Project Abstract
|
In previous work (Williams and Wallace, 2015) we used Green’s functions generated with the PyLith finite element code (Aagaard et al., 2013) to examine the effects of material heterogeneity on geodetic inversions of slow slip events (SSEs) along the Hikurangi Margin, New Zealand. We are extending this work to include interseismic coupling models for the North Island of New Zealand. For both of these projects we have used PyLith to generate Green’s functions and Defnode (McCaffrey, 1995; 2002) to perform geodetic inversions. We have developed a workflow that allows us to perform the necessary tasks in a semi-automated way.
The most difficult task is mesh generation. We have developed Python scripts to create geometry for use with the Trelis meshing package, using nodal information from Defnode. Additional scripts generate an initial mesh, impose a user- defined sizing function, and create a refined mesh suitable for generating Green’s functions. We generate Green’s functions with PyLith, and integrate them to provide the Green’s functions required by Defnode. Once the Defnode Green’s functions have been generated we use another script to launch many Defnode inversions in parallel. We are still improving our workflow, and once it is finished it will be publicly available. Our initial research has shown that using material properties from a seismic velocity model can have significant effects on predicted SSE slip distributions, and the same is likely true for interseismic coupling models. The workflow that we are developing should allow others to explore these effects in other regions. |