Dear Jonathan,
> I understand. That's tricky, [...]
Yes :-) .
> We could define apple to mean orange in
> future, for the sake of the existing datasets,
> but only if we are certain that no-one will
> ever want to talk about apples.
I am not aware of any situation in which someone actually meant to talk
about apples. Markus Fiebig from the World Data Centre for Aerosols
wrote the same
(http://mailman.cgd.ucar.edu/pipermail/cf-metadata/2017/059588.html). I
talked to two former colleagues, who confirmed it as well. Therefore, it
is quite save to assume that nobody talks about apples.
> We could just define and start using the new names,
> and be aware that the CMIP5 datasets used the
> wrong names (because the CF process somehow
> made a mistake), without defining aliases. Would
> that be acceptable?
With respect to my personal usage of the respective standard names I am
fine with just defining new standard names. I also see that it is the
simplest solution for the moment considering the work effort needed to
additionally define aliases.
But, we might run into trouble (and cause confusion), if both standard
names - apple and orange - are used to describe oranges. People, who
used "apple" in the past, probably keep using "apple" to describe
oranges because they are not aware of the changes. People who look up
standard names for their new data sets might also end up with "apple"
for describing an orange if "apple" is not marked as deprecated. Also
people comparing data sets following the old and the new conventions
(e.g. CMIP5 and CMIP6) might not be aware of this discussion. Hence, I
would prefer to define aliases.
Would it be feasible with respect to the required work, to define
aliases for all the ambiguous standard names? How could I support this
process? There seem to be 100 to 110 standard names involved:
- atmosphere_mass_content_of_X_dry_aerosol_particles (15)
- tendency_of_atmosphere_mass_content_of_X_dry_aerosol_particles*
(78, maybe less)
- mass_concentration_of_X_dry_aerosol_particles_in_air (15)
- tendency_of_mass_concentration_of_X_dry_aerosol_particles_in_air (1)
Best,
Daniel
On 03.01.2018 14:40, Jonathan Gregory wrote:
Dear Daniel
Is it feasible to rename all affected standard names?
It would be feasible (using aliases) but is it necessary? It seems to me that
your question has identified that there should be a distinction between e.g.
mass_concentration_of_particulate_X_in_air
and
mass_concentration_of_X_dry_aerosol_particles_in_air
for X=ammonium etc. These are different quantities: the former refers to the
mass of ammonium only, the latter to the dry mass of the aerosol of that type.
That is, we need new names for CMIP6, not aliases.
Yes, there should be a distinction between both standard names.
However, the latter name has been used as synonym for the first name
up till now (e.g. in CMIP5 or in a data set I published recently).
Additionally, the latter name has no real application - at least I
am not aware of an application (neither for model nor for
measurement data). Therefore, it might be reasonable for backward
compatibility to use aliases.
I understand. That's tricky, because we've established that the second name
is a valid concept but not correct. When we use aliases, it's because we've
decided on a clearer, more consistent or more precise formulation of the
name, but in this case, it seemed that we called something an apple when
it ought to have been called an orange. We could define apple to mean orange
in future, for the sake of the existing datasets, but only if we are certain
that no-one will ever want to talk about apples.
We could just define and start using the new names, and be aware that the
CMIP5 datasets used the wrong names (because the CF process somehow made a
mistake), without defining aliases. Would that be acceptable?
Best wishes
Jonathan
_______________________________________________
CF-metadata mailing list
CF-metadata@cgd.ucar.edu
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
_______________________________________________
CF-metadata mailing list
CF-metadata@cgd.ucar.edu
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
