Hi Phil,
> I can't recall if this topic has come up for discussion on the mailing
> list before - apologies in advance if that is the case.
>
> Although the current CF conventions do not appear to be tied explicitly
> to a particular version of the netCDF library, there are certain
> sections (e.g. 2.2 Data Types) which imply a fairly strong linkage with
> the netCDF-3 'classic' data model.
>
> Thinking about the new features that ship with the netCDF-4 library -
> named groups and richer data types, for instance - would it be correct
> to assume that the same metadata attributes that make a netCDF-3 file
> CF-compliant would also render the equivalent netCDF-4 file CF-
> compliant? (I'm assuming here that the data and metadata are placed
> within the default root group in the case of the netCDF-4 file.)
>
> I think the answer is Yes, but it would be useful to hear the view of
> others.
Yes, a netCDF-3 file that is CF-compliant should also be CF-compliant as
a netCDF-4 file, becasue the CF conventions all apply to classic model
data that's represented in netCDF-4 format files.
> Alternatively, asking the question from the opposite perspective, are
> there updates that need to be factored into a future version of CF to
> cater for the new capabilities offered by netCDF-4?
Yes, but I think it may be premature to propose CF conventions
concerning the new features of netCDF-4 until there is more experience
with using them. So far, most netCDF-4 users seem to be employing the
"classic model", in which data follows the simpler netCDF-3 data model
but may use performance-oriented features of netCDF-4, such as
compression, chunking, larger variable sizes, and efficient schema
changes. Use of such performance features don't require readers to
change their programs, but merely recompile them, and the changes to
writing programs are minor.
I've written a draft document "Developing conventions for netCDF-4":
http://www.unidata.ucar.edu/staff/russ/nc4_conventions.html
that identifies some areas where conventions may be needed for new
features like groups and compound types and proposes a few conventions,
such as for assigning attributes to members of compound type. These are
likely to be "Users Guide" conventions, so CF could inherit them as
happened with the "units" attribute, for example.
Feedback on that draft document is appreciated.
--Russ
> Thanks in advance.
>
> Regards,
> Phil
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
Received on Tue Oct 28 2008 - 15:22:15 GMT
