The builtin output formats for wgrib are

      binary: single precision CPU native format
      ieee:   single precision big-endian IEEE (32 bits)
      grib:
      text:

Someone had the need to produce double precision output.  One
may question their wisdom because probably all GRIB data have 
less precision than the single precision floating point
format.  Nevertheless it was easy to modify wgrib to output 
double precision floating point numbers as the standard binary
format. (Writing dp ieee is much more difficult.)



Kludge number 102:

Edit the file wgrib.c by changing all occurances of float
to double.  Change the line

    #include <double.h>

back to 

    #include <float.h>

Remember that you must change ALL occurances of "float" and 
that word may occur more than once on a line.

Compile normally and if all goes well, the "-bin" option
will now produce double precision floating output.  Don't
forget the default format is "-bin".

Note:  The kludge looks good but hasn't been examined with
a fine-tooth comb.  I should probably change "float" to
FLOAT and use a typedef.  However, this change is really
low priority.


END OF KLUDGE NUMBER 102