Vector Splitting & Mesh Evaluation

[Ulvetanna]

@hjabbot is exploring different splitting conditions for different vector fields, and evaluating different meshes with respect to original data.

Plotting this work for the mesh paper

[hjabbot]

To date we have done work to verify that the mesh construction behaves as expected with vector splitting. There is some arbitrary-ness about the threshold value used, but we have also developed a method for choosing an appropriate split threshold value for each given source of data. Annoyingly, it doesn't look like it's inherently transferrable between data sources, or even different areas within the same dataset. This method could be extended to scalar datasets as well, as their splitting conditions are also somewhat arbitrary (although they tend to be based on some real-world justification).

Further work needs to be done to add upper/lower bound to the splitting conditions. This would not work for Reynolds splitting, but we're leaning towards just using curl as the split condition for vector fields anyway, and upper/lower bounds would be applicable to these fields.