Augmented surface wave tomography for the Australasian region

An improved 3-D shear wavespeed model for the Australian region has been developed from waveform inversion of Rayleigh wave data. The new model brings in additional data from temporary broad-band station deployments and makes use of an adaptive approach for the choice of starting models for waveform inversion. Hitherto surface wave tomography for the Australian region has been limited by the sparse path coverage in Western Australia. Data from 30 stations deployed in the WA Craton experiment (2000-2001) gives a marked improvement in path coverage and potential resolution. Further data from central and eastern Australia is also included in the analysis. The waveform inversion for each path is built on the automated procedure of Debayle [1999] with adaptive choices of starting model, based on the characteristics of present 3-D shear wavespeed models. Multiple inversions with different starting models allow an improved estimate of both the reliability of the 1-D path-specific models and their errors. The 1-D models are used both directly in a conventional two-stage inversion for 3-D wavespeed, and as estimators of multi-mode dispersion along the path in a three-stage approach via the intermediary of multi-mode phase speed maps as a function of frequency [Kennett and Yoshizawa, 2002]. The use of phase speed maps simplifies the extraction of information on the azimuthal anisotropy of the Rayleigh wavefield as a function of horizontal position and depth. Comparison of the resulting 3-D models highlights the complications introduced by the strong gradients in wavespeed structure in the region.