Hi Bryan,
This issue has soaked up a lot of my thinking time in the past and I'm not sure I've got a satisfactory answer yet. At the moment, I handle wavelengths, suspended particulate material size fraction and biological entity size as part of the term. I have often questioned the wisdom of this because it is not fully scalable, providing a potential source of uncontrollable growth in dictionary size - today we have spectral radiometer data at 1 nm resolution, but what if we get 0.01 nm resolution data in the future? However, the co-ordinate variable approach has worried me even more because I can no longer uniquely label measurements without external qualification turning a usage vocabulary into a discovery vocabulary.
We have two solutions to a problem here, neither of which really works and so we need to look for other possibilities. One possible answer lies in further development of parameter description semantic modelling, which I'll maybe look at once I've got the current semantic model built and operational.
The discovery angle also needs some thought. I'm not sure marine scientists would search for specific size fractions - they're more likely to look for 'size-fractionated chlorophyll' than '2-5um chlorophyll'. I guess the really important thing is that a search for 'size-fractionated aerosols' finds datasets labelled 'PM10', which brings us back to the need for an ontology covered by a thesaurus server.
Cheers, Roy.
>>> Bryan Lawrence <b.n.lawrence at rl.ac.uk> 9/27/2006 8:25 am >>>
hi Folks
The discussion of aerosol standard names raises some issues. I think
things are going along just fine, but in the longer term there are some
problems looming ...
Putting things off into coordinate variables. This is in principle
fine ... but when one comes to finding particular variables amongst huge
numbers of files, then the more thing we put off into coordinate
variables the harder discovery becomes.
In practice we start having to build "discovery standard names" which
have all the semantics (forget the syntax for now) of the coordinate
variables, the cell measures etc. People do look for pm10 for example,
it is a common search term, and it has meaning as a name to a huge
community, including the entire pollution world. And what will we put on
the plot title? This new "compound/discovery" name, not the standard
name.
I think we have to be pragmatic and identify some "coordinates" which
are also in in such common usage that they really are names.
So in the following discussion:
> PM10 is particulate matter with an aerodynamic diameter of up to 10 ?m,
> i.e. the fine and coarse particle fractions combined.
> > Use of dimensions:
> >
> > * Some of the names have numerical values in them. This is not in keeping with
> > usual practice in standard names. For instance, we do not have standard names
> > for surface air temperature at 1.5 m or 2 m height. Instead, we use coordinates
> > for independent variables. In names like these:
> > mole_fraction_of_CO_with_lifetime_of_25_days_in_air
> > pm10_aerosol_ambient_optical_depth_at_550_nm
> > mass_fraction_of_pm10_aerosol_at_50_percent_relative_humidity_in_air
> > I think the lifetime, particle radius, wavelength and relative humidity should
> > be coordinates (probably scalar coordinate variables), rather than being in the
> > standard name. This requires standard names to be defined for the independent
> > variables you need.
I disagree with this suggestion. Temperature at 1.5m is an obvious
coordinate thing, but pm10 is not a coordinate thing, so I don't think
particle radius should disappear off into ancillary information, it
really is integral to the name (as Christiane says, it even appears in
legislation). I'm not even convinced by the argument that wavelength and
RH should disappear off into coordinate variables either.
What is the use case for having them in coordinate variables? Do we ever
expect people to use these variables as coordinates per se? (We can
imagine it in the temperature case). We have to look past whether these
look like numbers and ask: are they labels? or are they things which
have properties like coordinates?
Bryan
_______________________________________________
CF-metadata mailing list
CF-metadata at cgd.ucar.edu
http://www.cgd.ucar.edu/mailman/listinfo/cf-metadata
--
This message (and any attachments) is for the recipient only. NERC
is subject to the Freedom of Information Act 2000 and the contents
of this email and any reply you make may be disclosed by NERC unless
it is exempt from release under the Act. Any material supplied to
NERC may be stored in an electronic records management system.
Received on Wed Sep 27 2006 - 02:12:11 BST
