[CF-metadata] "grid relative" vector components revisited

From: Jonathan Gregory <j.m.gregory>
Date: Thu, 5 May 2005 18:42:14 +0100

Dear Rich

I think we do need different names for northward/eastward and grid-relative
x/y components, because both may be needed in the same file. I agree there could
be quite a lot of such pairs, but I'm not sure it is an overwhelming problem
like some of the others we've tackled. With examples like yours
> 200 micron sand concentration * sea_water_x_velocity
we might instead consider whether any of this information could be factorised
out of the standard name. The 200 microns, for instance, could be a scalar
coordinate variable or perhaps a standard name parameter, or maybe even the
whole of the tracer identity could be factorised out somehow.

However, "relative to grid" does seem rather imprecise for a standard name,
in that does implicitly depend on coordinates being present and that the
software is clever enough to deal with it. Would it be a good idea to require
unit vector fields to be supplied (relative to E and N) to specify the
direction, so vectors could be drawn by software that didn't understand the
grid? This would be analogous to requiring lon and lat coordinates to be
supplied so the data can be geolocated.

Best wishes

Jonathan
Received on Thu May 05 2005 - 11:42:14 BST