Add dem_array to TerrainCorrectionParameters

[alkirkby]

Currently, dem_array and density array are passed in at the TerrainCorrector stage, separate from the params object.
It would make more sense to include them as part of TerrainCorrectionParameters.

Tasks

• Add dem_array and density_array to TerrainCorrectionParameters
• Remove them from TerrainCorrector
• Refactor TerrainCorrector to use dem from TerrainCorrectionParameters not internal dem array
• Work out how to save the array metadata to the output file

[craigmillernz]

Hi, how is this different from what it already is?

dem_source and density_dataset_source, are already part of TerrainCorrectionParameters

https://gns-science.github.io/gsolve/api/api_index.html#gsolve.TerrainCorrectionParameters

and TerrainCorrectionParameters uses a list of those parameters.

corrector = TerrainCorrector(params=[inner_zone_params, outer_zone_params])

Is your script up to date for the current version? I think dem_array and density_array are obsolete.

https://github.com/GNS-Science/gsolve/blob/main/examples/scripts/terrain_correction_example.py

[craigmillernz]

I dug a bit deeper and it seems TerrainCorrector can also take these dem_array and density_array which to me seems unnecessary. These should be set up through the TerrainCorrectionParameters not stuck in ad hoc in the TerrainCorrector. I suggest we remove them so that TerrainCorrector only takes a list of TerrainCorrectionParameters. This will simplify the code and reduce duplication. We can remove the method add_calculation_zone as to add another zone you just create another TerrainCorrectionParameter.

One option would be for TerrainCorrectionParameters to accept either a filename (as current) or an array if accepting an array (such as a density model in memory) is useful.

[alkirkby]

Yes, that's what we meant, thanks @craigmillernz I think we are all in agreement.
At the moment dem_array and density_array are only provided in TerrainCorrector; the purpose of this issue is to move these parameters to TerrainCorrectionParameters instead.

[craigmillernz]

I guess the question is are they (dem_array and density_array) even needed, or can they be read in file from using the existing dem_source and density_dataset_source. These are likely to be externally created models so reading them from file seems the dominant use case, unless you have an example workflow of having them already in memory from somewhere?