FES2012 support? about pytmd

[tsutterley]

Hey @robbibt,
I never knew if people would want to use the earlier models :)
I put in PR #212 to get things up to speed to running FES2012. I had to fix some variable names, and fix that the eps2 constituent is named e2 in FES2012. Using these parameters in a model definition file should let you use the 2012 model (set directory in the io.model class equal to the path to the fes2012 directory,). I checked the outputs using one of the notebooks.

format      FES
name        FES2012
model_file  fes2012/*_FES2012_SLEV.nc.gz
type        z
version     FES2012
variable    tide_ocean
scale     0.01
compressed  True
reference   https://www.aviso.altimetry.fr/en/data/products/auxiliary-products/global-tide-fes.html

[robbibt]

We didn't realise we would need the earlier models either, but we have some areas along the Australian coastline where FES2014 unfortunately performs really poorly compared to FES2012; there's a neat paper here:
https://www.mdpi.com/2072-4292/11/10/1211

Thanks so much for this update - I think this will give us a much-needed nudge to finally update to pyTMD > 2.0!

[robbibt]

Hey @tsutterley, I've been able to sucessfully model some FES2012 tides using the model definition approach above (modifying it slightly to run on the uncompressed files) - thanks so much for this! Definitely a big improvement with FES2012 in some of our problem areas, so this will be really useful (blue = satellite pixels that get wetter with increasing tide heights):

A related question: is there any way to essentially bypass the requirement for reading in an external definition file by declaring the parameters in the file in Python code? I noticed that pyTMD.io.model(...) seems to accept most of the params defined in the definition file - should I be able to just populate it with the content from the definition file directly? (I guess I could always write out the definition file on-the-fly when my code is run, but that feels a little clunky...).

[tsutterley]

Great! That's a major improvement!
I've actually been thinking about the subject of your related question lately. The existing solution with definition files aren't my favorite. I think having the parameters be able to be entered as kwargs might be a nice way of generalizing the model class. It should also (hopefully) require less overall upkeep. I'll work on it 👍