High-frequency global wavefields for local 3D structures by wavefield injection and extrapolation

This is a Preprint and has not been peer reviewed. This is version 1 of this Preprint.


Download Preprint


Marta Pienkowska, Vadim Monteiller , Tarje Nissen-Meyer


Earth structure is multiscale, and seismology remains the primary means of deciphering signatures from small structures over large distances. To enable this at the highest resolution, we present a flexible injection and extrapolation type hybrid framework that couples wavefields from a precomputed global database of accurate Green’s functions with a local three dimensional (3-D) method of choice (e.g. a spectral element of a finite difference solver). The interface allows to embed a full 3-D domain in a spherically symmetric Earth model, tackling large-scale wave propagation with focus on localized heterogeneous complex structures. Thanks to reasonable computational costs (10k CPU hours) and storage requirements (a few TB for 1 Hz waveforms) of databases of global Green’s functions, the method provides coupling of 3-D wavefields that can reach the highest observable body-wave frequencies in the 1-4 Hz range. The framework is highly flexible and adaptable; alterations in source properties (radiation patterns, source-time function), in the source-receiver geometry, and in local domain dimensions and location can be introduced without re-running the global simulation. The once-and-for-all database approach reduces the overall computational cost by a factor of 5,000-100,000 relative to a full 3-D run, provided that the local domain is of the order of tens of wavelengths in size. In this paper we present the details of the method and its implementation, show benchmarks with a 3-D spectral-element solver, discuss its setup-dependent performance, and explore possible wave-propagation applications.




Geophysics and Seismology


computational seismology, wave propagation, body waves, numerical solutions, wave scattering


Published: 2020-10-28 01:01

Last Updated: 2020-10-28 08:01


CC BY Attribution 4.0 International

Add a Comment

You must log in to post a comment.


There are no comments or no comments have been made public for this article.