[CF-metadata] proposed changes to various standard names
j.m.gregory at reading.ac.uk
Tue Jun 1 06:25:04 MDT 2010
> > It would be unprecedented to create an alias with two translations, I think. If that's OK, I agree it is a logical solution.
> Can we clarify the meaning of 'alias' again? To me it has value as an unambiguous mapping. This is clearly not in that category. Even without the alias, your statement is still true (future data will be unambiguous but no penalty is imposed on older data). I'd like to think use of the older term could be deprecated, but it would stay in the list.
Thanks for pointing out my thinking is unclear! You are right, since we don't
know for sure the intention of existing data, we should allow the existing name
to have either interpretation. In this case, we have to be ambiguous. Aliases
are introduced to give the translation of a deprecated standard name into its
approved equivalent, so that existing data remains valid, rather than simply
deleting the deprecated name from the table. It seems that we need to have two
translations for the deprecated standard name in this case.
> Also, thinking out loud a little: Regarding the two directional terms, is the implication that any directional terms may now also have the opposite term? I was wondering if a minus sign achieves the same end (that is, a negative value for surface_upward_mole_flux_* can be treated as downward mole flux). A guideline about when it is and isn't appropriate to include a directional component might be useful, if not already present.
If it's possibly ambiguous, a direction is always indicated in standard names.
There are already pairs of names for opposite directions in the table. We've
done this to accommodate whatever choice data-writers want to make, rather than
trying to prescribe how the data should be written.
More information about the CF-metadata