GRIB Data Format used for the COSMO-Model System

Last updated: February 2013

Features of GRIB

The national weather services are using coded messages for the international exchange of meteorological information comprising observational data and also for processed / simulated data. While observational data is coded using BUFR (Binary Universal Form for the Representation of meteorological data), the simulation data are coded in GRIB. The history of GRIB already spans some decades and there have been several changes and augmentations to the structure of the GRIB data representation, which are identified as different GRIB edition numbers. The current edition number is 2. However, GRIB edition 1 still remains in use. And GRIB 1 is yet used by the COSMO-Model. Note that the meaning of the abbreviation GRIB also changed with the edition numbers.

The GRIB code is specified in the Manual on Codes, which can be downloaded from the WMO pages. GRIB messages contain meta data, which describe the data under consideration, and the data itself. GRIB coded data consist of a continuous bit-stream made of a sequence of octets (or bytes: 1 octet = 8 bits). Thus, the representation of the data is independent of any particular machine representation. The octets of a GRIB message are grouped in sections. Most sections are used to specify the meta data, while one special section, the Data Section, contains the data in compressed form.


Back