[CF-metadata] 2D latitude, longitude coordinates detection

[Brian Eaton]

On Thu, Jul 03, 2003 at 05:20:38PM -0600, John Caron wrote:
> Brian Eaton wrote:
> 
> >Hi John,
> >
> >You are correct that the latitude and longitude auxillary coordinate
> >variables should be identified by using the coordinates attribute.
> >
> >The warning that the CF checker should have given is that latitude and
> >longitude are multidimensional coordinate variables whose name matches the
> >name of one of their dimensions.  We recommended against this in the CF doc
> >to avoid the potential confusion with coordinate variables as defined in
> >the netCDF User's Guide.  That confusion is manifested in this file's
> >metadata as apparently the author expects the latitude and longitude
> >variables to be identified as coordinate data since the variable names
> >match dimension names.
> >
> >I believe the checker is correct in saying there are no errors since it can
> >only check for syntax errors.  There's no way for it to know that the
> >author intended certain variables to have latitude and longitude
> >coordinates.  The checker could be improved, however, to give "expert
> >analysis" of the metadata.  For example, in addition to reporting syntax
> >errors, it could give a summary of the coordinates associated with each
> >variable.  A data writer could use this information to verify that his
> >intentions were being correctly expressed in the metadata.
> >
> >Brian
> > 
> >
> Hi Brian:
> 
> i thought CF required that there be lat and lon coordinates.  if they 
> are missing, isnt that an error?

I don't think so.  I can write a CF conforming file which contains, for
example, a time series of globally averaged surface temperature.  This file
would not need to contain lat and lon coords.