating the New Zealand domain, as done with the NZ-Wide2.2 velocity model (Eberhart-Phillips et al, 2020b), may reduce the amount of unused domain space, but also increases problem complexity. Custom made meshes could also be designed, such as those that only provide simulation domain for the largest epicentral distance of a given source, or incorporating the idea of wavefield adapted meshes as in Thrastarson et al (2020). These solutions trade computational cost for setup cost, and may be worthwhile for the resultant unified tomography model.…”