[CF-metadata] some concerns about the "ensemble axis" proposal

From: Jonathan Gregory <j.m.gregory>
Date: Thu, 8 Mar 2007 08:15:42 +0000

Dear Jon

I agree with you that the ensemble dimension is an "index" dimension which is
not ordered. You can subset it, which means extracting particular values of
the index e.g. those where the institution is "Met Office", but it is not
meaningful to extract a range, like you do for a continuous coordinate, e.g.
0--45degN.

We ought to distinguish between the issues of (a) the metadata model for the
ensemble axis and (b) how to aggregate datasets. We can discuss (a) in the
framework of netCDF files, as we have been doing, but we recognise that
increasingly the data may not actually be stored in such self-contained files.
So far CF is a standard that deals exclusively with self-contained netCDF
files. It doesn't have any conventions about how to treat several files as
a single dataset, because this has been done by different applications in
their own ways. I think that we should now try to define conventions for
aggregation, but I would suggest we should do that in a different thread.

Best wishes

Jonathan
Received on Thu Mar 08 2007 - 01:15:42 GMT