Dataset problem: ACCESS-CM2 is stored under CSIRO-ARCCSS, not CSIRO-ARCCSS-BoM

[ledm]

The CMIP6 model dataset for the model ACCESS-CM2 is stored on jasmin under the institude ID: CSIRO-ARCCSS, but recorded as CSIRO-ARCCSS-BoM in ESMValCore in CMIP6_CV.json.

ie the paths to this model are:

/badc/cmip6/data/CMIP6/CMIP/CSIRO-ARCCSS

However, in esmvalcore/cmor/tables/cmip6/Tables/CMIP6_CV.json, we record the institude ID for this model to be :

        "source_id":{
            "ACCESS-CM2":{
                "activity_participation":[
                    "CMIP",
                    "FAFMIP",
                    "OMIP",
                    "RFMIP",
                    "ScenarioMIP"
                ],
                "cohort":[
                    "Registered"
                ],
                "institution_id":[
                    "CSIRO-ARCCSS-BoM"
                ],
                "source_id":"ACCESS-CM2",
                "source":"ACCESS-CM2 (2018): \naerosol: UKCA-GLOMAP-mode\natmos: MetUM-HadGEM3-GA7.1 (N96; 192 x 144 longitude/latitude; 85 levels; top level 85 km)\natmosChem: none\nland: CABLE2.3.5\nlandIce: none\nocean: ACCESS-OM2 (GFDL-MOM5, tripolar primarily 1deg; 360 x 300 longitude/latitude; 50 levels; top grid cell 0-10 m)\nocnBgchem: none\nseaIce: CICE5.1 (same grid as ocean)"
            },

So ESMValTool is looking for this dataset in the non-existant folder:

/badc/cmip6/data/CMIP6/CMIP/CSIRO-ARCCSS-BoM/ACCESS-CM2/
``

[zklaus]

This was changed 2019-11-15 in the tables. The original change in the CV was on 2019-10-22 and released as 6.2.39.2.

[valeriupredoi]

we can probably fix this with a temporary mapping in config-developer.yml but really, it's a bad case of old ESGF data structure - I will email @agstephens and Ruth Petrie about it

[valeriupredoi]

OK email sent to Ag and Ruth (and you two troublemakers too) :beer:

[valeriupredoi]

OK email sent to Ag and Ruth (and you two troublemakers too) beer

nevermind, I'm being silly - yes, the (new) table is correct and in line with the Jasmin path :beer:

[bouweandela]

As a workaround for this issue (once there is a new release of the CMIP6 CMOR tables and we've updated our copy to use it) you can specify the correct institute in the dataset definition in the recipe.

[bouweandela]

We are now using the latest release of the CMOR tables (6.9.33),