[CF-metadata] Standard_name table entries for air quality data

From: Jonathan Gregory <j.m.gregory>
Date: Tue, 4 Oct 2011 15:39:25 +0100

Dear Martin

Unlike Philip, I don't have the expertise to comment on the chemical content
of your email.

> > * (actually more a question): How do I express the "daily 8 hour
> > maximum" value that is derived from hourly measurements of (surface)
> > ozone mole fractions? I would tend to apply cell_methods here, but how
> > should this attribute look like? To explain in more detail: first an 8-
> > hour running mean of hourly concentrations is calculated, then the
> > daily maximum value of these running means is picked as the indicator.
>
> [PJC] I don't have an answer, I just want to reinforce the usefulness of having such cell_methods. 1-hour means are also common.

Do you mean you end up with one value for each day, being the largest of the
24 running-8-hour means within the day? There is not current a way to describe
this in cell_methods. We'll have to devise one if this is a common need.

> > aerosol" ? (and would "Mie" be spelled with "M" or "m"?)
>
> [PJC] I couldn't find any capital letter in the std_name list, but I also couldn't find any proper nouns (such as people's names) either.

I think it would be lower case, as we use lower case for everything, even
when upper case is more common (e.g. toa for top of atmosphere). This is
to avoid irritating problems with wrong case being accidentally used. I am
sure it would be very unsafe to distinguish two names just on the basis of
lower vs upper case, so we lose nothing by ignoring case, in effect.

> > * Stumbling over the "group" or "lumping" concept again, I would
> > propose to add a new attribute to the CF standard which is specifically
> > used to describe the group compounds.

I am not sure whether you mean a variable att or a global att. I would
certainly favour the former, because I think it's better for variables to
describe themselves. Is a new att really needed, though? Could you provide
a more detailed description in the long_name, perhaps? You could standardise
the long_name, to allow automatic processing, for particular applications. On
the other hand, if there is a common need to distinguish different choices of
lumping, because the quantities are not really comparable for the purpose in
hand, that would be a reason for making the distinction in the standard name.

best wishes

Jonathan
Received on Tue Oct 04 2011 - 08:39:25 BST