search

plesager / ece3-postproc

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

MOC requires depthv in place of olevel #35

Open plesager opened 6 years ago

plesager commented 6 years ago

Hiresclim2 computes MOC by default now. But with EC-Earth 3.2.3 the level dimension in the grid_V file is named olevel in place of the expected 'depthv`. I pushed a fix that works on cca@ECMWF: a73e9cd0d8b722892648ed6359b75850ab523c4b. Will test on rhino@KNMI.

plesager commented 6 years ago

Works also on rhino@KNMI.

rfbilbao commented 6 years ago

Hi @plesager

Here at BSC we ran into the same issue and solved it in a different way. We generated a CDFtools namelist and renamed the variable 'deptht' to 'olevel', then we export the namelist:

export NAM_CDF_NAMES='path'/spinup.namlist

This worked on MN4 with CDFtools version 3.0.1.

Using a namelist with the variable names of the nemo output, which could be generated in the script, would not require to change the variable names in the temporary files. At BSC we can test this and try to remove all the ncrename commands, which could save time.

Thanks,

Roberto spinup.namlist.txt

@aearamos @jvegasbsc @etiennesky

etiennesky commented 6 years ago

Hi @plesager I think it would be much easier to replace the various band-aid fixes we have been using by using cdftools namelists.

The idea would be to set the grid and variable names in the platform config file (or something more flexible which can be easily scripted), and generate a cdftool namelist from this info. What do you think?

plesager commented 6 years ago

I was not aware of that namelist. Sounds like a really good idea, it would tidy up the code. But how do we manage the different versions of that namelist. How many do we need? Two? PRIMAVERA and 3.2.3 ?

plesager commented 6 years ago

Can this be tested in its own branch?

etiennesky commented 6 years ago

I think that a namelist generated on-the-fly would be the most flexible option, but initially we could have 3 namelists (primavera, 3.2.3, 3.2.2). This would be easier to do and might be good enough.