When creating approved variable lists in QC, if a variable has multiple modelling realms (e.g. sithick_tavg-u-hxy-si with seaIce ocean (see UKNCSP/CDDS-CMIP7-mappings#1692)) both realms are used. This produces a line in the approved variables file that looks like this:
seaIce ocean@day/sithick_tavg-u-hxy-si;<filepath>
as oppose to the expected:
seaIce@day/sithick_tavg-u-hxy-si;<filepath>.
This issue silently fails in archive meaning that there is a discrepancy between the variables produced and the number archived to mass.
A change should be made so that, when there are multiple realms given, the code uses the "primary" realm i.e. the first realm written. It may also be valuable to add a flag into archive to trigger a critical error if it identifies this sort of issue to avoid silent failures.
When creating approved variable lists in QC, if a variable has multiple modelling realms (e.g.
sithick_tavg-u-hxy-siwithseaIce ocean(see UKNCSP/CDDS-CMIP7-mappings#1692)) both realms are used. This produces a line in the approved variables file that looks like this:seaIce ocean@day/sithick_tavg-u-hxy-si;<filepath>as oppose to the expected:
seaIce@day/sithick_tavg-u-hxy-si;<filepath>.This issue silently fails in archive meaning that there is a discrepancy between the variables produced and the number archived to mass.
A change should be made so that, when there are multiple realms given, the code uses the "primary" realm i.e. the first realm written. It may also be valuable to add a flag into archive to trigger a critical error if it identifies this sort of issue to avoid silent failures.