@sbrus89, could you use this as a starting point to address #439? I think it's most of the way there but I'm not familiar with the test cases affected so could use your help testing/debugging.

@sbrus89, I'll set up a test environment compass_cartopy for you to test with as soon as I can. Stay tuned...

@xylar, sounds great. Thanks for setting this up!

Okay, an environment should be in place. To activate it, do:

source /usr/projects/climate/SHARED_CLIMATE/anaconda_envs/base/etc/profile.d/conda.sh
conda activate compass_cartopy

@xylar, I'm running into an error. When cartopy tries downloading the coastline data, I get a 403: Forbidden. I'd say this probably some kind of proxy issue, but I do have all of my proxy settings in place (https://hpc.lanl.gov/proxy_setup)

Shoot, that's a tricky one. We've dealt with this in MPAS-Analysis:
https://github.com/MPAS-Dev/MPAS-Analysis#download-natural-earth-data-for-cartopy

I'll move the data over from my laptop.

Oh good. I'm glad this has already been sorted out for MPAS-Analysis. Thanks for moving the data. I'll try again once it is place.

@sbrus89, I put the files at /usr/projects/climate/SHARED_CLIMATE/anaconda_envs/cartopy. I'm not aware of a way to configure the environment to look there, so maybe you can link your local path ~/.local/share/cartopy to that folder or if the folder is already there, link the shapefiles subfolder to be within that local folder.

SciTools/cartopy#1325 (comment) gives a suggestion on how to get the directory we need from an environment variable. That variable could be set when COMPASS is activated but we would need to add the relevant check. I'll do this.

@sbrus89, I just added activation/deactivation scripts to the compass_cartopy environment that should set CARTOPY_DIR when the environment is active (and unset it when you deactivate).

Could you delete your ~/.local/share/cartopy and see if things still work for you?

Make sure you deactivate and reactivate compass_cartopy and also that you get the commit I just pushed to this branch.

A couple of updates for this PR. First, the new definition of COMPASS here:
MPAS-Dev/MPAS-Tools#285
will include cartopy and not basemap so we can test this PR with that environment before the COMPASS metapackage becomes official.

Second, regarding the offline data for cartopy, this issue is being addressed by adding a cartopy_offlinedata package to conda-forge:
conda-forge/staged-recipes#10831

@sbrus89, I just created a new conda environment for compass 0.1.0 that includes both cartopy and cartopy_offlinedata (a conda package with all of the data cartopy would download).

Could you do the following for me as a test before we merge this PR?

mv ~/.local/share/cartopy ~/.local/share/old_cartopy

Then, make sure you can run one of your test cases that now uses cartopy by activating the new environment:

source /usr/projects/climate/SHARED_CLIMATE/anaconda_envs/base/etc/profile.d/conda.sh
conda activate compass_0.1.0

I'd like to switch to this new version of COMPASS by early next week. Could you let me know if you'll be able to test by then? I'm only at LANL until Wednesday.

@xylar, this sounds great. I will try to test over the next day or so. Thanks again for taking this on!

@xylar, I tested this on the NAEC60to30cr8 case and it works great. I didn't have any issues with using the offline data.

@pwolfram or @mark-petersen, this is ready to merge at your (earliest) convenience.

I still need to test the interpolate_time_varying_forcing.py script

Ah, sorry, then let us know when you're happy.

Sorry about jumping the gun there.

@xylar, Alright, things are looking more reasonable now. I increased the number of plotting levels and had to use different scales for the two plots. Something is definitely buggy because it should work without having to do that.

I'll try pushing the changes to your branch and will do a PR if I can't.

@xylar, I opened a PR to your fork.

Just so you know, I also tested this with the compass_0.1.0 environment and it works as well.

Okay, not perfect but I agree we can live with it for now. Sorry this is such a pain.

@mark-petersen or @pwolfram, which of you would like to merge this PR into ocean/coastal?

@xylar and @mark-petersen, I can take care of this later, especially since I'm cranking on coastal right now. Thanks!

Thanks for merging this, @pwolfram! We need to get it into ocean/develop at some point so we also don't have trouble with basemap there.

@xylar is cartopy useful for our work on V2 meshes this week? If so, I think this is an example that justifies a parallel identical PR into ocean/develop, as I won't merge ocean/coastal into ocean/develop for 2-4 more weeks.

Yes, we need this in ocean/develop because otherwise we're going to have the same broken imports unless we explicitly add basemap back into the testing compass conda envrionment. So let's include it in the test branch.