Changing output frequency for a run continuation?

Hi there,

I've run a cesm2.1.3 simulation for a number of years (compost = 1850_CAM40_CLM40%CN_CICE_DOCN%SOM_RTM_SGLC_SWAV). I want to continue the simulation, but change the output frequency for certain variables.

In particular, I modifying user_nl_cam as follows:

FINCL2 = 'DEI_rad', 'MU_rad', 'LAMBDAC_rad', 'ICIWP_rad', 'ICLWP_rad', 'DES_rad', 'ICSWP_rad', 'CLD_rad', 'CLDFSNOW_rad' nhtfrq = 0,-1 mfilt = 1,24 avgflag_pertape = 'A','I' ndens = 2,2

I then ran ./preview_namelists, and this updated ./run/atm_in as intended. But when I continued the run (submitted to queue using ./case.submit), the output frequency remained the same (i.e. was not modified as intended).

Any idea why the model is not following the data output specifications in atm_in?

Cheers,

Mike

Hi Mike,

Unfortunately, CESM does not allow changes to history streams during a continue run. To make changes to history streams, you will need to make a branch. The easiest way to do this is to use "create_clone" to clone your case into a new case, make your history changes in the new case, and then change the xml variable RUN_TYPE to "branch." Also, CONTINUE_RUN should be FALSE, and your REF_CASE should be your previous case. This will continue your previous run on as a branch with new history streams. There's more information on branch runs here: 5. Running a Case — CIME cime5.6 documentation

katec said:

Hi Mike,

Unfortunately, CESM does not allow changes to history streams during a continue run. To make changes to history streams, you will need to make a branch. The easiest way to do this is to use "create_clone" to clone your case into a new case, make your history changes in the new case, and then change the xml variable RUN_TYPE to "branch." Also, CONTINUE_RUN should be FALSE, and your REF_CASE should be your previous case. This will continue your previous run on as a branch with new history streams. There's more information on branch runs here: 5. Running a Case — CIME cime5.6 documentation

Click to expand...

Thanks @katec, that's super clear!

katec said:

Hi Mike,

Unfortunately, CESM does not allow changes to history streams during a continue run. To make changes to history streams, you will need to make a branch. The easiest way to do this is to use "create_clone" to clone your case into a new case, make your history changes in the new case, and then change the xml variable RUN_TYPE to "branch." Also, CONTINUE_RUN should be FALSE, and your REF_CASE should be your previous case. This will continue your previous run on as a branch with new history streams. There's more information on branch runs here: 5. Running a Case — CIME cime5.6 documentation

Click to expand...

Hii katec,
I am a new user of CESM and I also want to change my output frequency of history file in the middle on the run. I did similar changes as suggested in your answer but i got an error:
ERROR: Unknown error submitted to shr_abort_abort.
Please suggest the solution.
I am attaching the cesm log file.
Thank you.

Hi,
So, the log that you attached here has an error very early in CIME setup. It is very difficult to determine what went wrong from just this one log but I don't think this is due to changing history fields. There is something else setup wrong in your branch run. Perhaps looking at your cpl.log file will give a better idea of what is missing or incorrect. If you want more direct help with this error, then this post describes the kind of information we need to help you: Information to include in help requests

katec said:

Hi,
So, the log that you attached here has an error very early in CIME setup. It is very difficult to determine what went wrong from just this one log but I don't think this is due to changing history fields. There is something else setup wrong in your branch run. Perhaps looking at your cpl.log file will give a better idea of what is missing or incorrect. If you want more direct help with this error, then this post describes the kind of information we need to help you: Information to include in help requests

Click to expand...

Hii,
I checked the atm log file and find out that i did not change the ref. date in env_run.xml file. After changing it, Model ran successfully.
Thank you.

katec said:

Hi Mike,

Unfortunately, CESM does not allow changes to history streams during a continue run. To make changes to history streams, you will need to make a branch. The easiest way to do this is to use "create_clone" to clone your case into a new case, make your history changes in the new case, and then change the xml variable RUN_TYPE to "branch." Also, CONTINUE_RUN should be FALSE, and your REF_CASE should be your previous case. This will continue your previous run on as a branch with new history streams. There's more information on branch runs here: 5. Running a Case — CIME cime5.6 documentation

Click to expand...

Hi Katec,
In the link you shared, it says:

"Under this scenario, the new case is able to produce an exact bit-for-bit restart in the same manner as a continuation run if no source code or component namelist inputs are modified."

Click to expand...

Does this mean that if I don’t change any code or namelist settings, a “continue” run and a “branch” run should behave exactly the same?

I am trying to run an Antarctic sea ice forcing experiment using fully coupled CESM2.1.5. After running for over 30 years, the model crashed with an error in the ocean module:

(Task 197, block 1) Message from (lon, lat) ( 91.062, 1.740), which is global (i,j) (117, 194). Level: 14
(Task 197, block 1) MARBL WARNING (marbl_interior_tendency_mod:compute_large_detritus_prod): dz*DOP_loss_P_bal= 0.321E-012 exceeds Jint_Ptot_thres= 0.271E-013
(Task 238, block 1) Message from (lon, lat) ( 286.826, 12.523), which is global (i,j) (291, 235). Level: 14
(Task 238, block 1) MARBL WARNING (marbl_interior_tendency_mod:compute_large_detritus_prod): dz*DOP_loss_P_bal= 0.607E-012 exceeds Jint_Ptot_thres= 0.271E-013
(Task 30, block 1) Message from (lon, lat) ( 294.688, -65.331), which is global (i,j) (298, 27). Level: 1
(Task 30, block 1) MARBL ERROR (marbl_diagnostics_mod:store_diagnostics_iron_fluxes): abs(Jint_Fetot)= 0.347E-016 exceeds Jint_Fetot_thres= 0.951E-017
(Task 30, block 1) MARBL ERROR (marbl_diagnostics_mod:marbl_diagnostics_interior_tendency_compute): Error reported from store_diagnostics_iron_fluxes
(Task 30, block 1) MARBL ERROR (marbl_interior_tendency_mod:marbl_interior_tendency_compute): Error reported from marbl_diagnostics_interior_tendency_compute()
(Task 30, block 1) MARBL ERROR (marbl_interface:interior_tendency_compute): Error reported from marbl_interior_tendency_compute()
(Task 30, block 1) MARBL ERROR (ecosys_driver:ecosys_driver_set_interior): Error reported from marbl_instances(1)%set_interior_forcing()
ERROR reported from MARBL library

Click to expand...

When I tried restarting with the “continue” option using the latest restart files, the same error happened again. But when I did a “branch” run from the same restart point (without changing the code and namelist, just a new case name), the model continued running fine. I’m wondering whether the outputs from this “branch” run are valid for analysis together with the original run? Any suggestions would be very helpful. Thanks a lot!

Best regards,
Zhu Zhu

​

Zhu Zhu said:

Does this mean that if I don’t change any code or namelist settings, a “continue” run and a “branch” run should behave exactly the same?

Click to expand...

Yes, and I suggest that you create a separate test-case to confirm this statement beyond doubt. Of course this suggests that something changed between your "continue" and your branch or that the branch is not being started correctly.