[v2] Interpolation onto the NEMOMED8 curvilinear grid

Hello everyone,

I am looking to interpolate LMDZ-regional output onto the grid of the Mediterranean regional version of NEMO (NEMOMED8) using CDO and bilinear interpolation.
I found this well detailed procedure on the NEMO wiki:
https://forge.ipsl.jussieu.fr/nemo/wiki/Users/SetupNewConfiguration/cdo-interpolation
I followed the procedure and the result looks pretty good on the Mediterranean Sea and the near Atlantic zone (attached picture). However, as you can see, there are some issues on the north-west African cost and north-west Spain. By masking with LMDZ land-sea mask before interpolation, it seems that there is a problem with longitude and latitude coordinates on these regions.

Do you have any clue to improve the interpolation? Though it’s not a big deal since it looks OK on the Med Sea.

Thanks in advance
Tristan

PS: I can provide the files if needed

Hi Tristan,

Is it possible that the NEMO (destination) lat/lon fed to CDO are coming from a file (output or mesh_mask) featuring land suppression?

Best
Charles

Hi Charles,

Thanks for your reply. Yes indeed lat/lon comes from the mesh_mask of NEMOMED8 file but I am not sure what you are referring to with “featuring land suppression”, could you explain me ?

Best
Tristan

If the mesh_mask file has been generated from a run featuring land suppression, see Fig 8.6 on

Then the lat/lon are not meaningful over land subdomains, and interpolations can go wrong. Could you share that mesh_mask file?

Sure, here it is:

Tristan

Your mesh mask does not feature land suppression, so that is not the problem. I am not familiar with that grid, but it kind of folds over land around the Gibraltar strait, which I suspect confuses CDO.

It may be worth trying to remap taking into account a destination mask

It is not an ideal fix, because then your interpolation is mask-dependent (i.e., you would need to re-interpolate if the bathymetry or rn_hmin change), but that is the best I can come up with.

I tried with the destination mask method and it looks better (attached figure) ! (though it was a bit tricky to do it, I am thinking to make a tutorial for the potential user).

Thanks for the great help Charles. I take not in case of bathymetry or rn_hmin change.

Best
Tristan

1 Like

Yes it would be very helpfull.