GRIB is a
WMO format for gridded data. GRIB is used by the operational meteorological
centers for storage and the exchange of gridded fields. GRIB's major
advantages are files are typically 1/2 to 1/3 of the size of normal
binary files (floats), the fields are self describing, and GRIB is an
open, international standard.
What is in this GRIB File?
A major advantage of GRIB is that is self describing. Each record
has information such as: resolution of the grid, time, variable,
level, who created the field. There are a number of programs
to create GRIB inventories. My favorite is the Ginsu of grib
inventory programs, wgrib.
This program was written in C and works on all reasonable machines
ranging from PCs to Crays. Once you've installed wgrib, an
inventory is a single command line away. You should never have to
ask, "What is in this file?".
How do I Read GRIB?
There are a number of GRIB decoders available. My favorite
is the Ginsu of grib decoders, wgrib.
Once you've installed wgrib, decoding grib takes a single command
line. The output from wgrib can then be read by fortran, C or even
BASIC programs. The GRIB format may look complicated but reading
GRIB is painless. You should never have to ask, "How do I read this data?".
How do I Display GRIB Files?
Gempack and IDL can directly or indirectly display GRIB data.
However, my favorite is the the Ginsu of display programs,
GrADS. Grads is powerful enough that
some people do many of their computations using the GrADS scripting
language. The demonstration program on the BAMS cd-rom (March 1996)
and the annual Reanalysis CD-ROMs shows a sample of GRADS' abilities.
GrADS will run under MS-DOS and on many UNIX boxes. Very powerful,
freely available. and highly recommended.
Suggestions
One's first impulse is to convert the GRIB file into some
"understandable" format. It is better remove the unwanted data
and leave the rest in GRIB. Then you extract the data
at the last minute. The advantages are that it saves disk space
(who has enough disk space?), it is self documenting (what is
record 33?) and is machine
independent. This is similar strategy to using "compressed" files. You
don't need to know the format of a compressed file except
that you can easily uncompress it, and you only uncompress the
data just before use.