July 10, 1998


                        Bug fix for wgrib v1.7


  A user was reading the source code and found a mistake in the
decoding section.  When the GRIB message has a bitmap (i.e., has
undefined values) and the precision of the data is greater than 
24 bits then the decoded data is wrong.  The error was introduced
when I optimized the decoding section.  Now for the good news: 
the error is probably "theoretical."  NCEP and the ECMWF do not 
store data to more than 16 bits precision.  I really can't imagine 
anyone storing meteorological data to 25 bits precision.  Even the 
typical single precision number only has 23 bits precision.  If 
you want to find the precision of a GRIB message, type 

                   wgrib -V gribfile

and look for the "num bits".  (Summary: don't worry but you
might want to check the precision of your GRIB files.)

  Along with the bug fix, v1.7 also has the source code in a tar 
file.  The tar file contains the original source code before the 
modules are combined into one source code file.  This version is
for people who want to read the source code.

  If you are still using wgrib v1.5 or before, you really should
upgrade.  Improvements include Y2K fix, the NCEP operational 
table, updated Reanalysis table support, thinned grid support and
new command line options.


                 grib2ctl.pl is at version 0.9.12.5p17

   Grib2ctl.pl make GrADS control files.


                  gribw library has been updated

   The gribw library is a set of C routines to read and write GRIB
files.  The routines have been pretty stable.  The library to be in 
"alpha" testing primarily because there hasn't been enough testing 
and the documentation is sparse.  If you want to test/use the library 
send me an e-mail.


                            Wesley Ebisuzaki
                            Climate Prediction Center
                            NCEP

ebis@wesley.wwb.noaa.gov
http://wesley.wwb.noaa.gov/