CLM output restart file missing

[syang]

Dear all,

I successfully run the CLM5 case I2000Clm50BgcCruGs and the output files included "hist" files (***.clm.h0.***2550***.nc) and "restart" files (***.clm.r.nc). However, I re-setup the case and would like to do spinups for 100 years. This time no "restart" files (***.clm.r.nc) coming out, only with "***.datm.rs1.**.bin file". I have no idea why does this situation happen? And how can I get the restart files so I can use it as initial conditions. THANKS!!!

 

[oleson]

It looks like your restart file is there:

Heihe_005x005_OWNFC_500spinup.clm2.r.2550-01-01-00000.nc

 

[erik]

Hopefully, this is resolved, but I'm moving it to the CLM forum.

 

[syang]

It looks like your restart file is there:

Heihe_005x005_OWNFC_500spinup.clm2.r.2550-01-01-00000.nc

Thanks for your reply. "Heihe_005x005_OWNFC_500spinup.clm2.r.2550-01-01-00000.nc" was the previous restart file. The new outputs yield no restart files. Then, I re-created a I2000Clm50BgcCruGs case. This time there are restart files now! So, I think it could be a bug or something wrong with my old case. Anyway, it is solved now! THANKS!

 

[sowon@postech_ac_kr]

Hopefully, this is resolved, but I'm moving it to the CLM forum.

Dear Erik and all,

I'm struggling due to this problem (CESM2.2.0).
Before running the IHISTClm50Bgc, I run the I1850Clm50*(I1850Clm50Bgc, I1850Clm50BgcCrop, I1850Clm50BgcCropCru) compsets for spin-up.
However, there is no restart file after running these compsets (I1850Clm50*).
Also, I got the error when running I1850Clm50BgcSpinup.

How can I solve this problem?
Thank you in advance

 

[oleson]

Is the model completing successfully? Can you attach the CaseStatus file that is in your case directory?

 

[sowon@postech_ac_kr]

Is the model completing successfully? Can you attach the CaseStatus file that is in your case directory?

I1850Clm50Bgc, I1850ClmBgcCrop (also I1850ClmBgcCropCru case) compsets: success, Only historical files are created...
I1850ClmBgc.Spinup: failed.

What's the difference btw. I1850Clm50Bgc and I1850Clm50Bgc_Spinup?

Thank you for your help in advance.

 

[sowon@postech_ac_kr]

Is the model completing successfully? Can you attach the CaseStatus file that is in your case directory?

I also attached log file (I1850Clm50BgcSpinup).
Thank you in advance.

 

[oleson]

That compset (I1850Clm50BgcSpinup) is definitely not working out of the box with release-cesm2.2.0.
That compset is designed to use "CPLHIST" atmospheric forcing, which is generally forcing from a coupled model simulation, to spinup CLM for a coupled model historical simulation. The compset comes out of the box with "UNSET" for coupler history xml variables (e.g., DATM_CPLHIST_DIR). That's why you are getting an error. You would have to specify these variables for this to work.
However, if you are simply running a historical simulation out of the box then there will be a spunup initial file provided by that case (specified by finidat in lnd_in).
If you making changes to the model and want to spinup the model prior to running a historical simulation then I would recommend following the instructions in the User's Guide instead of using that compset:

1.5.5. Spinup of CLM5.0-BGC-Crop — ctsm release-clm5.0 documentation

Basically, the spinup is in two steps, an accelerated decomposition (AD) step and a post-AD step. The instructions specify the I1850Clm50BgcCropCru compset but you could also use the I1850Clm50Bgc or I1850ClmBgcCrop compsets depending on your choice of atmospheric forcing or model configuration.

 

[wvsi3w]

I1850Clm50Bgc, I1850ClmBgcCrop (also I1850ClmBgcCropCru case) compsets: success, Only historical files are created...
I1850ClmBgc.Spinup: failed.

What's the difference btw. I1850Clm50Bgc and I1850Clm50Bgc_Spinup?

Thank you for your help in advance.

Hello,
I was wondering if you have used CESM 2.1.3 and if yes, could you share the restart file of your AD spin-up for IHistClm50BgcCrop compset??? and if you by any chance have done this what is the soil layer structure that you have used?

I am taking this opportunity to ask all the members and fellow researchers of the CESM forum (@oleson @slevis ) that how can we get access to these restart files. because it is time-consuming and lengthy for certain people to do AD spin up (like me) and by sharing these files we can save a lot of time.

is there a link or somewhere that I am missing out?

 

[oleson]

If you create a case using IHistClm50BgcCrop at f09 using release-cesm2.1.5 you will get a spunup initial file by default, as specified by this in your lnd_in:

finidat = '/glade/campaign/cesm/cesmdata/inputdata/lnd/clm2/initdata_map/clmi.I1850Clm50BgcCrop.1366-01-01.0.9x1.25_gx1v6_simyr1850_c171213.nc'

 

[wvsi3w]

If you create a case using IHistClm50BgcCrop at f09 using release-cesm2.1.5 you will get a spunup initial file by default, as specified by this in your lnd_in:

finidat = '/glade/campaign/cesm/cesmdata/inputdata/lnd/clm2/initdata_map/clmi.I1850Clm50BgcCrop.1366-01-01.0.9x1.25_gx1v6_simyr1850_c171213.nc'

Dear Keith,
Thank you for your response.

Q1- What about its soil layer structure? how can I find out what soil layer it has used?

Q2- how about other releases of cesm? like CESM2.1.3 that I am using and this is what I have found in its lnd_in file:
finidat = '/inputdata/lnd/clm2/initdata_map/clmi.I1850Clm50BgcCrop.1366-01-01.0.9x1.25_gx1v6_simyr1850_c171213.nc'
It seems like it is the same file as you mentioned.

Q3- So I am guessing this spunup file was using the forcing of the year 1366 (?). how could I find more information about this spunup file? I mean the reason we do AD spin up is to reach steady state for the desired variables (in this case of Bgc it would need 1000 year of AD and 200 years of ND spin up as suggested before). So how can I be sure that this spunup file you mentioned has reached equilibrium for Bgc variables?

Thanks for your time

 

[oleson]

1. The initial files provided are for the default configuration of the model. For this version of the model, the default soil layer structure is 20SL_8.5m.
2. release-cesm2.1.5 is simply a software update to release-cesm2.1.3, mainly for the purpose of being able to run on our HPC, Derecho. That's why the provided initial file is the same.
3. 1366 simply refers to the number of years run in the last stage of the spinup. The simyr1850 refers to the actual year the initial file is valid for. So this file is appropriate for initializing a historical run for 1850-... Our initial files that we provide have reached equilibrium for BGC variables.

 

[wvsi3w]

1. The initial files provided are for the default configuration of the model. For this version of the model, the default soil layer structure is 20SL_8.5m.
2. release-cesm2.1.5 is simply a software update to release-cesm2.1.3, mainly for the purpose of being able to run on our HPC, Derecho. That's why the provided initial file is the same.
3. 1366 simply refers to the number of years run in the last stage of the spinup. The simyr1850 refers to the actual year the initial file is valid for. So this file is appropriate for initializing a historical run for 1850-... Our initial files that we provide have reached equilibrium for BGC variables.

Thank you Keith for your thorough answer.
So it took 1366 years for the spinup of that simulation (I guess 1000y AD + 366y ND?) using the 20SL_8.5m soil layer. and in case I want to start my transient simulation from 1850 onward I can use that file. But if I want to change anything like the forcing start and end date or lets say soil layer structure or anything else this file is not useful for me and I should conduct my own AD spinup (almost 1200y of simulation). Am I right?

I am referring to this part of documentation regarding my question above:

1.5.5. Spinup of CLM5.0-BGC-Crop — ctsm release-clm5.0 documentation

 

[wvsi3w]

Oh one more thing, in case I want to run my transient simulation with that finidat file then are there any restrictions or guidelines to follow? like:
there is no need to do spinup then "CLM_ACCELERATED_SPINUP" shouldn't be touched, and "CLM_FORCE_COLDSTART" should be on? how about "DATM_CLMNCEP_YR_START and END" should I change them or "RUN_TYPE, or RUN_STARTDATE or STOP_DATE"?
I remember the forcing was "on" even when I was not running any spinup job, so maybe I should turn anything related to spinup off when I am using that spunup finidat file?
The STOP_DATE I think is not necessary since it says in the compset definition that "1850: Pre-Industrial; 2000 present day", but is 2000 the end year or is it 2015 like some other compsets that I have read on the Forum?

 

[oleson]

No, as I said, it took 1366 years for the last phase of the spinup (ND).
Yes, generally you should do your own spinup if you make significant changes to the model. In some cases, you may be able to shorten the spinup by using that file to start an AD spinup instead of a cold start.

 

[wvsi3w]

No, as I said, it took 1366 years for the last phase of the spinup (ND).
Yes, generally you should do your own spinup if you make significant changes to the model. In some cases, you may be able to shorten the spinup by using that file to start an AD spinup instead of a cold start.

Thank you very much for the hint.

It will certainly be helpful if I need to do these runs in the future. However, for now, I have to proceed with my own AD + ND spinup (which is currently running). In my case, the surface temperature needs more time to reach the modified depth than the typical ~150 years (1850 onward). After completing a 1000-year AD and 200+ year ND spinup, I plan to run ~1000 years of transient simulations using NorESM data (capturing the last 1000 years of high and low variability, from around 1015 to 2015).

This leads me to wonder about something:

Is it possible to run transient simulations with NorESM data (or any similar dataset) for more than 1000 years using CLM5? I understand we can use custom forcing data, but that forcing is usually 1 year or 10 years or something like that and the purpose of that is to reach equilibrium but if I set up the forcing data in CLM5 to cover a different period, say from 1015 to 2015 (in case of using NorESM data), does that effectively change the start year in the compset from 1850 to 1015? because it should and the whole purpose of my transient run is to use that >1000y of data to simulate the land and to see the impacts of surface temperature reaching that depth I modified (~500m).

I got very nervous thinking about it so I hope my question is clear – sorry if my wording is a bit unsophisticated.

I should mention that my current AD and ND spinup runs are not using any other forcing data, they are using the model's default forcing

 

[wvsi3w]

Thank you very much for the hint.

It will certainly be helpful if I need to do these runs in the future. However, for now, I have to proceed with my own AD + ND spinup (which is currently running). In my case, the surface temperature needs more time to reach the modified depth than the typical ~150 years (1850 onward). After completing a 1000-year AD and 200+ year ND spinup, I plan to run ~1000 years of transient simulations using NorESM data (capturing the last 1000 years of high and low variability, from around 1015 to 2015).

This leads me to wonder about something:

Is it possible to run transient simulations with NorESM data (or any similar dataset) for more than 1000 years using CLM5? I understand we can use custom forcing data, but that forcing is usually 1 year or 10 years or something like that and the purpose of that is to reach equilibrium but if I set up the forcing data in CLM5 to cover a different period, say from 1015 to 2015 (in case of using NorESM data), does that effectively change the start year in the compset from 1850 to 1015? because it should and the whole purpose of my transient run is to use that >1000y of data to simulate the land and to see the impacts of surface temperature reaching that depth I modified (~500m).

I got very nervous thinking about it so I hope my question is clear – sorry if my wording is a bit unsophisticated.

I should mention that my current AD and ND spinup runs are not using any other forcing data, they are using the model's default forcing

please ignore or delete this message of mine. thanks