Hybrid run in gray-radiation aquaplanet

[leahb]

Hi,

I'm using the FGRAYRAD compset in cesm2_3_alpha17f. I want to do a 2xCO2-like experiment (instantaneous increase in longwave optical depth), but when I tried to do a hybrid run it just initializes to the default starting temperature rather than starting from the reference case. Do the simpler models support hybrid runs?

Here's my process:
1. create case
`./create_newcase --case ~/cases/fgray.ghg_reg.winter --compset FGRAYRAD --res f09_f09_mg17 --run-unsupported`
`cd ~/cases/fgray.ghg_reg.winter`
`./case.setup`

2. set hybrid
`./xmlchange RUN_REFCASE=fgray.ref.winter`
`./xmlchange RUN_REFDATE=0003-01-01`
`./xmlchange RUN_STARTDATE=0003-01-01`
`./xmlchange GET_REFCASE=FALSE`

3. change namelists
add `frierson_tau_eqtr=7.8`to `$CASEDIR/user_nl_cam`
add `frierson_tau_pole=1.95`to `$CASEDIR/user_nl_cam

4. run
`./xmlchange STOP_N=12,STOP_OPTION=nmonths`
`qcmd -- ./case.build`
`./case.submit`

 

[leahb]

Oops! I was missing a few steps. Here's the corrected workflow:

1. boilerplate
1. `cd ~/code/$CESM/cime/scripts`
2. `./create_newcase --case ~/cases/fgray.ghg_reg.winter --compset FGRAYRAD --res f09_f09_mg17 --run-unsupported`
3. `cd ~/cases/fgray.ghg_reg.winter`
4. `./case.setup`
2. set hybrid
1. `./xmlchange RUN_TYPE=hybrid`
2. `./xmlchange RUN_REFCASE=fgray.ref.winter`
3. `./xmlchange RUN_REFDATE=0004-01-01`
4. `./xmlchange RUN_STARTDATE=0004-01-01`
5. `./xmlchange GET_REFCASE=TRUE`
6. `cp -u -p /glade/derecho/scratch/$USER/fgray.ref.winter/run/* /glade/derecho/scratch/$USER/fgray.ghg_reg.winter/run`
3. change namelists
1. `echo frierson_tau_eqtr=7.8 >> user_nl_cam`
2. `echo frierson_tau_pole=1.95 >> user_nl_cam`
4. set run time
1. `./xmlchange STOP_N=12,STOP_OPTION=nmonths`
5. boilerplate
1. `qcmd -- ./case.build`
2. `./case.submit`

But I still see initialization from the default global mean surface temperature of ~290 K, even though the logs note that the restart files from my previous run were loaded successfully.

 

[islas]

Hi Leah,

I'm able to reproduce your problem. It seems like it's going back to the analytic initial conditions for a hybrid run, which I don't think it should be doing. It seems like it works for a branch run. I will ask the software engineers about this. In the mean time, do you think you can do any of what you need to do with a branch run?

Isla