The NETwork Common Data Format is an extremely flexible storage format for a variety of classes of data. The MRG CDFconvert package is specifically designed to address data conversion issues for gridded datasets stored under the COARDS convention, used almost exclusively for meteorological datasets. CDFconvert is capable of converting regular Cylindrical Equidistant (Lat/Long) and Gaussian (Spherical) netCDF grids into either the Canadian RPN Standard File (FST) or GEMPAK (GEM) file formats. MRG CDFconvert has the flexibility to handle netCDF files generated by a number of sources, including NCEP and ECMWF. User-definable conversion tables make the extension of the package to different datasets possible.
The MRG CDFconvert package is built with GNU Autotools in order to simplify the build process across different platforms. In general, installing the package should be as simple as downloading the latest version of the source code, unpacking the distribution with the gunzip -c cdfconvert.tar.gz | tar -xf - command, configuring for your system by running ./configure from the top level of the distribution, building with make all and installing with make install. For more configuration options, including installation path and additional library definitions, run ./configure --help. This will list the options understood by the configure script.
Note that the linking of the Fortan90 code in the CDFconvert package with older versions of GEMPAK built with Fortran77 compilers may be problematic. On Sun workstations, an EXTRALIBS="-lF77 -lM77" argument to configure may be required. On SGI platforms, the easiest solution to matching ABI architectures will likely be to compile a newer version of GEMPAK with Fortran90. On a Linux RedHat Fedora Core 1 system, an EXTRALIBS="-lpghpf -lg2c" argument was required at configure time.
After MRG CDFconvert is installed, invoke the GUI from the command line by issuing the command cdfconvert. The main window of the application will appear, allowing you to select the input file, output file (and type), projection, date range, optional scalings, data frequency and table ID. These values allow the user to quickly parse the input dataset into manageable sections during the decoding process. For standard data sources, no other settings are required and the Execute button should be used to invoke the decoder once these values are set. However, unknown identifiers and/or multiple decodes into the same output file may require the use of the Advanced and Tables menus. The Advanced options (for each of the RPN Standard and GEMPAK output types) are explained in detail in their associated popup windows. Custom tables can be created and managed through the Tables menu, allowing for the correct decoding of fields whose names do not appear on the standard decoding tables.
A command line version of CDFconvert is available as of v2.2.0 (August 2005). Batch processing is invoked using the cdfconvert --batch command. (Note that input and output file names are also required in this mode.) For more information about the options available on the CDFconvert command line, check the CDFconvert man page.
Questions or Comments? Contact our WebMaster.