It sounds to me that the NCO utilities may be what you want? Check out
http://nco.sourceforge.net, particularly the ncks operator.
On Monday 19 September 2005 09:01 am, Brian Etheridge wrote:
> Hi All,
> I posted the same question last Friday, but I didn't get any response so
> I am going to rephrase it now that I have explored the problem myself a
> little more.
>
> My program is applying subset and reduction criteria to a netcdf file
> and outputting another netcdf file. I am also allowing the user to
> choose which data variables they get in the output. The problem I am
> having is that when I write out the metadata I have conflicting
> requirements for how that metadata will look. How do I reconcile
> differences in the subset and reduction criteria which has ben applied
> to the data variables. This does not seem to be posible given the
> structure of the netcdf file.
>
> When looking closely at the metadata which is written to the output
> netcdf file I can see that the length of each coordinate variable is
> set. This doesn't make sense when I am subsequently gong to write out
> data variables which either don't use one of the coordinate dimensions
> at all, or have different lengths for the coordinate variables.
>
> I suspect I am trying to do something that the netCDF file format does
> not support.
>
> Thanks
> Brian
--
John Evans: IMCS Rutgers University
71 Dudley Rd, New Brunswick, NJ 08901-8521
ph: 732-932-3435
"When people don't want to play the blame game, they're to blame"
-- Jon Stewart
