[CF-metadata] Variable Associations (was CF convention for vector quantities)

From: John Caron <caron>
Date: Mon, 20 Mar 2006 10:24:19 -0700

I have to agree with Bryan and Bert, that indicating vector quantities is useful and important.

It would be a mistake, IMO, to overload the standard names with this info, and expect a program to parse these. Better to use another, more explicit mechanism (im unclear if an actual proposal for this has been made). Adding another dimension is such an obvious thing to do, i would lean towards supporting that also, although there are tradeoffs with keeping each component a seperate variable.

The principles I would suggest to govern these cases:

1) CP should cover conventions about information that is intrinsic to the data from the data provider's POV. This would eliminate, eg color tables or metadata indicating that the data could be used for K-12 education. (the provider can include this or anything else they want, im talking only about CF conventions).

2) It should be a common case in the CF world. So we might argue how common vector quantities are, but including them doesnt require us to include less common cases.

It may not be a bad idea to also require a "canonical example file", and at least one implementation of generic (visualization?) software that can successfully parse it. But thats a bigger discussion.
 
Received on Mon Mar 20 2006 - 10:24:19 GMT