Surface Wave Tomography


Principal Investigator

Eric Debayle

Research School of Earth Sciences


x14 - VPP

Surface wave tomography is a technique used by
seismologists for the delineation of the 3D internal
structure of the Earth mantle. With sufficient coverage of the Earth, lateral variations in seismic parameters as small as 250 km can be resolved for the upper 400 km of the Earth mantle. These observations provide direct insights on the distribution of temperature, composition and deformation in the upper mantle. This work is part of the continuing effort of the Seismology Group at RSES to achieve a high resolution picture of the crust and mantle beneath the Australian continent. It aims to image the Australian upper mantle at a scale which allows confrontation with surface geology. It also reveals at an unprecedent scale the deep structure of a continental plate and its relations with the deeper mantle.


What are the results to date and the future of the work?

At the beginning of this project, a preliminary 3D tomographic model had been obtained, constrained by the vertical component of 668 seismic waveforms (Debayle, 1999). The use of a vectorised code on the VPP 300 allowed us to process larger dataset, resulting in a 3D model constrained by 2194 seismic waveforms, with an improved coverage of the continent compared to previous studies. With the dense path coverage, it is possible in the tomographic inversion to extract simultaneous estimates of the variation in seismic wavespeed and the azimuthal anisotropy associated with the dependence of the seismic velocity on the azimuth of wave propagation. While the variations in seismic wavespeed can be associated with variations in temperature and/or composition in the mantle, azimuthal anisotropy represented by the directions of fast seismic propagation is likely to be associated with the reorientation of mantle mineralogy through deformation e.g. alignment of anisotropic minerals in the mantle parallel to the direction of maximum extension.

The current anisotropic model reveals a major change in the inferred fast direction for seismic propagation beneath Australia, between the uppermost 150 km of the model and the deeper structure. In the uppermost 150 km, a complex anisotropy is observed, likely to be related to the long history of deformation of the Australian continent. At larger depth, a smoother North-South pattern provides strong evidences of large-scale deformation in the Earth mantle, likely related to the northward drift of the Australian tectonic plate. The distribution of seismic

- Appendix A



wavespeed confirms earlier work and shows the complex structure of the Australian plate which appears to be thinner on the eastern margin of the continent compared to central and western Australia.

The existence of large scale deformation at the bottom of continental tectonic plates has been controversial for many years. As our current results provide new strong evidence for the existence of such deformations, it is important to complete our model with independent observations. Our aim is now to better constrain the depth extent of anisotropy by including the horizontal component of the seismograms in the inversion. A better knowledge of the quality factor of the Earth will also help to locate low viscosity layers in the mantle, where such deformations are likely to occur. This requires us to determine a 3D model for the quality factor beneath Australia, which can be done through measurements of surface wave amplitudes.

What computational techniques are used?

A 3D Earth model is retrieved from the path-averaged seismic parameters measured for a large number of source-receiver paths. The algorithm consists of finding a smooth function of the model parameters which explains the path-averaged measurements. This is an under determined inverse problem which requires the introduction of a priori information on the model to ensure the uniqueness of the solution. The estimation of this a priori information largely dominates the computation time as the model is described by more than 150 000 parameters, but the algorithm vectorises to about 95 % and is one of the most efficient codes for surface wave tomography.


E. Debayle, SV-wave azimuthal anisotropy in the Australian upper mantle: preliminary results from automated Rayleigh waveform inversion, Geophys. J. Int., in press.

Appendix A -