[CF-metadata] Re: Array Subscript Convention

From: Tom Gross <Tom.Gross>
Date: Wed, 11 Feb 2004 09:59:40 -0500

Dear all:

   I guess I disagree in a way which disagrees with a lot of COARDS
methodology. I think that attributes and units are wonderful things and
that the names of the variables should not include such information. I
think that attributes such as upward or downward should be separate from
the name of the thing. Perhaps COARDS is influenced by the Germanic
languages which form new words out of lengthy strings of adjectives.
But I am not on the COARDS naming panel, so I guess I will submit to the
collective wisdom.

In finite element modeling a list of unstructured nodes is accompanied
by an element incidence list and usually (but optionally) a list of the
boundary nodes. The element list is a counterclockwise index list of
the three nodes making up each triangular element. I generalize to
quadralaterals etc. by incuding nface, the number of sides (or maximum
number of sides). So I guess I need a COARDS standard name for the ele
and bnd arrays? My quoddy netcdfs have:

dimensions:
        node = 7258 ;
        nele = 13044 ;
        nbnd = 2000 ;
        nface = 3 ;
        nbi = 4 ;
        sigma = 15 ;
        time = UNLIMITED ; // (5 currently)
variables:
        int ele(nele, nface) ;
                ele:long_name = "Horizontal Triangular Element Incidence
List" ; ele:units = "index_start_1" ;
        int bnd(nbnd, nbi) ;
                bnd:long_name = "Boundary Segment Node List" ;
                bnd:units = "index_start_1" ;
        float time(time) ;
                time:long_name = "Time" ;
                time:units = "days since 1954-06-28 0:00:00 00:00" ;
                time:base_date = 1954, 6, 28, 0 ;
                time:standard_name = "time" ;
        float lon(node) ;
                lon:long_name = "Longitude" ;
                lon:units = "degrees_east" ;
                lon:standard_name = "longitude" ;
        float lat(node) ;
                lat:long_name = "Latitude" ;
                lat:units = "degrees_north" ;
                lat:standard_name = "latitude" ;
 

> Date: Mon, 9 Feb 2004 22:42:48 +0000
> From: Jonathan Gregory <j.m.gregory at reading.ac.uk>
> To: cf-metadata at cgd.ucar.edu
> Subject: [CF-metadata] Array Subscript Convention
> Message-ID: <20040209224248.GA8452 at met.reading.ac.uk>
> Content-Type: text/plain; charset=us-ascii
> MIME-Version: 1.0
> Precedence: list
> Message: 1
>
> Dear Tom
>
> Instead of introducing a separate attribute, perhaps we could
> indicate in the
> standard names of the quantities themselves what their numbering
> convention is?
> Presumably you will need standard names to identify the various
> arrays of
> indices? If so, including the convention in the name seems
> analogous to
> including "upward" and "downward" in vertical velocity components,
> for example.
> Making it part of the standard name means you can't avoid including
> it.
> Cheers
>
> Jonathan
Received on Wed Feb 11 2004 - 07:59:40 GMT