[CF-metadata] string valued coordinates

From: Ben Hetland <Ben.A.Hetland>
Date: Fri, 3 Oct 2014 14:16:28 +0000

On 2014-10-03, Jim Biard <jbiard at cicsnc.org> wrote:
> The 'no units' case is covered by leaving off the units attribute.

Wouldn't that be in violation of the CF convention?
I was under the impression that _all_ dimensional quantities need that attribute, and a "coordinate" would by its very nature always imply a dimensional aspect, or is there something I've missed?

If it were a 'dimensionless' value, using just "1" as a udunits string would work fine, but here I suppose the request is more along the lines of an _undefined_ unit (as in "missing", "unknown" or "not applicable").


> On 10/3/14, 6:59 AM, Hedley, Mark wrote:
> But I believe that I can create a CF auxiliary coordinate with string values without any concern.

Perhaps you could provide some concrete examples where you think a string as such (*) may be used as a coordinate or dimension?


-+-Ben-+-

[*] i.e., the string is not just an alternative representation of some numeric value or scale, which it could have been converted into by means of a defined process.
Received on Fri Oct 03 2014 - 08:16:28 BST