[CF-metadata] CF complexity and CF compliance

From: Jonathan Gregory <j.m.gregory>
Date: Fri, 23 May 2008 08:41:46 +0100

Dear Joe

I agree with Karl. It is important to be able to record the data on its own
model grid, and the auxiliary coords are essential then for locating the data.
I don't know whether there are plans to support them in Ferret and GrADS - I
hope so - but even if not, that doesn't mean they aren't essential. When I
analyse data from the NEMO ocean model on its tripolar (non-lat-lon) grid
using IDL, for instance, I use the auxiliary coordinates.

As far as climate modellers are concerned, data-writers and data-consumers
are largely the same people. The CF convention started with climate modelling,
and we have included the metadata judged necessary to describe and analyse the
data. Since data tends to persist for a long time in archives and remain of
interest, it is important to devise and record the metadata at the time you
write the data, even if software hasn't been written to use it automatically
at that time. It can still be used, by using simpler tools like the netCDF
library directly, or by a human reading the file!

It wouldn't make sense to me to exclude useful metadata from being recorded
on the grounds that it was difficult to keep up the software development to
use it. Rather than restricting the development of the convention, perhaps what
is needed is some discussion of priorities for implementation in software? Some
metadata is certainly more generally needed. For example, I would say it is
more important to implement auxiliary coords than to support grid mappings.
That's because aux coords allow you to visualise and do a good deal of analysis
with data on any grid; grid mappings become necessary if you want to do your
own accurate calculations about the coordinate system.

Best wishes

Jonathan
Received on Fri May 23 2008 - 01:41:46 BST