Dear Mark I suppose it therefore makes sense for David and me to write an alternative proposal, to insert some words in the convention document to make clearer what we understand to be the intention of scalar coordinate variables. OK?
As you know, we think they are not a distinct concept in the data model, just a convenient shorthand for storing coordinates of size one in the CF-netCDF file, to avoid defining dimensions of size one. That's why they were introduced to the convention, and I'm not convinced there is a need to change the interpretation. I will draft some words to clarify that interpretation. I now think the discrete sampling geometry is a more complicated case, where we are using CF mechanisms to pack several data variables into one data variable, in effect, using one of the four different representations according to the situation. Therefore I think the interpretation in terms of the data model may need to be different. I suggest we should stick to CF 1.5 until we have resolved the question of scalar coordinates i.e. leave out the discrete sampling geometries until we have our other ideas sorted out. The existing trac tickets are about CF 1.5, as you know. Best wishes Jonathan ----- Forwarded message from "Hedley, Mark" <[email protected]> ----- > From: "Hedley, Mark" <[email protected]> > To: "[email protected]" <[email protected]> > Date: Wed, 12 Jun 2013 09:33:17 +0000 > Subject: [CF-metadata] scalar coordinate variables > > > It's been suggested to me that it might help open up this discussion if I > could post up a suggested text for scalar coordinate variables, which > recognises their semantics within NetCDF files. > > If there is some feeling that this is a helpful step, I will raise a trac > ticket to discuss the finer points of detail on wording and implications and > look to possible approval by the community. > > These are suggested alterations to sections in the conventions text. I think > all of the current examples (5.11, H4, H5 and H9) remain valid as they are. > I think the conventions would benefit from further examples, I leave the > discussion on which cases could be most useful to illustrate the usage and > implications of scalar coordinates to a potential trac ticket. > > mark _______________________________________________ CF-metadata mailing list [email protected] http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
