November 13, 2001

MEMORANDUM TO:  Record
FROM:           Dennis A. Keyser -- NP22
Subject:        Changes to IBM-SP Data Preprocessor (PREPDATA)
                (February 17, 2000 Version -- UPDATE # 8)


The February 17, 2000 data preprocessor program (PREPDATA) was
updated for the eighth time.  UPDATE #8 was implemented on the
IBM-SP machine at 12Z on November 13, 2001.  Except where noted,
the changes below affect all five versions: ETA, AVN, FNL, RUC
and CDAS.

**************** I B M - S P   U P D A T E   # 8  ***************


I. GENERAL CHANGES

     1) Reads in switch MP_PROCESS from a new namelist TASK. This
is the process number for the MPI run (normally 0-11 when this
program is multi-tasked on 12 nodes, defaults to 0 when multi-
tasking not invoked, as currently in the RUC).  It is set in the
ush script prepobs_makeprepbufr.sh where the namelist TASK is
concatenated to the top of the standard input datacards.  The
process number is then encoded in the PREPBUFR file under the new
mnemonic PROCN.  This was added so the separate mass and wind
pieces of the same original report could be easily identified
since they must have the same PROCN and SQN (report sequence
number).  SQN is generated sequentially in each multi-process
run, thus it is not unique across processes.  [MAIN, W3FIZZ,
FIZZ01, BUFRTABLE]

     2) Modified to pass two different unit numbers representing
Global sigma first guess files into W3LIB routine GBLEVENTS.  If
the hour in the center date of the PREPBUFR file is not a
multiple of three, it is expected that two spanning Global sigma
first guess files, valid at times that are a multiple of three on
either side of the center date, will be passed into GBLEVENTS. 
GBLEVENTS has already been modified to linearly interpolate the
spectral coefficients from the spanning guesses to the PREPBUFR
center hour. [Note: The compilation of this program will bring in
these changes.]  This change really affects only the RUC2A
network, since all other runs are at PREPBUFR center hours which
are multiples of three (in which case only the Global sigma guess
valid at the center time is passed into GBLEVENTS, as before).
[MAIN, W3FIZZ]



     3) The number of bits for mnemonic SQN (report sequence
number) is increased from 17 to 19 to handle larger numbers of
data.  [BUFRTABLE]


II. OERATIONAL CHANGES SPECIFIC TO A DATA TYPE

  A. CHANGES TO ATOVS RETRIEVAL DATA PROCESSING

     1) CDAS only: Subroutine W3CNVXTOVS modified to return
satellite id (BUFR code table 0-01-007) in word 6 of unpacked
IW3UNPBF format (integer).  Prior to this work 6 was returned as
missing. [W3CNVXTOVS]

     2) CDAS only: Now encodes satellite id in PREPBUFR file
(under mnemonic SAID) for either ATOVS or RTOVS retrievals
(originating from EITHER 40-level format from unpacked NMCEDS
format).  Prior to this, the satellite id was missing for this
type of data. [MAIN, SATEDS, SATBFR]


III. NON-OPERATIONAL CHANGES SPECIFIC TO A DATA TYPE

  A. CHANGES TO QUIKSCAT DATA PROCESSING

     1) Namelist switches IQKSLM and IQKSSP are no longer read
and are removed.  This program no longer performs q.c. of
quikSCAT data w.r.t. probability of rain and cell number on
orbital swath edge.  This q.c. has been moved upstream to the
program WAVE_DCODQUIKSCAT (optional, but currently it is being
performed). [MAIN, GETSCATT, DATACARDS]

     2) Modified to handle superobs (now generated in upstream
program WAVE_DCODQUIKSCAT) by recognizing S in character 2 of the
report id. [MAIN, GETSCATT]

     3) No longer thins data by processing only every other
report read in.  The upstream program WAVE_DCODQUIKSCAT now
has the option to superob these data (and in fact it is
superobing them). [MAIN, GETSCATT]

     4) The interface to these data in the reprocessed BUFR data
dump, W3UNPKB7, now returns the number of original reports used
to generate a superob in word 10 of the header (the upstream code
WAVE_DCODQUIKSCAT can now superob these data and it writes this
information to the reprocessed dump file).  For non-superobed
reports this defaults to 1.  This information is currently not
transferred to the PREPBUFR file.  [W3UNPKB7]