"grid contains non-ocean point" in aquaplanet simulation using eulerian dycore

[thakur.abubakar]

Dear All,

I'm trying to run a set of experiments using the Eulerian dycore in aquaplanet mode. The model can build the case successfully, and it executes until initialization is complete, after which it throws the error.
ERROR: TPHYSAC error: grid contains non-ocean point
I have attached all the log files.

I've created the case as follows:
./create_newcase --case /mnt/lustre/cas2/casabu/Abu/cesm21/cases/eul0 --compset 2000_CAM60_SLND_SICE_DOCN%AQP3_SROF_SGLC_SWAV --res T85_T85_mg17 --run-unsupported --machine sahasrat

I have modified the T85_T85_mg17 entry in config_grids.xml to null mask (attached). I have also modified the domain files (attached) mentioned in the domain entry for the T85 grid to change all land points to ocean.

I have mentioned the initial condition dataset in user_nl_cam (attached) and modified the said dataset to have 32 vertical levels.

I may be missing something or might be doing something wrong.
Any pointers would be highly appreciated.

Best regards,
Abu Bakar Siddiqui

 

[thakur.abubakar]

Update::

Quite remarkably, the error fixed itself when I created the same case again. The model was becoming unstable and throwing segmentation faults. I managed to fix this too, using eul_divdampn=2 in user_nl_cam.

 

[brianpm]

Curious. Just in case it is of use later, that error comes from the physics package. It arises during energy budget checks. From physpkg:

   1922     if (aqua_planet) then
   1923        labort = .false.
   1924        do i=1,ncol
   1925           if (cam_in%ocnfrac(i) /= 1._r8) then
   1926              labort = .true.
   1927              if (masterproc) write(iulog,*) 'oceanfrac(',i,')=',cam_in%ocnfrac(i)
   1928           end if
   1929        end do
   1930        if (labort) then
   1931           call endrun ('TPHYSAC error: in aquaplanet mode, but grid contains non-ocean point')
   1932        endif
   1933     endif

So the the surface models for some reason returned at least one grid point that had non-ocean fraction.

 

[thakur.abubakar]

Thank you for your response, Brian. I had checked this segment of code and made all the modifications to the domain files. However, the model kept throwing that error (even though I did a fresh build every time I made a change.). When I created a fresh case (using the create_clone command), that problem got resolved. I think the changes I made took effect only in the new case.