Hi John,
I was wondering what the next step is for this initiative? You have
had only two responses since you posted this request some two weeks ago,
though one (by John) listed a good number of points. Does the CF
conventions committee plan to publish a draft setting out its views for
comments, and if so when might this happen?
As an interested observer of the CF standard names activity, there are
several issues that have struck me:
1. The non-scalability of the current approach based on a single long
flat list of names, as others have also observed. It seems to me that a
necessary way forward is to move to an approach based on compound names,
built from lists of authorised terms in various categories, along with
grammar rules which define legitimate combinations of these terms.
(Simply tagging a name with extra attributes, e.g. 'air_pressure' tagged
with 'cloud_base' will not be sufficient, since that can't handle cases
involving e.g. a ratio between two quantities. You need a grammar.)
Otherwise the community will end up with tens of thousands of names,
with a lot of shared structure.
2. The support for multiple formats for the compound names allowed by
the above approach, e..g a plain-text format (much as is used at
present), and an XML format (and possibly an RDF format?). This is
analogous to plain-text and MathML syntaxes for representing a
mathematical expression.
3. A detailed analysis of what a "standard name" for a variable actually
is. It is clearly not a "name" in the normal sense of the word (i.e. a
unique identifier), since one dataset or model can legitimately have
more than one variable with the same "standard name". It therefore
seems (to this outsider) more like a standard type, label, or
characterization of a variable, rather than a standard name as such.
Sorry if this seems obvious within the CF-metadata community, but I've
looked carefully through the associated documentation (e.g. section 3.3
of the CF Conventions), and can't see this stated explicitly.
4. Not sure this is an "issue", but I think it would be very helpful to
publish a set of typical use-cases for standard names. What actual
processing do people do which makes use of standard names? Given the
widespread adoption of the CF-metadata standard naming scheme, there
must be many examples of use, probably well known within your community,
but not easy for an outsider to find. Having such a set of use cases
would go some way to conveying what a standard name "is".
In the particular area I'm currently involved in (plant-environment
modelling), the main uses will be:
- linking a variable in a dataset to an input variable in a model; and
- linking an output from one model to an input in another, when
constructing composite models.
Standard naming/typing/labelling schemes can clearly help in this,
through flagging invalid combinations and suggesting valid ones, though
manual intervention will remain essential (to stop air_temperature
measured in Australia being an input to a European model...).
Cheers,
Robert
John Caron wrote:
> The CF Conventions committee is looking to clarify the issues surrounding CF standard names and the process for creating them. There are a number of proposals and discussions for changing or augmenting standard names, and not yet much consensus on what direction to take.
>
> We'd like to come up with a clear statement of what standard names are (or should be), and what are the problems and issues that we should be focusing on next.
>
> You are invited to send input to this email group, ideally as concisely as possible. You are welcome to add your ideas of possible solutions, but it would be helpful to keep those separate for now.
>
> The CF Conventions committee will attempt to capture some kind of consensus on what the issues are, which may help clarify what solutions are possible and what still needs to be explored.
>
> Thanks!
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
>
>
--
The University of Edinburgh is a charitable body, registered in
Scotland, with registration number SC005336.
Received on Mon Nov 24 2008 - 07:29:13 GMT
