TabCDF

The TabCDF package consists of two utilities: tab2nc and nc2tab, which convert plain tab-separated value (TSV) text files to NetCDF data and vice versa, respectively. This process is facilitated by a simple scripting language that uses the .tnc extension by convention (TNC scripts). For more information on NetCDF, see

http://www.unidata.ucar.edu/software/netcdf/

Such a scripting language has distinct advantages:

it is shorter and more intuitive than the corresponding CDL (Common Data Language) description of the data
the same format description - a TNC script - can be used for many files of the same format
it eliminates the need to write specialized programs to convert textual data to NetCDF

Example

As a simple example of TabCDF's capabilities, suppose we have a file of tab-separated values containing the population of several cities over a century:

	1900	1940	1970	2000
Los Angeles	0.102 	1.504 	2.812 	3.695
Washington	0.279 	0.663 	0.757 	0.572
New York	3.437	7.455	7.896	8.008
Seattle	0.081 	0.368 	0.531 	0.563
London	6.528	8.197	7.452	7.322

A corresponding TNC script to describe the layout of the data would be:

	
    { string empty; int year[n] }
m * { string city[]; float population[][n] }

In contrast, the resulting NetCDF file expressed in CDL is much less compact:

	
netcdf out {
dimensions:
		__string = 11 ;
		n = 4 ;
		m = 5 ;
variables:
		char empty(__string) ;
		int year(n) ;
		char city(m, __string) ;
		float population(m, n) ;

// global attributes:
				:__str_len = 11 ;
data:

 empty = "" ;

 year = 1900, 1940, 1970, 2000 ;

 city =
  "Los Angeles",
  "Washington",
  "New York",
  "Seattle",
  "London" ;

 population =
  0.102, 1.504, 2.812, 3.695,
  0.279, 0.663, 0.757, 0.572,
  3.437, 7.455, 7.896, 8.008,
  0.081, 0.368, 0.531, 0.563,
  6.528, 8.197, 7.452, 7.322 ;
}

See the README more comprehensive instructions.

Downloads and installation

TabCDF is available for download as a typical installation package for unix-like systems at

http://biit.cs.ut.ee/~arak/pub/software/tabcdf/

Typically, installation consists of unpacking the tarball, changing to the extracted directory and issuing the commands

./configure
make
make install

TabCDF has two software dependencies:

netcdf >= 3.6.1
argp (on non-GNU systems, bundled with GNU libc)

TabCDF should work on any recent unix-like system, and has been to confirmed to work on the following platforms:

GNU/Linux 2.6 i686, x86-64
Windows (Cygwin)
NetBSD 3.0
OS X 10.4

See INSTALL for more information. TabCDF is distributed under terms of the BSD license. See COPYING for details. Changes by version are outlined in the ChangeLog.

Contact

Tambet Arak ( mail/arak ), Meelis Kull ( mail/kull )
BIIT research group, University of Tartu