August 22, 2006

MEMORANDUM TO:  Record
FROM:           Dennis A. Keyser -- NP22
Subject:        Changes to IBM-CCS Data Preprocessor (PREPDATA)
                (September 9, 2004 Version -- UPDATE # 4)


The September 9, 2004 data preprocessor program (PREPDATA) was updated for the
fourth time.  UPDATE #4 was implemented on the IBM-CCS-II machine at 12Z on
August 22, 2006.  Except where noted, the changes below affect all six
versions: NAM, GFS, GDAS, RUC, CDAS and (new) RTMA.

*************** I B M - C C S   U P D A T E   # 4 **************


I. GENERAL CHANGES

     1) Added new parm cards for new RTMA (Real Time Mesoscale Analysis)
network.  This network has been added to production. [DATACARDS]

     2) Added new subroutine BALLOON_DRIFT which contains the balloon drift
logic.  This was moved out of subroutine STOROB where it was executed in two
places (once for mass and again for wind processing - wind processing needs it
if mass processing did not occur).  Subroutine STOROB now simply calls
BALLOON_DRIFT in these two places.  [STOROB, BALLOON_DRIFT]

     3)  The W3LIB routine GBLEVENTS, which this program calls, was modified 
to have the first guess vertical interpolation use the pressure profile
directly instead of the pressure profile converted to sigma as before.  This
change will allow this routine to read in and properly interpolate a hybrid
guess.  In production a sigma guess is still read in - this change will result
in a very small change in the values of the interpolated sigma guess.
[GBLEVENTS]



II. OPERATIONAL CHANGES SPECIFIC TO A DATA TYPE

   A. CHANGES TO UPPER-AIR (RADIOSONDE, PIBAL, RECONNAISSANCE, DROPWINSONDE)
      DATA PROCESSING
      
     1)  BALLOON-DRIFT UPGRADE [BALLOON_DRIFT]
          a) Expanded the degree of diagnostic print in the balloon drift
calculation for any reports specified in namelist switch STNPRT passed in from
datacards.
          b) Corrected logic errors in balloon drift processing which had led
to previously undetected array overflows and level indexing mis-matches.  This
seemed to happen mainly where there was a problem in the balloon drift
calculation from one level to the next (e.g. a drift of > 1 degree in latitude
or longitude) and the previous level values were moved to the level where
there was a problem in the calculation.  One result of the array overflow
problem was a sudden jump in the drift time calculation between adjacent
levels (> 3 hours!).
          c) Previously, drift time and lat/lon was not calculated on any
levels at or above a level where there was a negative drift time calculation
from the level below to the next level above.  Will now allow the calculation
to proceed if the time difference is < 5 seconds (the time and lat/lon for the
level below is used and then the calculation continues).  Many reports have a
small 1-2 second negative time calculation between adjacent levels at the very
bottom of the profile.

     2) Now prints a diagnostic warning and sends a diagnostic to the
production joblog file is the number of mass and wind levels processed (and
sent to subroutine FILLX) for a particular report are not the same (they
should still always be the same at this point).  [STOROB]

     3) The wind quality mark is set to 15 if a reported wind is found on a
level which is below ground.  [STOROB]

     4) GFS, GDAS only: Gulf Stream dropwinsonde moisture observations
(specific humidity) are no longer flagged.  These should now be assimilated by
the Global-SSI.  (They are still flagged in all other networks as is USAF drop
moisture in all networks.)  [DATACARDS]