[CF-metadata] Ocean CTD data following CF Conventions v1.6

From: Jonathan Gregory <j.m.gregory>
Date: Wed, 4 Apr 2012 17:55:28 +0100

Dear Andrew

Thanks for your email.

> 1) What distinguishes an 'auxilliary coordinate variable' from a plain old
> 'coordinate variable'?

A plain old coordinate variable is as defined by the Unidata netCDF user's
guide. It is 1D, monotonic, and its name is the same as the name of its
dimension. An auxiliary coordinate variable is a CF concept. It can be
multi-D and it is not necessarily monotonic. Auxiliary coordinate variables
are linked to the data variable by being named by the coordinates attribute;
that is the function of this attribute. Coordinate variables are implicitly
associated with the data variable through the name of their dimension.

> 2) Is it permissable for a 'auxilliary coordinate variable' to have
> either the scalar form
> e.g "float lat" or the array form lat(lat) with lat=1 declared as a dimension?

lat(lat) is a (Unidata) coordinate variable. It is permissible (but not
required or recommended) for it to be listed in the coordinates attribute.
lat (scalar) must be named in the coordinates attribute; otherwise, it would
not be associated with the data variable. Hence, it is an auxiliary coord var.
CF calls a scalar auxiliary coordinate variable a "scalar coordinate variable".

> 3) Do we need the coordinates attribute to "look-up"
> the corresponding auxilliary cordinates variable? For example if we had a
> variable declared as salinity(z) then to look up its lat, long and
> time coordinates for say display purpose
> would the "salinity:coordinates = "lat long time"" attribute used
> for the lookup.

Yes. That is the only way you could associate lon, lat and time with the
salinity variable, since it doesn't have dimensions of lon, lat or time.

> 4) However if we had the salinity variable be declared as
> salinity(lat,long,time,z) then
> the application would lookup the co-ordinates from the co-ordinates
> variables declaration names
> and we wouldn't need the salinity:coordinates attribute?

Exactly.

> I do notice that a lot of the netCDF CTD data out there doesn't use
> the coordinates
> attribute. Is this a feature that came in later with CF v1.5 or v1.6?

No, it's existed since the start of CF.

Best wishes

Jonathan
Received on Wed Apr 04 2012 - 10:55:28 BST