CESM1.0.5 POP2 with new grid

[cyoo@ewha_ac_kr]

I have added a new grid to CESM1.0.5 to run a C_ compset and have a trouble. I think that I am quite familiar with procedure as I have done this to CESM1.1 and CCSM4 and they work fine. But since I want to use this particular CESM1.0.5 version, I would appreciate your comments/suggestions to resolve this issue. The error occurs after POP finishes its initialization and it involves invalid tavg id. I have checked tavg_content file many times, but found no clue. Here is the part of the ocn.log file:Many thanks in advance.========================================================================
End of initialization========================================================================
(tavg_requested) id = 0
(tavg_requested) FATAL ERROR: invalid tavg id
------------------------------------------------------------------------POP aborting...
FATAL ERROR: invalid tavg id

 

[jjakacki@iopan_gda_pl]

Based on my expirience, the model is not perfect and not all of the options were tested. I had similar problems and i did not have any choice. I had to find the place where the problem with tavg id exist and the fix it in the code or in the tavg_contents file. POP is able to figure out that tavg_id does not exist, but do not print any part of tavg name (in example vvc). I do not believe it will help, but you will understand the developing level of the code.
Regards,
jaromir

 

[mlevy]

It's difficult to pinpoint exactly where you are running into an error from the information given, but here is one suggestion for you to try (this fix is being implemented in CESM 1.2): In vertical_mix.F90, comment out the "if (implicit_vertical_mix) then" on line 474 and the corresponding "end if" on line 484. The updated block of code should look like

!   if (implicit_vertical_mix) then
     do n = 1,nt
       call define_tavg_field(tavg_DIA_IMPVF_TRACER(n), 'DIA_IMPVF_'     /&
                                       &/ trim(tracer_d(n)%short_name),3, &
          long_name=trim(tracer_d(n)%short_name)                         /&
    &/ ' Flux Across Bottom Face from Diabatic Implicit Vertical Mixing', &
          units=trim(tracer_d(n)%flux_units), grid_loc='3113',            &
          scale_factor=tracer_d(n)%scale_factor,                          &
          coordinates='TLONG TLAT z_w_bot time')
     enddo
!   endif

By default

implicit_vertical_mix

is set to true, so this will only effect your run if you manually set it to false. Similarly, if you changed

vmix_choice

from 'kpp' to something else, there is another set of changes that may help. If you are still having issues, could you post more information about your setup? If you don't mind sharing your grid and what you used with "create_newcase", I can try to recreate your problem on the NCAR system to help debug.

~Mike