Create new auxiliary history file for cplhist test with ndep data

[adrifoster]

Brief summary of bug

Now that we are reading in ndep data for cplhist cases, we need to update the auxiliary history file that we use for test SMS_D_Ld1_PS.f09_g17.I1850Clm50BgcSpinup.cheyenne_intel.clm-cplhist

This also means that old histaux files are no longer compatible with the latest CDEPS (and CTSM, when we update externals).

General bug information

CTSM version you are using: ctsm5.1.dev106

Does this bug cause significantly incorrect results in the model's science? No

Configurations affected: cplhist

To do

@mvertens is planning to run a short F compset to write out new histaux files with ndep data. We then need to update the testdefs for the above test to reference these new files.

[billsacks]

We'll need to use the upcoming new DATM CPLHIST mode for CMEPS-style cplhist data (which will be called something like CPLHISTCMEPS). This mode isn't implemented yet, but probably will be soon.

[jedwards4b]

I'm a bit confused about this - the lnd_in namelist includes an ndep stream file:
/glade/p/cesmdata/cseg/inputdata/lnd/clm2/ndepdata/fndep_clm_WACCM6_CMIP6piControl001_y21-50avg_1850monthly_0.95x1.25_c180802.nc

Is datm also supposed to provide an ndep data stream? How does that work?

[billsacks]

@jedwards4b very good question. We were just discussing this at yesterday's ctsm-software meeting and realizing we're all confused about the current state of affairs. I know @mvertens explained it to me a month or two ago, but I forget now. The short story is that we're in a state of transition from CTSM reading ndep itself to always getting it from DATM/CAM, but this isn't fully in place yet, and we need to investigate or hear from @mvertens where we are getting ndep from in different situations currently.

But for the purpose of this issue, the relevant point is that I think CDEPS's DATM now always expects an ndep field.

[mvertens]

@jedwards4b @billsacks - to clarify the situation - datm always sends ndep right now - but clm is not yet listening to it. The issue is that cam also needs to send ndep in non-waccm cases and so needs to read in ndep the way that clm does when its not computed prognostically. I started putting in the hooks into cam to do this but did not get a chance to finish it before I left. So there were 3 steps to the plan:

1. Have datm always send ndep
2. Have cam always send ndep
3. Have clm no longer read in ndep but receive it from either datm or cam

Does that make sense? I think the next step is to do 2. I can try to take that on. Please let me know if everyone agrees with this.

[billsacks]

Thanks a lot for clarifying, and for all of your work on this @mvertens !

[billsacks]

Just a quick summary from the discussion I just had with @mvertens - The big challenge right now is that there is a lot of complex logic for how CTSM determines the right forcings & forcing years to use for NDEP in various cases, based on the use case, sim_year and sim_year_range. @mvertens is going to work through how to transfer this logic to DATM and CAM.

[billsacks]

I was wondering how @olyson managed to get cplhist mode working recently given this issue (see also ESCOMP/CDEPS#230). It looks like he did this by doing:

./xmlchange DATM_PRESNDEP=none

I'll do that in our cplhist test to get it working, but longer-term we'll want to resolve ESCOMP/CDEPS#230 and then remove this workaround.