The first year has seen activity on a number of fronts, including resolution measures for surface waveforms. One focus has been on use of irregular parameterizations in inversion of 2-D reflection data refraction data. Here the unknowns are crustal seismic velocities and depth to the Moho discontinuity. The figure shows an example using a controlled source data set of 36,000 air-gun shots fired from the research vessel Rig seismic and recorder by 44 digital and analogue recorders distributed across Tasmania. (This data set has recently been extended in the TIGGER project presented elsewhere in this report.) A comparison of inversion results with regular and irregular parameterizations shows that the irregular scheme uses about half the number of nodes to represent the Moho surface compared the regular case, and yet produced significantly increased resolution across the region. Note that significant structure is recovered, but regions of the Moho, particularly beneath central Tasmania are still poorly resolved, due to a lack of data coverage.
Figure 1: Comparison of inversion results using regular and irregular grids and wide-angle data from Tasmania. In all diagrams, stars indicate recorder positions and small triangles indicate shot points from which data were picked - contiguous triangles form shot lines. (a) Ray interface hit points and surface patches for the irregular grid. Refracted ray hits are denoted by blue crosses and reflected ray hits are denoted by brown crosses. Interface nodes are represented by dots and thin blue lines represent surface patch boundaries. (b) Same as (a) but for the regular grid. (c) Diagonal elements of the resolution matrix for the irregular grid solution. (d) Same as (c) but for the regular grid. (e) Solution model for the irregular grid. (f) Solution model for the regular grid.
Another area has been the development of interactive visualization package (GeoViz) for interrogating 2-D and 3-D seismic models. A proto-type MATLAB based data handling and visualization system has been built (with the help of ANU visualization laboratory VIZLAB). This combined with public domain visualization tools and animations (see Figure 2.) will allow interactive interrogation of 2-D and 3-D models in real time.
During 2003 the focus will be on development and application of flexible multi-scale parameterizations for dealing with large scale three-dimensional tomographic problems at regional and global scales. Another important issue will be finding mechanisms for robust adaptive smoothing (regularization) of tomographic problems. (A web page has been developed to disseminate results: publications, animations and software from this work). http://rses.anu.edu.au/seismology/projects/tireg )
Figure 2: Data sampling from the Tasmanian Moho inversion. Refracted rays are shown in red; reflected rays are shown in green. Pink dots denote sources and red dots denote receivers. The Tasmanian coastline is shown in purple.
Comments on the maintenance of these frames to Ray Martin:
ray@rses.anu.edu.au