Question regarding NATIVE6 and ERA5 variables mapping

[rbeucher]

Hi all,

This question is directed primarily at the @ESMValGroup/technical-lead-development-team, but it may also be beneficial for the broader community.

Regarding the Native6 ERA5 fixes, could you please explain how the mapping from ERA5 variable names to CMIP names is handled? I understand that the data is organized in the Tier3 folder on Levante, using the following directory structure: Tier{tier}/{dataset}/{version}/{frequency}/{short_name}, where short_name refers to the CMIP variable short name.

The reason I ask is that we are currently using a similar directory structure with symlinks pointing to the actual NetCDF files stored in the ERA5 national collection at NCI. This setup introduces some maintenance overhead, as the links need to be updated when new data is added. I'm exploring ways we could directly use the ERA5 collections to streamline this process.

Any insights or suggestions would be greatly appreciated.

Romain

[bouweandela]

Regarding the Native6 ERA5 fixes, could you please explain how the mapping from ERA5 variable names to CMIP names is handled? I understand that the data is organized in the Tier3 folder on Levante, using the following directory structure: Tier{tier}/{dataset}/{version}/{frequency}/{short_name}, where short_name refers to the CMIP variable short name.

There isn't really any mapping or safety mechanism in place, any files found in the folder rootpath/Tier{tier}/{dataset}/{version}/{frequency}/{short_name} are considered to represent the correct ERA5 variable. It is up to the person downloading the data to put the files with the right variable in the right directory. We have fairly extensive documentation available on how the rootpath and drs settings and config-developer.yml work here and here.

To directly use the ERA5 national collection at NCI you could add a an extra entry here

ESMValCore/esmvalcore/config-developer.yml

Lines 100 to 101 in 4c21ca6

	input_dir:
	default: 'Tier{tier}/{dataset}/{version}/{frequency}/{short_name}'

and if you need extra facets that are automatically added you can add those as documented here.

To see an example of how this could be done, you can have a look at #2178.