Hi Mike,
The recommendation that non space/time dimensions be added to the left of
space/time dimensions was taken from the COARDS document, and is part of
CF's attempt to remain COARDS compliant. However, as you note, this
recommendation can't be followed if you wish to keep time as the unlimited
dimension. Several solutions come to mind. The one you mention is a
possibility. I'd be tempted though to use the order (time, wavelength,
latitude, longitude), since then wavelength is in the slot normally used by
the vertical level and applications that expect a level coordinate there
might be made to work with wavelength instead. Another possibility is to
not use wavelength as a coordinate and instead put that into the variable
name so that each wavelength (or range of wavelengths) is in a variable
with dimensions (time, latitude, longitude) which makes COARDS conforming
apps really happy. In the end I think it comes down to making sure that
the applications that need to read your data can handle the dimension
ordering you choose. My impression is that we're not yet at the stage
where there are lots of robust implementations of the full CF conventions,
so you still need to be concerned with dimension ordering.
Brian
On Wed, Sep 27, 2006 at 12:06:09PM -0700, Godin, Michael wrote:
> Under 2.4, Dimensions, Recommendations, the CF standard says "In files
> that are meant to conform to the COARDS subset of CF, any dimensions of a
> variable other than space and time dimensions should be added "to the
> left" of the space and time dimensions as represented in CDL."
>
> So, if I have a variable that varies with time, latitude, longitude and
> wavelength, then the recommendation is to use the dimensions:
>
> wavelength, time, latitude, longitude
>
> However, such a set of dimensions doesn't allow me to define a compliant
> file with an expanding time dimension and an extra dimension. Is there a
> subset to CF/COARDS that would otherwise follow COARDS, except for the
> dimension ordering, which would allow me to define a much more manageable
> variable with dimensions such as:
>
> time, latitude, longitude, wavelength
>
> Thanks,
> Mike
>
> _____________________________________________
> Michael A. Godin
> Software Engineer
> Monterey Bay Aquarium Research Institute
> Phone: 831-775-2063 http://www.mbari.org <http://www.mbari.org/>
>
>
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://www.cgd.ucar.edu/mailman/listinfo/cf-metadata
Received on Wed Sep 27 2006 - 14:11:55 BST
