Add a transient FATES compset alias and make sure do_transient settings are correct by default for FATES

[ekluzek]

Transient simulations are started to be done for FATES. We should add an alias for this and also make sure the namelist is setup to work by default so that do_transient_pfts is FALSE.

Definiton of done:

• Add IHistClm60FatesRs compset to config_compsets.xml (also IHistClm60FatesSpRsGs?)
• Change build-namelist to ensure that a transient compset is used when use_fates_luh is TRUE and abort if not. Abort if it's FALSE and it IS a transient compset
• Add settings to bld/namelist_files/use_cases/*_transient.xml for FATES LUH settings
• Change build-namelist testing that the above works correctly for the options
• Make sure build-namelist testing works
• Adjust testlist to use transient compsets
• Run standard testing as normal

[JoshuaRady]

I've been uinsg 2000_DATM%GSWP3v1_CLM50%FATES_SICE_SOCN_SROF_SGLC_SWAV and recently switched to 2000_DATM%GSWP3v1_CLM50%FATES_SICE_SOCN_RTM_SGLC_SWAV because it is more similar to what can be done with anomaly forcing. I'd be interested to know what might be wrong with this. I've never been able figure out if I should be using 1850_DATM for historical transients. Frankly I don't fully understand with that specifically changes.

The example case uses a compset that hasn't worked in some time (I2000Clm50FatesGs).

[ekluzek]

@JoshuaRady I edited the example to use I2000Clm50FatesRs, which is one that exists now.

The difference between the two compsets you give above is just adding the RTM river model rather than the stub-river. This would only be helpful if you are actually going to look at your RTM output and compare to streamflow for example. So I suspect you would be better off with the stub-river. If you aren't looking at the output, you might as well save a little bit of CPU time.

The "2000" at the beginning is the "period" to simulate for. So 2000 means year 2000, and 1850 means pre-industrial for year 1850. A transient case is HIST for Historical. And there's also options for the future scenarios SSP-RCP which are transient simulations also that go up to year 2100. What this part changes is some of the DATM_* and CLM_* XML case settings, which then get reflected in the namelists. Does that help?

[JoshuaRady]

Thanks for updating the example case.

I don't need the RTM data currently but had problems getting the SROF to work with anomaly forcing. @jkshuman turned me on to using RTM. An example anomaly compset that works for me is SSP585_DATM%GSWP3v1_CLM50%FATES_SICE_SOCN_RTM_SGLC_SWAV.
Correction: I had problems with regional simulations with SROF, not anomaly forcing.

I understood what the descriptors mean but not all the settings that make them different.
Some time ago I think I tried 1850 and HIST comsets but had issues. My current compsets started with the example compset which I have modified over time. I make a bunch of XML setting changes on top of it to, as far as I can tell, make work like the HIST compset. There is a good chance I missed something though. I have pretty much stayed with those settings out of fear of breaking something. If there are 1850 and HIST compsets that are now supported I very glad to switch to them. Thanks for your work on this.

[ekluzek]

Here's how I think this should work:

#2304 (comment)

[ekluzek]

In talking with @glemieux we can do this in small steps on b4b-dev. It's isn't a high priority for either of us, but as it's a usability issue I'd like to prioritize it for the CESM3 release.