FHIST_BGC spinup problem

[Besuyi]

Hi everyone,

I would like to use CESM2.1.3 to study the radiative forcing change caused by land use and land cover change (LULCC). I am going to conduct a historical transient control run and land_changed run, and I am trying to run a FHIST_BGC case on our school’s HPC now. As a novice, something is confusing me:

1. If there is an initial files, for example, FHIST_BGC compset with default finidat = 'b.e20.BHIST.f09_g17.20thC.297_01_v3.clm2.r.1979-01-01-00000.nc', does it mean there is no need for spinup? Besides, the simulation start time cannot be changed, only from 1979?

2. If 1 holds, then my steps would be:
> ./create_newcase --case ./cases/FHIST_BGC_Ctl --res f09_f09_mg17 --compset FHIST_BGC --mach mymach
> cd FHIST_BGC_Ctl
> ./xmlchange RUN_TYPE=branch
> ./case.setup
> ./case.build --skip-provenance-check
> ./xmlchange RESUBMIT=3,STOP_N=10,STOP_OPTION=nyears
> ./case.submit
Above as the control experiment, then change the flanduse_timeseries data, and follow the same steps to simulate again as the land_changed run.
Is this right?

3. If I use the FHIST_BGC for simulation, the difference between this and the “land-hist” in the first phase of LUMIP in CMIP6 is that the latter is land only and the former is CAM+CLM? At the same time, I would also like to know if there is a similar experiment in CMIP6 that can achieve the same thing, so as not to duplicate work.

Thanks in advance!

Best,
Suyi

 

[oleson]

1. If you start from 1979 there shouldn't be any need for a spinup as the case provides an initial file from a BHIST run which should have reasonable initial conditions. There aren't any other years available in inputdata. However, there might be other years available on request, I'm not sure. You could post your request on CESM General and see if the CESM Data Manager has any files.

2. Depending on your flanduse_timeseries file, you might need to do a hybrid run for the land_changed run and interpolate your CLM initial file (use_init_interp=.true.)

3. Correct. And yes, there were some FHIST simulations conducted for CMIP6, in particular you may be interested in the AMIP simulations which I think start in 1950. See:

CESM2 Experiments, Data & Diagnostic Output

Multi-decadal model runs of given component sets at the target resolution, followed by scientific review of the model output diagnostics

csegweb.cgd.ucar.edu

 

[Besuyi]

1. If you start from 1979 there shouldn't be any need for a spinup as the case provides an initial file from a BHIST run which should have reasonable initial conditions. There aren't any other years available in inputdata. However, there might be other years available on request, I'm not sure. You could post your request on CESM General and see if the CESM Data Manager has any files.

2. Depending on your flanduse_timeseries file, you might need to do a hybrid run for the land_changed run and interpolate your CLM initial file (use_init_interp=.true.)

3. Correct. And yes, there were some FHIST simulations conducted for CMIP6, in particular you may be interested in the AMIP simulations which I think start in 1950. See:

CESM2 Experiments, Data & Diagnostic Output

Multi-decadal model runs of given component sets at the target resolution, followed by scientific review of the model output diagnostics

csegweb.cgd.ucar.edu

Thanks for your reply, Mr.Oleson.I will check it！

 

[Besuyi]

Dear Mr.Oleson

I have completed my historic transient control run (default configuration,1979~2014) and then I fixed the values of all landuse variables:

PCT_CROP,PCT_NAT_PFT,PCT_CFT,HARVEST_VH1,HARVEST_VH2,HARVEST_SH1,HARVEST_SH2,HARVEST_SH3,GRAZING,FERTNITRO_CFT,UNREPRESENTED_PFT_LULCC,UNREPRESENTED_CFT_LULCC

in flanduse_timeseries file (landuse.timeseries_0.9x1.25_hist_78pfts_CMIP6_simyr1850-2015_c170824.nc) at yr1850 using NCL, named landuse.timeseries_simyr1850_2015_at1850.nc.

Next, I create land_changed case:

> ./create_newcase --case ./cases/FHIST_BGC_LC --res f09_f09_mg17 --compset FHIST_BGC --mach mymach
> cd FHIST_BGC_LC
> ./xmlchange RUN_TYPE=hybrid
> # edit user_nl_clm
flanduse_timeseries = '/.../landuse.timeseries_simyr1850_2015_at1850.nc'
use_init_interp = .true.
......

This simulation has been running to 2002 now.

But I'm not sure whether it is correct, should I do a spinup firstly? The reason is, I think the initial condition of finidat is 1979, but landuse data begin in 1850 (and matintain until end of run). If this is correct, can I assume that the adjustment of the surface forcing by the atmosphere is very fast, and it will soon back to equilibrium?
Or, should I change finidat to yr1850, like 'b.e20.BHIST.f09_g17.20thC.297_01_v3.clm2.r.1850-01-01-00000.nc'?
Besides，I noticed this thread had similar issue
FHIS_BGC spinup problem
in which you said

"I don't really see that you need to spinup since you just want to compare climate with modified pfts against unmodified pfts.
The assumption is that both simulations have the same initial conditions prior to modifying the pfts."

Can you explain this question for me?
Thanks^^

Best,
Suyi

 

[oleson]

I think you could just discard the first few years of each run, e.g., 5 years, for spinup purposes and analyze the rest of the simulation. I think the atmosphere adjusts fairly quickly to changes like this.
But I would also recommend doing a literature search on modeling the effects of landcover change to see how these issues are dealt with.

 

[Besuyi]

I think you could just discard the first few years of each run, e.g., 5 years, for spinup purposes and analyze the rest of the simulation. I think the atmosphere adjusts fairly quickly to changes like this.
But I would also recommend doing a literature search on modeling the effects of landcover change to see how these issues are dealt with.

Thank you for your prompt reply and assistance!
Have a nice day!

 

[MikeYuan]

I think you could just discard the first few years of each run, e.g., 5 years, for spinup purposes and analyze the rest of the simulation. I think the atmosphere adjusts fairly quickly to changes like this.
But I would also recommend doing a literature search on modeling the effects of landcover change to see how these issues are dealt with.

Dear Mr.Oleson,
Happy new year. Now, I wanna explore the LULCC(land use and land cover) between 2090-2100 by using the ISSP585Clm50BgcCrop. In order to get the initial file for 2090 to do a 2090-2100 run, should I spin-up 75 years(2015-2090), and then run 10 years? In addition, must I turn on the "CLM_ACCELERATED_SPINUP" before the model spin-up?

Looking forward to your reply~
Thank you very much
Mike