Dear Martin
I like your idea of identifying the chemical compound in an extra tag in the
standard_name table. Coincidentally this is very similar to what I proposed
a couple of days ago in ticket
https://cf-pcmdi.llnl.gov/trac/ticket/79 about
vector components. There, I suggested that the standard_name table could
identify the components in a separate tag e.g. eastward in eastward_wind.
The reason is the same: it's difficult for a program to parse standard names.
In fact, I think this is generally a good way to address a long-standing
issue. Standard names are not really intended to be parsed. Although we try
to construct them systematically and consistently, they are not necessarily
capable of being analysed into semantic elements. This is true particularly
because we deliberately use words and phrases from natural language to make
the standard_name more self-explanatory. For instance, we use "wind" rather
than "air velocity", although the latter would be analogous to "sea water
velocity" and "sea ice velocity". To get round this, we can record semantic
markers in the standard_name table. That means the table will become a
dictionary, providing some explanation of what the names mean (for use by
programs - humans should be able to understand them anyway), not just a list.
For consistency, I think the semantic markers should correspond to the largely
systematic ways in which we have constructed the names, e.g. the phrasetypes
in my lexicon
http://www.met.reading.ac.uk/~jonathan/CF_metadata/14.1/#lexicon.
As usual, I would not advocate doing this comprehensively, just for the cases
we need, when we need them. Getting back to your particular case, I have a
couple of comments:
* What if there is more than one compound involved in a standard name? This
could be the case with the existing construction using expressed_as, and you
can probably imagine better than me other situations where standard_names
might want to name more than one species.
* Is it specifically only compounds that you want to identify, that is,
excluding elements, lumped groups of components, ions or other chemical
species? In my draft lexicon, I have a category "species". This includes
all chemical species, but also biological species and some other things such
as aerosol and grapel. That's because they appear in similarly patterened
standard_names. Do we need to draw a semantic distinction? That would increase
the number of standard_name patterns.
Best wishes
Jonathan
Received on Tue Sep 11 2012 - 01:07:40 BST
