Re: limits of netCDF library

Stephen Walker (Stephen.Walker@ml.csiro.au)
Tue, 30 Apr 1996 11:30:52 +1000

X-sub: netcdf-group@ml.csiro.au

Peter Halvorson writes:

[stuff deleted]

> As my program iterates along, it will write all the 0d, 1d, and 2d data
> at each step, but only writes the large 3 dimensional arrays at selected
> steps (to conserve disk space). ...

I do this by using several output files. An example fragment
of my hydrodynamic model parameter file looks like this:

# Output files

OutputFiles 2

file0.name out/out125_eta.nc
file0.tstart 1461 days
file0.tinc 2 hours
file0.tstop 1642 days
file0.bytespervalue 8
file0.vars eta wind_x wind_y

file1.name out/out125_all.nc
file1.tstart 1461 days
file1.tinc 10 days
file1.tstop 1642 days
file1.bytespervalue 8
file1.vars ALL # This means all the model variables

The model reads this specification and produces output files accordingly.
I usually store the small (0, 1 and 2-d) variables in a file being written
often, and the small and large (3-d) variables in a file written less often.
This means the small variables occasionally get written twice
(they are in both files), but I find that having them in the same file as
the 3-d variables often tends to make later analysis and plotting easier.
It's not a particularly elegant solution, but I find it quite workable.

Stephen Walker Email: walker@ml.csiro.au
CSIRO Division of Oceanography Fax: +61 02 325123 (International)
GPO Box 1538, Hobart Phone: 002 325298 (in Australia)
Tasmania, AUSTRALIA : +61 02 325298 (International)