FHIS_BGC spinup problem

[heyaqian1987@gmail_com]

Hi everyone,

I would like to use FHIS_BGC to conduct my research. However, since I modified the pft types in surface data, I feel I need to spinup BGC. Am I right?

However, when I did spinup (follow the link: 1.5.5. Spinup of CLM5.0-BGC-Crop — clmdoc release-clm5.0.01 documentation), the time for FHIS_BGC is from 1979 to 2015, I do not know how can I spin up for 200+ years. Can anyone give some suggestions or references?

Thanks in advance!

Yaqian

 

[heyaqian1987@gmail_com]

Here is my script:

STEP 1: spin up
cd /glade/p/cesm/releases/cesm2_1_1/cime/scripts
./create_newcase --case ~/cases/bgc_spinup --res f09_g17 --compset FHIST_BGC --project UDRT0013 --run-unsupported
cd ~/cases/bgc_spinup
# change accelerated spinup mode
./xmlchange CLM_ACCELERATED_SPINUP='on'

# now setup
./case.setup

# edit/check user_nl_clm for my own surface data
flanduse_timeseries = '/glade/u/home/yahe/lulcdata/landuse.timeseries_0.9x1.25_hist_78pfts_CMIP6_simyr1850-2015_c170824_netet_floor_c4g.nc'

./preview_namelists
qcmd -- ./case.build

# the following sets Resubmit to 3 times in env_run.xml
# the following sets STOP_DATE, STOP_N and STOP_OPTION to jan/1/0201, 20, “nyears” in env_run.xml
./xmlchange RESUBMIT=3,STOP_N=50,STOP_OPTION=nyears,STOP_DATE=02010101

#Now run normally
./case.submit


Thanks in advance!

Yaqian

 

[oleson]

One suggestion is to do some spinup in land-only mode. You could try setting up an I2000 compset with the default initial file (finidat in lnd_in) and your new pft files, and run in AD mode, then post-AD mode. Much cheaper in land-only mode. Then use the resulting initial (restart) file to start off your FHIST_BGC run.

 

[heyaqian1987@gmail_com]

Dear Oleson,

Thanks a lot for your very helpful suggestions! Can I ask two more questions based on your suggestion? 1) I am wondering do you mean I could follow this link to do the spinup (1.5.7. Running with atmospheric forcing from a previous simulation — clmdoc release-clm5.0.01 documentation)? 2) since I modify annual pft from 1982-2013, so do you think an IHIST compset is more suitable?

Thanks in advance.

Yaqian

 

[oleson]

1) Right, that's one way to do it. You could also spinup with the GSWP3 atmospheric forcing that is used with CLM5. It is essentially a reanalysis based on observations. If for example, your final simulation will start in 1982, you could spinup using GSWP3 data for 20 years prior repeatedly (cycle over 1962-1981 GSWP3 data), then use the resulting initial file for your 1982-2013 simulation. 2) Generally, we do spinups for the year when the actual simulation starts. For example, we spinup using 1850 surface data (and other data such as nitrogen deposition) before starting an 1850-2014 run. You could spinup using 1962-1981 surface data in an IHIST compset but it probably doesn't matter much as the pfts likely don't change that much during those 20 years. Plus you'd have a jump in pft distribution (and other forcing fields) when looping from 1981 back to 1962. So it might be best to simply use 1982 surface data to spinup.

 

[heyaqian1987@gmail_com]

Thanks a lot for your answers, Oleson! I am wondering if I use 1982 surface data to spinup, I need to create 1982 surface data, right? And since I modified pfts in landuse.timeseries only from 1982-3013, spinuping without this period is ok? In my understanding, I need to spin up BGC model is because that I modified the pfts, so the spinup needs to include the modified time-period? Or it does not matter? Thanks in advanced.

 

[oleson]

Well, that is a good point. If you are only interested in seeing what the effects of changing the pfts for 1982-2013 are, then you don't really need to spinup. Presumably, there is an initial file associated with the FHIST_BGC compset for 1979 that comes from a default transient simulation. I assume you'll be comparing your simulation to a control simulation (unmodified pfts) for 1979-2013....

 

[heyaqian1987@gmail_com]

Thanks a lot, Oleson. I understand it now.

Yes! I want to compare with modified pfts from 1979-213 with unmodified pfts. Finally, I want to run FHIST_BGC with modified pfts to explore land use and land cover changes impact on precipitation changes.

 

[heyaqian1987@gmail_com]

Hi Oleson, Happy New Year!

I am wondering if I want to spin up land-only mode with modified 1982-2013 pfts, how many years do you think I need to run?

During last few days, I was trying to spin up CLM5 using IHIST_BGC from 1982-2013 with modified pfts, after 200 years AD mode run, it is not equilibrium at all. I feel it may be because I changed pfts for 32 years, so I need to run 200*32 years?

I am also wondering if I only changed pfts and did not change any other surface data, as you mentioned in the previous reply, I actually do not need to spin up IHIST_BGC or FHIST_BGC, right?

Thanks in advance!

Yaqian

 

[oleson]

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.

 

[heyaqian1987@gmail_com]

Thanks Oleson. Then I will try without spin-up to see if there is any unexpected things.