NOTICE: This version of the NSF Unidata web site (archive.unidata.ucar.edu) is no longer being updated.
Current content can be found at unidata.ucar.edu.
To learn about what's going on, see About the Archive Site.
All, The way we here go about creating so many NetCDF files is a process that I like to call "templating". We create a CDL that has the coordinate variables' data, except for "time". There is no data in the CDL for the data variable or for time. We then run ncgen on the CDL, and then that "template" NetCDF file is "populated" by a program (usually C, but not necessarily). All the program has to do is write out grids to the data variable, add a value to the time coordinate vector, and update the actual_range attribute of both variables. A simpler program is easier to get right than one that has to create all the dimensions, variables, and attributes, as I'm sure all would agree. ;) What occurs to me is the complication that NetCDF-4 introduces. I'm told that, to take advantage of any compression mechanism in the NetCDF-4 file, this must be set on file creation. Well, we create the files with ncgen. So, unless ncgen (or CDL) is enhanced to specify this compression mechanism, we will lose out. So I hope that such an enhancement will be considered for ncgen/CDL with regard to NetCDF-4. Thanks, -Hoop
netcdfgroup
archives: