-
Notifications
You must be signed in to change notification settings - Fork 0
Encountered issues
Running recipe_radiation_budget.yml with CNRM-CM6-1 dataset instead of the given ones produces the following error:
ERROR esmvalcore.preprocessor:338 Failed to run preprocessor function 'fix_metadata' on the data
esmvalcore.cmor.check.CMORCheckError: There were errors in variable rls:
rls: positive should be down, not up
The same error occurs for CMIP6 and EPOC data of the same model. But only on the rls variable. The problem with these data is that the ‘positive’ attribute should be in the current CMIP DRS set to ‘down’ instead of ‘up’. Since rls (short for ‘Net Longwave Surface Radiation’) is a field variable, ‘up’ and ‘down’ defines the z-axis orientation for the radiative fluxes. This error could be solved by CMORizing the data. In view of the fact that all CNRM-CERFACS production is like that, we asked the ESMValTool support for help. A shortcut for the problem is to set the check level to relaxed.
esmvaltool run recipe_radiation_budget.yml --check_level relaxed
When trying to pass CMIP6 data to a recipe written for cmip5 data we can encounter the problem of changed DRS. This is for example the case for the arctic ocean recipe. Here the areacello variable can be found in the fx mip for cmip5 but for CMIP6 it is stored in Ofx. To find these differences between CMIP DRS having a look at CMIP Data Request makes things easier. The related error message looks like this:
ERROR esmvalcore._recipe:1292 Unable to load CMOR table (project) 'CMIP6' for variable 'areacello' with mip 'fx'
Running recipe_monitor.yml on CNRM-CM6-1 did produce this error:
ERROR esmvalcore.preprocessor:338 Failed to run preprocessor function 'mask_landsea' on the data
[<iris 'Cube' of sea_surface_temperature / (degC) (time: 180; -- : 294; -- : 362)>]
loaded from original input file(s)
['/data/scratch/globc/dcom/CMIP6/CMIP/CNRM-CERFACS/CNRM-CM6-1/historical/r1i1p1f2/Omon/tos/gn/latest/tos_Omon_CNRM-CM6-1_historical_r1i1p1f2_gn_185001-201412.nc']
with function argument(s)
mask_out = 'land'
…
ValueError: Use of shapefiles with irregular grids not yet implemented, land-sea mask not applied.
Opened a GitHub discussion on this error. The problem here is related to not-available model data : CNRM-CM6-1 model we use did not produced (for none of the experiments) the land/sea fraction on the ocean grid (“sftof_Ofx”).
A quite usually appearing error when working with observation data or cmip5 data is that the rootpath isn’t correctly defined in the config-user file. This problem produces for example such an error message:
ERROR esmvalcore._recipe_checks:110 Looked for files matching: ESGF:/thetao_Omon_MPI-ESM-LR_historical_r1i1p1_gn*.nc
This error just points out that there is no rootpath matching the data. For fixing this just search the file in your expected directory and compare it with the definition in the config-user file. In this case, the config-user file defined the rootpath ending with /CMIP5 but to match with the actual directory it needed to end with /cmip5.