Recipe `recipe_perfmetrics_land_CMIP5.yml` fails because of wrong dataset name `CERES-EBAF_Surface`

[jlenh]

Describe the bug
Running recipe_perfmetrics_land_CMIP5.yml results in an error when looking for the dataset CERES-EBAF_Surface which is named CERES-EBAF in cmor_config.

I am wondering, following #3325, if the current CMORizer should be updated to fit CERES-EBAF_Surface or if the recipe should be adapted here. If somebody (e.g. @bouweandela ?) has more information especially regarding obs4MIPs standard names?

[bouweandela]

The data from the obs4MIPs project is already CMORized and available from ESGF. The CMORizer scripts are used to CMORize data for the ESMValTool specific OBS and OBS6 projects, so not related to obs4MIPs data (yet). There is hope that in the future we will be able to contribute CMORized datasets to obs4MIPs that has been CMORized with our CMORizer scripts (cf #3960), but this is still to be discussed with the obs4MIPs community.

[jlenh]

Ah I understand, it is not referring to the observational dataset produced by cmorization in ESMValTool.

Would the solution here then be to update in the recipe CERES-EBAF_Surface to CERES-EBAF?
On DKRZ/Levante the dataset is available at /work/bd0854/DATA/ESMValTool2/OBS/Tier1/CERES-EBAF/.

[bouweandela]

We prefer to use publicly available data in recipes, so if the data is available from the obs4MIPs project on ESGF that is preferred. What is the problem with using the data from ESGF?

[jlenh]

I think the dataset entry needs to be updated following another example like here so it is looking for that data in the right place:

ESMValTool/esmvaltool/recipes/ipccwg1ar6ch3/recipe_ipccwg1ar6ch3_fig_3_42_b.yml

Lines 821 to 823 in 5dae745

	# OBS
	- {dataset: CERES-EBAF, project: obs4MIPs, level: L3B,
	version: Ed2-8, tier: 1, start_year: 2001, end_year: 2005}

Not all variable seem to be in the folder /work/bd0854/DATA/ESMValTool2/download/obs4MIPs/CERES-EBAF/, some are in /work/bd0854/DATA/ESMValTool2/OBS/Tier1/CERES-EBAF/.

[bouweandela]

You can use the options search_esgf: always or search_esgf: when_missing to automatically download the required data. I ran the recipe on Levante with that option and the data is now available in /work/bd0854/DATA/ESMValTool2/download/obs4MIPs/CERES-EBAF_Surface/.

[jlenh]

Hmm I had the settings for DKRZ/Levante uncommented so it should have tried to download 🤔

Thanks for downloading the data and running the recipe, guess it was an issue on my side after all :)