[CF-metadata] CF2 and standard names

From: Jonathan Gregory <j.m.gregory>
Date: Tue, 7 Feb 2006 19:47:01 +0000

Dear Bert

Thanks for your questions.

> when it comes to the use of standard names, I'm still not certain what to do.

Standard names, like most features of CF, are optional. You don't have to use
them if you don't think they will help you.

> some kind of standardisation can
> be helpful for users to understand model output from various sources and as
> such it may be a valuable source of information when setting up a coupling
> between model components of different origin

That is the first reason to your question
> (1) what are the main reasons for using standard names
It's not only for coupling, but for any activity where data from different
sources is to be brought together, for example in the comparative analysis of
data from various climate models. The second reason is that within a data
source, standard names may be a convenient and precise help in distinguishing
among quantities.

> I feel a bit uneasy about
> introducing the standard name "sea_surface_height_above_sea_level" in the
> output of modelling systems that have so far been using "water level" or
> similarly simple looking phrasings.

You know what you mean by "water level" in the context of your own model, but
the standard name is a more precise description of a particular quantity.
However, it may not be the quantity you want, or precise enough for your
purposes. We should introduce further standard names which are more appropriate
for other kinds of application, as they arise. There is an awkwardness about
"lake", "sea" and "river" which we haven't properly dealt with yet. The choice
of "sea" reflects the origin of the standard names. Some word was needed to
distinguish water in the atmosphere (as cloud or vapour) from water in the
ocean.

> reference level is not everywhere the same: a
> river in Tibet may be modelled using a completely different reference level
> than a lowland river like the Rhine or the Mississippi.

Quite right. It has previously been proposed that we should have a facility,
via a standard name parameter, to specify the reference surface. This is not
necessary in all applications (for instance in climate models, which have
idealised Earth geometry), so it should be optional. Different kinds of
application require different levels of precision, so standard naming has to
offer such levels.

> (3) or, wouldn't it be better to define a name space and some mapping as has
> been suggested in relation to CF2 standards

If sufficiently precise names exist in other domains, we could devise some way
of mapping them into the standard name space. This would save effort and
duplication in the community. However we have to be careful that in doing this
we don't accidentally allow different domains to give alternative names to the
same quantity without the equivalence being noted, because if that happened the
names would not be useful for their original purpose of indicating which data
are comparable.

Best wishes

Jonathan
Received on Tue Feb 07 2006 - 12:47:01 GMT