Dear Jonathan
> > consider the following example:
> >
> > dimensions:
> > station = 11 ;
> > time = 2 ;
> > max_len_id = 10 ;
> > vec_dim = 4 ;
> >
> > float a_vector(station, vec_dim) ;
> > char station_id(station, max_len_id) ;
> > ...
> >
> > How can the application determine that (station) is a spatiotemporal dimension,
> > but (max_len_id) and (vec_dim) are non-spatiotemporal dimensions?
>
> Since station_id is an auxiliary coordinate variable, it should be named in
> the coordinates attribute. In 6.1, we have "An application processing the
> variables listed in the coordinates attribute can recognize a labeled axis by
> checking whether or not a given variable contains character data." In general
> all the dimensions of an auxiliary coordinate variable must be dimensions of
> the data variable, but there is an exception for string arrays. In the
> conformance requirements for section 5, we say "The dimensions of each
> auxiliary coordinate must be a subset of the dimensions of the variable they
> are attached to. An exception is a label variable which will have a trailing
> dimension for the maximum string length." Hence, because station_id is of
> character type, you know that its trailing dimension max_len_id is not a
> spatiotemporal dimension.
Thanks for the clarification. In other words the application always has
to check the variable type first in order to correctly interpret the type
of the dimensions.
But this still leaves me with the question of how a general application
can determine that <station> is a spatiotemporal dimension/coordinate,
whereas the <parcel>-dimension of the example in section 6.1 is not a
spatiotemporal dimension, but a kind of "iteration" dimension.
Wouldn't a scalar coordinate variable as shown below be
helpful?
netcdf test {
dimensions:
station = 11 ;
pressure = 3;
time = 2 ;
max_len_id = 10 ;
variables:
float station ;
station:axis = 'X' ;
...
Best wishes
Juerg
Received on Mon Jul 28 2003 - 07:29:29 BST
