read_bufr.F will convert KMA bufr conventional observation
  files to little_r format for ingest to OBSPROC.

  The reasons for keeping the little_r format are
  1) OBSPROC uses the little_r memory structure to store the data.
  2) OBSPROC uses the NCEP bufrlib (for creation of prepbufr files
     for ingest into wrfvar), while the KMA bufr files are based
     on the ECMWF bufr library. It's better to keep the libraries
     separate.
  3) The ECMWF bufr library, while freeware, is licensed and 
     cannot be distributed with OBSPROC. The ECMWF library is
     designed and built differently than OBSPROC and cannot be
     merged easily. Since KMA already has the bufr library, it's
     unnecessary to distribute it with the OBSPROC code.
  4) Only KMA needs this converter. There is no need to include
     it in the released OBSPROC code.
  5) The extra step of converting to little_r is not very time-consuming
     since conventional datasets are rather small. Large datasets
     (radiances, reflectivity, etc.) are not converted to little_r
     format.

     The decoder uses the ECMWF bufr library and tables.

To use:

   Put read_bufr.F in the example directory of the ECMWF library
   distribution.

   Add the following lines to the Makefile:

   EXECS    = decode_bufr read_bufr

   read_bufr: read_bufr.o 
	   $(FC) $(FFLAGS) -o $@ read_bufr.o -L$(PLACE) -lbufr$(R64)

   make.

   Create a file called 'flist' containing a list of bufr files
   to be decoded. These must be the file names only, not the full
   path. Use ln -s to create a soft link to the files if they reside
   in a different directory from where the decoder will be run. 
   The reason for this is that the bufr files contain no indication
   of what type of data are in the files. This information is
   gleaned from the first part of the file name.

   Then simply run the executable 'read_bufr'. An ascii file called
   littler will be created.


Known bugs as of 2/16/2006

For ship and synop files, pressure and sea level pressure are encoded
as mb or hPa, when the units are supposed to be Pa. read_bufr will need 
to be corrected if the bufr files are corrected.

For both pilot and temp ttbb and ttdd files, pressure and height are 
missing. This renders the data useless.
Pressure is available in the gts-coded message, so it should be available
in this data file.

(ttaa and ttcc files seem to be OK)

For acars and aireps, the minute value is not encoded (it's always zero).

For satobs, the satellite ID is not encoded (value is zero). The pressure
is not encoded properly. The table entry 7004 (pressure) should be used
rather than 10004. In the sample file, all pressures are zero, making the
data unusable.


Jim Bresch
NCAR/MMM
16 February 2006