[CF-metadata] Re: Array Subscript Convention

From: Jonathan Gregory <j.m.gregory>
Date: Wed, 11 Feb 2004 17:45:21 +0000

Dear Tom

CF makes a lot of use of attributes. Much information is factorised out of
the standard name in this way. For instance, we don't have a standard name
for 2 m air temperature (unlike the AMIP table) because we regard 2 m as
a vertical coordinate, which should not be included in the standard name.

One argument for including the 0 or 1 convention in the name is because it is
absolutely essential for interpretation of the data that this convention be
stated. The same is true of whether a flux or a velocity is upward or downward.
By putting them in the standard name we make it impossible for the convention
not to be stated. I think it's virtually certain that if these conventions
were recorded in other attributes they would sometimes be omitted.

> I need a COARDS standard name for the ele
> and bnd arrays?

Yes, I think that would be right, and I argue that these standard names
should have the 1 or 0 convention suffixed to them.

> ele:long_name = "Horizontal Triangular Element Incidence
> List"
> bnd:long_name = "Boundary Segment Node List" ;

Is it possible to make these more self-explanatory so that they could be
used as standard names? Maybe they refer specifically to some kind of grid?

Best wishes

Jonathan
Received on Wed Feb 11 2004 - 10:45:21 GMT