Jonathan Gregory writes:
> Dear John
>
> The coordinate information is there so that the data can be used by any
> software without having to be able to interpret a higher-level (but more
> compact) grid description. Even though we may add map projections etc. to CF,
> applications might not support them and if they don't the data is useless.
> The basic question is not how to change the standard, but whether to change
> it, I think. Obviously it is easy to omit the coordinate information, but
> the consequence is that the data will not be usable by the majority of
> applications (I suspect). What criteria could be used to decide whether this
> would be the right choice to make?
>
> If there are many such fields, perhaps the coordinate information could be
> sent in a separate file, just once. This is not CF-compliant, strictly
> speaking, but data-reading software may be able to aggregate many files into
> one dataset, like cdms can do.
I think it is inevitable that we move toward defining datasets that
aren't single files... mosaics, various filesize limits (netCDF,
apache...) are all going to run into this problem.
There are mechanisms for assuring integrity of linkages between multiple
files in a dataset, and _efficient_ software for the aggregation layers
is a requirement as well.
--
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 Aug 29 2005 - 11:04:26 BST
