Historical transient run reference cases and configuration

[aswann2]

I'm trying to set up a run with time-varying atmospheric forcing (CO2, greenhouse gasses, aerosols) and land use.

I used compset (matches short name FHIST_BGC): HIST_CAM60_CLM50%BGC-CROP_CICE%PRES_DOCN%DOM_MOSART_CISM2%NOEVOLVE_SWAV
and grid: f09_g16

The default build of this creates a "startup" run with a start date of 1979 ( <entry id="RUN_STARTDATE" value="1979-01-01">)
I don't see it creating any "streams" files, but I do see streams files specified in the built namelist for land lnd_in

I want to:
1) verify that this will make a run with transient forcing (backup option is to just run it and analyze after the fact, but seems like there would be a setting I can check somewhere!). I do see streams files specified in the build namelist file lnd_in, but it wasn't clear to me if they were in atm_in . Can someone verify what I should see in atm_in or elsewhere to indicate that it will use time varying forcing?

2) should this be changed to a "hybrid" run instead of a startup with an initial condition of spun up land? If so where do I find the name of an existing historical run (say from CMIP DECK historical runs) to point to? I tried looking in /glade/collections/cmip/CMIP6/CMIP/NCAR/CESM2/historical/r1i1p1f1 but this didn't seem right

3) eventually I want to also put in a custom land use change file. I think I should modify this file from lnd_in and point it to my custom file in user_nl_clm: flanduse_timeseries = '/glade/p/cesmdata/cseg/inputdata/lnd/clm2/surfdata_map/landuse.timeseries_0.9x1.25_hist_78pfts_CMIP6_simyr1850-2015_c170824.nc'

Any help getting pointed in the right direction would be appreciated!

 

[sacks]

Hi Abby,

I think I can answer most of these questions. I'll try, but will also ask one or two others to look this over to make sure I got this right, because I'm not 100% confident about my answers. I'm assuming you're using CESM2.something here, but it would help to know if you're using the scientifically-supported CESM2.1 series or more recent development code.

(0) Typically for F compsets for 1-degree you would use grid f09_f09_mg17. There are two points here: (a) f09_g16 has been phased out and replaced by the newer g17 ocean grid/mask in CESM2 (where the land owns the Caspian Sea); (b) F compsets typically use a resolution where the (data) ocean model is run at the same resolution as the atmosphere, hence f09_f09_mg17 (where the second f09 specifies an ocean grid of f09, matching the atmosphere grid, and mg17 specifies the mask – i.e., the same ocean mask as would be used in a fully-coupled f09_g17 case).

(1) A HIST compset is typically the right choice for transient forcing, but I don't know what to look for in the atmosphere settings to verify that they are set up the way you want. I'll get someone else to help with that piece.

(2) If you are using the scientifically-supported CESM2.1.z series, then this compset should be a hybrid out-of-the-box as long as you use the supported grid, f09_f09_mg17. If you are using a more recent development version, then this is not set up as a hybrid case because we typically reserve hybrid cases for scientifically-supported configurations (in part because they can be hard to support in rapidly-changing development code). However, CLM should still be using spunup initial conditions even in this "startup" run (I know, it's confusing). You can verify that by checking finidat in lnd_in. Typically the out-of-the-box initial conditions should be reasonably compatible with your configuration, at least for a standard compset and grid like this. But they may not be completely in equilibrium, especially if you're using a recent development version that has evolved somewhat from when the initial conditions were created. If you have questions about getting the initial conditions into equilibrium, I can get someone else to help.

(3) Yes, that's right: modify flanduse_timeseries in user_nl_clm.

 

[aswann2]

Thanks Bill!

I'm using cesm2.1.3, which I think is a release version. Not looking to do anything with newer code, so could go back to an earlier version if that is easier.

(0) - I chose this grid because I am planning to also run a slab ocean version in addition to a prescribed SST version, and the slab ocean qflux file I have is on the f09_g16 grid. I had tried the f09_f09_mg17 grid first and couldent get the slab ocean version of my setup to run. I looked here for qflux files here:

/glade/p/cesmdata/cseg/inputdata/ocn/docn7/SOM

and I only see f09_g16 and f09_g17 options

(2) hmm. When I build the HIST simulation env_run.xml has the following by default:

<entry id="RUN_TYPE" value="startup">
<entry id="RUN_REFCASE" value="case.std">

As you said it sets an finidat in the lnd_in and I do think the main things that would be out of synch would be carbon pools and leaves, so these are probably close enough as long as I have a spinup period. By default it set to this:

finidat = '/glade/p/cesmdata/cseg/inputdata/lnd/clm2/initdata_map/clmi.BHIST.2000-01-01.0.9x1.25_gx1v7_simyr2000_c181015.nc'

curiously when I change only the ocean to slab (i.e. go from DOCN%DOM to DOCN%SOM) it switched to a different finidat - but I can just replace it so they match.

 

[sacks]

CESM2.1.3 is a good choice.

(0) I couldn't tell from your response if you are tied to your particular SOM forcings (at f09_g16) or if you are happy to use any recent forcings that are available. In general, I would expect f09_g17 forcings to be more recent and compatible with the recent versions of CESM. But I will try to get someone else to respond about what forcings and resolutions are typically used for SOM cases.

(2) I think you're getting a startup case because of the resolution. The default RUN_TYPE is dependent on both the compset and resolution, and for this compset, I think the default is to only have a hybrid case if you're running at f09_f09_g17 resolution. As you have said, the most important thing here is to use equivalent land initial conditions. I was able to reproduce what you found, that switching DOCN%DOM to DOCN%SOM changes the default land initial conditions. It looks like the reason that happens is because the F compset (DOCN%DOM) is set up with a start year of 1979, whereas the SOM case has a start year of 1850. CLM chooses different initial conditions as being most appropriate for these different start years.

It sounds like you therefore have three reasonable out-of-the-box choices for CLM initial conditions:

(a) Year-1979 initial conditions from the refcase b.e20.BHIST.f09_g17.20thC.297_01_v3/1979-01-01

(b) Year-2000 initial conditions from the default finidat pointed to by CLM in the F compset (clmi.BHIST.2000-01-01.0.9x1.25_gx1v7_simyr2000_c181015.nc)

(c) Year-1850 initial conditions from the default finidat pointed to by CLM in the SOM compset (I'm getting clmi.B1850.1171-01-01.0.9x1.25_gx1v7_simyr1850_c181029.nc for that).

As you say, probably any of them are reasonable starting points as long as you have a spinup period, but I'll see if anyone else here has more guidance about what they would typically choose in this situation.

 

[dbailey]

Hi Abby,

Bill alerted me to this thread. I have an FAQ about the SOM here:

FAQ: Data ocean slab mode (DOCN-SOM)

I get numerous questions about the new SOM in the data ocean model of CESM. So, I thought it appropriate to add a sort of FAQ here. 1. Why does the SOM mode no longer run out of the box? Earlier versions were set up with a test SOM forcing file (pop_frc.1x1d.090130.nc) This forcing file was...

bb.cgd.ucar.edu

Basically, the SOM always runs on the sea ice / ocean grid because SOM cases use the full CICE model. Also, we derive slab forcing from the fully-coupled simulations from POP. If you want a paired F compset with an E compset, then I would use f09_g17 or f19_g17 for both. Then you can create the SST/IFRAC forcing from the coupled model for the DOCN to read in for the F compset. Ideally the SST/IFRAC would be from a similar period to the full SOM forcing.

Dave