[CF-metadata] new chemical species

From: Jonathan Gregory <j.m.gregory>
Date: Mon, 16 Jul 2007 23:07:20 +0100

Dear Philip

> First, the list of CF standard names will ultimately become very long
> (with concentrations, tendencies, emissions, depositions, and many other
> quantities for each species)

I agree that if we carry on as we are presently working, we could add hundreds
of names to the standard name table for chemistry. I don't think that is in
itself a problem, though, so long as it is easy to search the table, and it's
available in machine-readable form for programs.

> Second, unless proposers are very diligent, or there is some mechanism to
> avoid it, there will end up being a lot of 'holes' in the convention.
> For example, long lists of wet_deposition_of_X, and emission_of_Y, could
> be added, but with X and Y being different sets which only partially
> overlap.

Yes, but that also doesn't necessarily matter, does it? We already have plenty
of holes like that in other areas, because we have not added names until some-
one has needed them. One way in which holes may be a problem is that when you
find one you need to be filled, you have to request a new name. But if it is
an obvious interpolation, there should be no problem with including it.

> Would it make more sense to represent chemical quantities under CF by
> splitting the standard name into two pieces, e.g. 'Quantity_of_X' and
> 'X_is_species_name' ?

We could do something like that. For consistency with other similar cases,
we would make one of them a coordinate variable with string values. I expect
that in some chemistry models there are data variables which have a dimension
in chemical species, and this approach would represent that kind of
organisation. We have started to do this with land-surface types. However,
I do not actually think we need to do it for chemical species, because they
are generally well-defined and discrete, even though numerous. Land-surface
types are rather more model-dependent and so far we have not standardised them;
it is convenient to put the non-standardised part of the quantity in a
coordinate variable.

Best wishes

Jonathan
Received on Mon Jul 16 2007 - 16:07:20 BST