Re: netcdf4 and ncdump

Karsten Bolding <karsten@xxxxxxxxxxxxxxxxxxxx> writes:

> That is a pity. I do not know anything about HDF and as such do not know
> if it is because of HDF that the conventions used in NetCDF3 where a
> variable with the same name as a dimension is a coordinate variable (or
> maybe that is only in COARDS).

Yes, that is the reason. It is a netCDF-3 convention that the variable
with the same name as the dimension.

> What I really like about NetCDF4 is the abillity to have multiple
> unlimited dimensions (in my case it will always be time) since I'll use
> it for storing regions (hyperslabs) in higher temporal resolution than
> the rest of the domain. Often these hyperslabs will share one or more
> dimension with the root dimensions e.g. root dimensions are e.g. (z,lat,lon)
> and surface layer group is (time,1,lat,lon). Here I would find it
> natural to be able also to re-use the coordiantes associated with
> lat,lon. 
> A frequent use of the multiple un-limited dimensions will be storing
> high frequency data for positions corresponding to tide-gauge stations 
> and buoys - could easy be more than 50 for the North Sea/Baltic Sea setup 
> I mainly work with. Having 50 times the same vertical coodinate variable
> in the same file is not very elegant.

There is no good reason why the coordinate var needs to be in the same
group as the data variable.

The coordinate vars could be in a root group, and the data in other

When accessing the coordinate variables, you would just need to
provide the root group id...



Ed Hartnett  -- ed@xxxxxxxxxxxxxxxx

To unsubscribe netcdfgroup, visit: