Create FATES BCI user_mods on Derecho so we can run it for tests and FATES applications

[adrifoster]

We'd like to have a Barro Colorado Island (BCI) config for FATES - and call it "scientifically supported" so that we'd have an out-of-the-box configuration for FATES that CTSM folks can use.

For this, we just need relevant input files for BCI - from @rgknox or @ckoven or @glemieux ?

Thanks!

Definition of done:

• Add a rule to the mksurfdata_esmf Makefile to create the surface dataset
• Add a test using the new surface dataset
• Add BCI site to ccs_config (Replace 1x1_brazil grid with 1x1_barro_co_islandPC grid ESMCI/ccs_config_cesm#120)
• Update our tests to use the site name in ccs_config

[adrifoster]

@glemieux has files for this already on Cheyenne so I will use those for now

[ekluzek]

@adrifoster in the ccs_config issue I propose the name be...

1x1_barro_co_islandPC

That follows the convention used for other sites, but we can modify it it would be better...

[ekluzek]

Note, that doing this also provides us at least one site without a cold start so resolves this issue:

#1277

[ekluzek]

This relates to #2356

[wwieder]

In the FATES call today Charlie mentioned that BCI runs are forced with observed, high frequency meterology. This makes me wonder if a test that uses global (or subset) DATM data is warranted as a system test for running a 'generic single point' using subset_data? In the SE meeting @slevis-lmwg also suggested that the smallville tests should accomplish this?

@ekluzek and @slevis-lmwg I'll let you evaluate if we have sufficient testing for the CTSM5.2 PR.

[ekluzek]

@wwieder this is a post ctsm5.2.0 activity. It's in the CTSM5.2 board as post CTSM5.2 work.

But, I am excited that we have MET data for BCI, that is really awesome. So that takes care of having a case that has tower site data that's not part of PLUMBER2 or NEON. I'd really like to see this case functional to make running a calibrated FATES site easy for the CTSM community.

I'm going to look into the other idea of having a test for subset_data with datm data and open an issue if that's missing. I need to check into the existing smallville tests to make sure it would be useful. I agree if we have that case covered there is no need for it...

[ekluzek]

We talked about this as something we would still like to do and have available. It would be nice to have it available in the CESM3 release for example as a calibrated FATES site that users can use. But, it isn't the highest priority and we don't have resources to do it right now.