Jonathan Gregory writes:
> CF doesn't have a convention for associating separate nc files and treating
> them as though they were a single file. However, some software can do this
> e.g. CDMS/CDAT.
As Bryan mentioned earlier in this thread, there is much to be said for
indicating links between files. I'm including this in the comprehensive
grid proposal, where in fact this is definitely needed.
I do see this being used not only for grid specification, but also for
linking vector components, including those which may not be collocated,
i.e on a staggered grid.
There is indeed a tension between Jonathan's minimalist approach (where
smart software fills in many blanks) and including enough information in
the metadata to make it somewhat human-readable. I've tried to do this
in the not-yet-human-readable gridspec draft proposal by using optional
keywords that redundantly provide information that could otherwise be
deduced from the 'required' elements, by software, and by industrious
humans.
Thanks,
--
V. Balaji Office: +1-609-452-6516
Head, Modeling Systems Group, GFDL Home: +1-212-253-6662
Princeton University Email: v.balaji at noaa.gov
Received on Mon Apr 03 2006 - 09:28:01 BST
