Energy conservation error and advective heat flux to ocean

[hshin74]

Hello. I have been running CESM2.1.3 with B1850 compset but reset to year 2000 on a large computing system of Korea Meteorological Administration. I have encountered the same problem mentioned in a previous message (ERROR: ice: Vertical thermo error) twice. At the first time the model run stopped after simulating 50 years, I resolved the problem by changing a clubb_gamma coefficient a bit as suggested on the board. But the model stopped again after simulating 190 years. This time, I am looking into the source code and tracking down the energy terms.

Among the terms, I found out that the advective heat flux to ocean, fadvocn, is initialized to zero and then not updated with POP or CICE source codes, but that the subroutine conservation_check_vthermo in the ice_therm_vertical.F90 module printed an error message with a value of -570.xxx for fadvocn when the model stopped recently. I wonder where the actual value of fadvocn is calculated/updated after the initialization. I attached the cesm.log file to this message. Any help will be greatly appreciated. Thanks in advance!

 

[dbailey]

The fadvocn should not be the problem. We do just get intermittent thermodynamic errors. The flux conservation error here is very small and fits into the category of point 5 here:

FAQ: CICE Thermodynamic convergence errors.

Sometimes the CICE model produces an error like: Thermo iteration does not converge istep1, my_task, i, j: .... (a number of messages and details) (shr_sys_abort) ERROR: ice: Vertical thermo error This is normally indicative of a problem...

bb.cgd.ucar.edu

 

[hshin74]

The fadvocn should not be the problem. We do just get intermittent thermodynamic errors. The flux conservation error here is very small and fits into the category of point 5 here:

FAQ: CICE Thermodynamic convergence errors.

Sometimes the CICE model produces an error like: Thermo iteration does not converge istep1, my_task, i, j: .... (a number of messages and details) (shr_sys_abort) ERROR: ice: Vertical thermo error This is normally indicative of a problem...

bb.cgd.ucar.edu

Thank you for your comment! Owing to your detailed explanation in the previous message, I could get the clue for my stopped simulation.

In my inquiry, I put the background of my current situation, and my true question was intended to be the advective heat flux where the non zero value came from. I apologize if I made my inquiry misleading, or if my English still do not make sense.

Based on my twenty years of experience in climate modeling, I have some doubt that it could be a problem if a variable prints out unintended output. And I wondered if 1e-3 is small enough so that it can be ignored when the variables were set in double precision. I am not pointing out these can be a problem affecting the scientific result. My concern is that I I just want to make sure if I understand the code correctly.

Thank you again. I appreciate your kind guidance.

 

[dbailey]

We typically use the threshold of 1.0e-3. So, even if it is slightly above this it is fine. There are roundoff issues in temperature and energy iteration in the thermodynamics that can lead to this.