#Panoply netcdf software
Since the netcdf output is in CF format any standard software can be used for display (e.g. One netcdf file is written for either global or nested output which contains all species and all time steps. In the netcdf module netcdf_output_mod a parameter write_releases determines at compile time if the releases should also be written to the netcdf file. Invokation from the namelist switch lnetcdf in the namelist COMMAND file, or by adding 8 to IOUT parameter in the standard text format COMMAND file (example: IOUT=9 corresponds to IOUT=2 with the standard output). Optional netcdf output has been included with changeset:30 (Version 9.2beta, see FpRoadmap).
So I guess Harald will include the netcdf output (at least as an option) in the upcoming 9.1 release. On the other hand, doing so is not really more difficult than compiling the grib_api. But it comes at the cost of correctly compliling a netcdf library with hdf5 support, which might be an additional hurdle for some users. The chosen format follows the CF conventions and can directly be visualised with standard netcdf tools like ncview. We had implemented a netcdf output for the FLEXPART COSMO version and I also put it into the ECMWF version that Harald is working on. So the sparseness is "only" considered in that it is very efficient to compress a long series of zeros. So in the end it comes down to a zip algorithm of the data that is integrated in the library (as far as I understand it). You need netcdf4 with hdf5 and zlib (or similar) support. Does this hold for any degree of sparseness in the output field?
Using netcdf library version 4 makes the flexpart netcdf output of similar or smaller size than the binary format used so far.
0 kommentar(er)
