[CF-metadata] Coordinate System nomenclature in netcdf-java-2.2 / CDM:

[Brian Eaton]

Hi John (and Russ),

To me "axis" has strong connotations of 1D and straight lines.  I searched
through the online Oxford English Dictionary entry for "axis" and found the
string "straight line" matched 8 times.  Using "coordinate variable" for
the 1D object, and "coordinate axis" for the multi-dimensional objects is
backwards from the meanings I would expect.

Earlier in the draft stages of CF I tried using "coordinate axis" for 1D
coordinate variables, and "coordinate variable" as the generalization.  But
it was awkward to be constantly explaining "it's a coordinate varible, but
not a netCDF coordinate variable".  In the end consistency with the NetCDF
User's Guide usage won the day.

If there's any chance that the NetCDF-4 User's Guide usage could change,
please consider starting from scratch with the meaning of "coordinate
variable".

Thanks,
Brian


On Wed, Dec 22, 2004 at 02:23:11PM -0800, John Caron wrote:
> Brian Eaton wrote:
> 
> >On Wed, Dec 22, 2004 at 02:28:16PM -0500, Rich Signell wrote:
> > 
> >
> >>This seems to be agreeing with Brian's earlier statement that saving 4D 
> >>fields of Z are allowed already, enabled by the "coordinates" 
> >>attribute.   Can one of you guys give an example of how this would be 
> >>done?  I have only seen the "coordinates" attribute used to specify 
> >>two-dimensional longitude and latitude fields.  
> >>   
> >>
> >
> >In general, the "coordinates" attribute is used to point to variables that
> >contain coordinate information that can't be represented by a coordinate
> >variable (note that the conventions always use the term coordinate variable
> >to mean a 1D variable whose name matches the name of its dimension).  The
> >only restriction on an auxiliary coordinate variable are that its
> >dimensions are a subset (not necesarily proper) of the variable to which it
> >is attached (and of course there's an exceptional case for dealing with
> >label coordinates).  
> >
> FYI, we have started using the following nomenclature in the Coordinate 
> System layer in netcdf-java -2.2 and the common data model:
> 
> A "coordinate variable alias" is really a coordinate variable (1D, 
> monotonic) but its name doesnt match its dimension.
> 
> A "coordinate axis" is a generalization of a coordinate variable, it may 
> have more than 1 dimension and may have any name. it acts like a 
> coordinate variable in that it assignes a coordinate value to each value 
> in a data variable. It is the same i think as CF's "auxiliary coordinate 
> variable". A coordinate variable is also a coordinate axis.
> 
> A "coordinate system" for a data variable is a set of coordinate axes.
> 
> A "georeferencing coordinate system" for a data variable is a set of 
> coordinate axes that give us lat, lon (or x, y on projection plane),  z, 
> and time coordinates for each value in the data variable.
> 
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://www.cgd.ucar.edu/mailman/listinfo/cf-metadata