[CF-metadata] Standard names for anomaly-type parameters

From: Jonathan Gregory <j.m.gregory>
Date: Wed, 2 Apr 2008 18:07:09 +0100

Dear Phil

> Yet is it not the case that the physical property described by the
> 'canonical' name and its associated 'anomaly' name is fundamentally
> the same? All that's changed is the use of a different reference datum
> (i.e. climatological mean instead of zero).
> So rather than create "_anomaly" variants of potentially hundreds of
> CF standard names, would an alternative solution be to add the term
> "anomaly" to the list of standard name modifiers (cf. Appendix C in CF
> spec)? In which case we'd be able to use names such as:
> "air_temperature anomaly"
> "lwe_precipitation_rate anomaly"
> "air_pressure_at_sea_level anomaly"
> and so on

Following an email thread started by Julian Hill, he and I also thought we
would propose exactly this, and I mentioned the proposal at the GO-ESSP meeting
in Paris last June. However it wasn't received enthusiastically. It is a bit
different in concept from the other standard_name modifiers, which are
generally intended for "fields of metadata" - standard errors and so on.
The anomaly of a quantity is arguably more of a different quantity altogether;
a time-interval, for instance, is a different thing from an absolute time.

I agree that in principle any quantity *could* be an anomaly, but in practice
are we *actually* going to need anomaly quantities for all of them (or a
fraction of them which is nearer 100% than 10%, say)? As you say, very few have
so far been needed. Perhaps we should decide on that kind of basis how to deal
with it.

Other views would be helpful.

Best wishes

Jonathan
Received on Wed Apr 02 2008 - 11:07:09 BST