search

plesager / ece3-postproc

Suite of processing tools for EC-Earth3 output
5 stars 8 forks source link

error message when hireclim2 computes Nino3.4 SST #1

Closed plesager closed 6 years ago

plesager commented 7 years ago

at high res (ORCA025) the nemo_post.sh log reports:

##[16:24:52] master_hiresclim.sh[189]> /usr/local/apps/cdftools/3.0/bin/cdfmean aGez_1m_19890101_19891231_grid_T.nc sosstsst T 470 668 519 479 0 0
 npiglo =          199
 npjglo =          -39
 npk    =           75
 npt    =           12
 nvpk   =            1
 depth dim name is deptht
  Problem in getvar for e1t
 ERROR in NETCDF routine, status=         -57
 NetCDF: Start+count exceeds dimension bound

Note that that does not prevent the code to write a EXP_YEAR_sosstsst_nino34.nc file.

oloapinivad commented 6 years ago

We saw the same issue a few ago, I think it is related to a mismatch between the points provided to compute the ENSO region and the ORCA025 grid for NEMO 3.6 that is different - as long as I understand - to the older one (that was used in older EC-Earth versions).

Now that we have Barakuda operational I think it is not very important, but still it would be nice to fix it.

abalanzino commented 6 years ago

Hi Philippe,

as wrote Paolo, I obtained new Nino3.4 coordinates using Barakuda operational (I added a print in one of the python modules). So the code line ( ORCA025L75 ) in nemo_post.sh should be:

Nino3.4 region (in ij coordinates) nino34_region='469 669 507 547'

etiennesky commented 6 years ago

I had a problem in my installation with the Nino3.4 coordinates of the ORCA1 grid, am I the only one?

abalanzino commented 6 years ago

The correct Nino3.4 coordinates of the ORCA1 grid should be: Nino3.4 region (in ij coordinates) nino34_region='119 168 133 161'

etiennesky commented 6 years ago

Great thanks! right now the coordinates are the same as for the ORCA025 grid.

nino34_region='470 668 519 479'

abalanzino commented 6 years ago

Yes, there was a mistake! Moreover the coordinates (ORCA025) nino34_region='470 668 519 479' relate to old mesh.nc and mask.nc, when the grid was 1442x1021 (now the grid for nemo3.6 is 1442x1050 and the coordinates are '469 669 507 547').

plesager commented 6 years ago

Great! It may be worth documenting somewhere in the README how to retrieve these coordinates if someone changes the mesh grid. Actually I was wondering if we could make that automatic?

abalanzino commented 6 years ago

I think it is possible documenting in the README. We still haven't thought about how to make that automatic, but we could work on that one.

plesager commented 6 years ago

Thank you @abalanzino, I put your fixes in 5cf743d.