A full waveform tomography algorithm for teleseismic body and surface waves in 2.5 dimensions

Abstract

We describe a 2.5-D, frequency domain, viscoelastic waveform tomography algorithm for imaging with seismograms of teleseismic body and surface waves recorded by quasi-linear arrays. The equations of motion are discretized with p-adaptive finite elements that allow for geometric flexibility and accurate solutions as a function of wavelength. Artificial forces are introduced into the media by specifying a known wavefield along the model edges and solving for the corresponding scattered field. Because of the relatively low frequency content of teleseismic data, regional scale tectonic settings can be parametrized with a modest number of variables and perturbations can be determined directly from a regularized Gauss-Newton system of equations.Waveforms generated by the forward problem compare well with analytic solutions for simple 1-D and 2-D media. Tests of different approaches to the inverse problem show that the use of an approximate Hessian serves to properly focus the scattered field. We also find that while full waveform inversion can provide significantly better resolution than standard techniques for both body and surface wave tomography modelled individually, joint inversion both enhances resolution and mitigates potential artefacts. © The Authors 2014. Published by Oxford University Press on behalf of The Royal Astronomical Society.