Looping over a single year using HIST compsets

[hsjung]

What version of the code are you using?

I am using CESM 2.1.5

Have you made any changes to files in the source tree?

No

Describe your problem or question:

Hi everyone,

I am trying to loop over a single year multiple times to get a spin-up run for an experiment.
To do this I have been trying to branch from a restart file from one of the CESM2-LE runs which is run with the compset BHISTsmbb.
I understand that compsets that start with 1850 or 2000 will loop over the forcings representative of that time period while the HIST compsets have a transient forcing.
Is there anyway that I can configure the BHIST compset to loop over a certain year of forcing (for example loop over year 2010 and force it to behave like B2010)?

Thanks!

 

[xnnzka]

Hello Hee-Sung,

Regretfully, I do not know if there is B2010. So I also want to listen suggestions from others.
From my view, I think you could modify the namelist, repeating forcings from one year over variable dynamics use.
I use CAM to as an example:
In user_nl_cam:
&chem_inparm
ext_frc_type = 'CYCLICAL'
ext_frc_cycle_yr = 2010
srf_emis_type = 'CYCLICAL'
srf_emis_cycle_yr = 2010

&chem_surfvals_nl
flbc_type = 'CYCLICAL'
flbc_cycle_yr = 2010

I must say this might be a little tedious, you have to make such modifications in all namelists.

 

[hsjung]

Ahh I see,
Thanks for the advice!
If there is a streamlined method for adjusting each component to cycle over a single year's forcing, I would also love to learn about them too.

 

[nusbaume]

Hi Hee-Sung Jung,

I checked with one of our top CESM users here at NCAR and they provided the same recommendation as @xnnzka, which is that you'll need to make modifications to various namelist variables to set the year that you want to repeat using the historical input data.

To make it easier, they recommended setting up a BHIST and a B1850 case, and then examining the differences between the two cases' namelist files (e.g. all of the XXX_in files that exist in the run directory). That should flag all of the locations that you will need to specify the year and time interpolation type.

Hope that helps, and have a great day!

Jesse

 

[hsjung]

Hi Jesse,

Thank you for the confirmation and suggestion!
I also thought of an alternative to editing the namelist variables and would like to know if this would also be a valid approach.

The basic idea is that I would run a BHIST compset over one year. (ex) run from 2010-01-01 to 2010-12-31
Then I would take the restart file from the end of the initial run and start a hybrid run where I can set the start date as 2010-01-01.
Repeating this multiple times in my view should have the save effect as looping over the year 2010.

I would like to know if this can be a valid alternative to modifying the namelists.
Thanks!

 

[nusbaume]

Hi Hee-Sung Jung,

That may work, although I believe with a hybrid run some variables are initialized to "cold-start" values even if a restart file is present. My recommendation would be to try it for a short time, and then see if the variables you care about have a continuous-looking time series. If yes then you may be good, but if not then it could mean that the hybrid method is re-setting relevant variables every time you restart, in which case using the namelist modification method would likely be the safer option.

Hope that helps!

Jesse

 

[hsjung]

Hi Jesse,

Thanks for all the help!

Hee-Sung