Running CESM2 for a short run/different restart files

[Joey Wong]

I ran two experiments using CESM2, both with the same settings except for STOP_N and STOP_OPTION.

In Experiment A, it looks like this:

./xmlchange STOP_OPTION=nday

./xmlchange STOP_N=1

But in Experiment B it is like this

./xmlchange STOP_OPTION=nhours

./xmlchange STOP_N=6

./xmlchange RESUBMIT=3

For both experiments, the model will end after running for a day. If the start date is 0001-01-02-00000, I get a restart file *.cam.r.0001-01-02-00000.nc.

But I found that the two experiments produced different restart files. That's the instantaneous value in the restart file, right？

If this is true, then both experiments give instantaneous values at 0001-01-02-00000.

Why do I get different restart files?

Can anyone give me some advice?

 

[peverley]

Hi.

Can you clarify what is different about the files? Are you comparing the cam.r.0001-01-02-00000.nc file with the file of the same name from the resubmitted run? For the resubmitted run, you will have 4 restart files, one for the end of each run. The last restart will be the cam.r.0001-01-02-00000.nc file. For the ndays run, you'll only have one restart file.

Hope that helps.
Courtney

 

[Joey Wong]

Hi.

Can you clarify what is different about the files? Are you comparing the cam.r.0001-01-02-00000.nc file with the file of the same name from the resubmitted run? For the resubmitted run, you will have 4 restart files, one for the end of each run. The last restart will be the cam.r.0001-01-02-00000.nc file. For the ndays run, you'll only have one restart file.

Hope that helps.
Courtney

Hi,Courtney！
Thanks for your replying. Sorry, I made a mistake above. My start date is 0001-01-01-00000 instead of 0001-01-02-00000. Experiment A and Experiment B have the same initial fields, but the run length is different.
In Experiment A:
./xmlchange STOP_OPTION=nday
./xmlchange STOP_N=1
But in Experiment B
./xmlchange STOP_OPTION=nhours
./xmlchange STOP_N=6
./xmlchange RESUBMIT=3
You are right, Experiment A will generate 1 restart file and Experiment B will generate 4 restart files, but they will all generate a restart file named *.cam.r.0001-01-02-00000.nc. However, EXPA.cam.r.0001-01-02-00000.nc produced by Experiment A and EXPB.cam.r.0001-01-02-00000.nc produced by Experiment B are different, such as the value of Q. I don't know why.Can you give me some advices?
regards
Joey

 

[peverley]

Hi again Joey,

Couple of questions:
1. What compset/resolution/compiler are you using?
2. Are the differences more than round-off?

Thanks!
Courtney

 

[Joey Wong]

Hi again Joey,

Couple of questions:
1. What compset/resolution/compiler are you using?
2. Are the differences more than round-off?

Thanks!
Courtney

Hi Courtney, sorry for the late reply!
1. My compset is BC5L45BGC ,compset is f09_g16 and runtype is startup.
I just did the following steps:
Experiment A:
./create_newcase --compset BC5L45BGC --res f09_g16 --mach wxy --case ./case/Experiment_A --run-unsupport
cd ./case/Experiment_A
./case.setup
./xmlchange STOP_OPTION=nday
./xmlchange STOP_N=1
./case.build --skip-provenance-check
./case.submit

Experiment B:
./create_newcase --compset BC5L45BGC --res f09_g16 --mach wxy --case ./case/Experiment_B --run-unsupport
cd ./case/Experiment_B
./case.setup
./xmlchange STOP_OPTION=nhours
./xmlchange STOP_N=6
./xmlchange RESUBMIT=3
./case.build --skip-provenance-check
./case.submit
The difference between these two experiments is only the setting of STOP_N and STOP_OPTION, but they are all running for one day.

2. Yes, for example, the difference of variable U in the restart files(*.cam.r.0001-01-02-00000.nc) in these two experiments is up to 10.

 

[Joey Wong]

Hi again Joey,

Couple of questions:
1. What compset/resolution/compiler are you using?
2. Are the differences more than round-off?

Thanks!
Courtney

Also, I wonder if the variables in the restart file are instantaneous values during the run of the model?In other words,are the variables in the *.cam.r.0001-01-02-00000.nc instantaneous values at the moment 0001-01-02-00000?
I tried outputting the instantaneous value at each step,
&cam_history_nl
empty_htapes = .true.
avgflag_pertape = 'I'
nhtfrq = 1
mfilt = 50
fincl1='U','V','T','Q'
I found that the value in restart file is not equal to the instantaneous value at the corresponding moment.
I want to know what data is stored in restart file.
Could you please give me some advice?
Looking forward to your reply.
Thanks!
Joey