Documentation for GPCP Satellite-Derived (IR) Pentad Rainfall Estimates

   To get the file via anonymous ftp, do the following:
        1.  ftp ftp.ncep.noaa.gov
        2.  login with userid anonymous  (use your e-mail ID as the password)
        3.  cd pub/precip/gpi/pentad
        4.  set ftp to binary mode (bin)
        5.  get data gpi_pent_2.5_mmday_198601-yyyypp ( yyyypp - year and number
            of pentad of the last available data) 

**********************************************************************************
*  NOTE!! An inter-satellite calibration  correction for geostationary satellite *
*     data  has been applied to remove calibration disparities between           *
*     geostationary satellites.                                                  *
*                                                                                *
*  NOTE!! A reduction correction has also been applied to GPI estimates          *
*     produced from geostionary IR coverage at zenith angles(ie >25 deg).        *
*     A systematic bias was found by comparisons of GPI estimates computed       *
*     separately from each geostationary satellite's IR data in overlap regions. *
*     The satellite with the larger zenith angle yielded a larger amount of      *
*     cloud colder than GPI's temperature threshold of 235K relative to the      *
*     other satellite.  This correction removes (partially at least) this bias.  *
**********************************************************************************
  

    The data are IR-based raiunfall estimates which are an intermediate
product of the Global Precipitation Climatology Project which is
a WMO/WCRP effort.  An appropriate reference for these estimates 
is:  "Rainfall variations in the tropics during 1986-89, as estimated
from observations of cloud-top temperature", J.E. Janowiak and
P. A. Arkin, J. Geophs. Res., Vol. 96, Supplement, pgs. 3359-3373.
    The data are pentad rainfall rates (units are mm/day) of estimated rainfall
on a 2.5 deg. lat/lon grid.  The data are stored in 144 X 32
arrays, and each value represents the estimated areal average 
rainfall for a 2.5 X 2.5 deg "box".  The array orientation is:
    --->>>>>>>   Missing values are identified as "-9999". <<<<<<<<------

  40N  ----------------------------      -----------------
       |       |       |       |         |       |       |
       | (1,1) | (2,1) | (3,1) |   . . . |(143,1)| (144,1)
       |       |       |       |         |       |       |  
 37.5N ----------------------------      -----------------
       |       |       |       |         |       |       |
       | (1,2) | (2,2) | (3,2) |   . . . |(143,2)| (144,2)
       |       |       |       |         |       |       |  
                                     .
                                     .
 37.5S ----------------------------      -----------------
       |       |       |       |         |       |       |
       | (1,32)| (2,32)| (3,32)|   . . . |(143,32) (144,32)
       |       |       |       |         |       |       |  
 40S   ----------------------------      -----------------
       |       |       |       |         |       |       |
       0      2.5E     5E     7.5E       5W     2.5W     0
 
 The dataset is a series of 144 X 32 arrays of pentad rainfall estimates (in mm/day)
from pentad #1 in 1986 to the last available pentad. The data can be read in FORTRAN 
as follows:

      real precip(144,32)
  
      open(unit=20,access='direct',file=input_file,
     2     status='old',form='unformatted',recl=144*32*4)
   
      do 200 irec=1,wrec
c     wrec - number of all pentad in this dataset 
c     wrec = yy(number of full years)*73(number of pentads in one year)+
c            pp(number of pentad in the last year)
        read(20,rec=irec) precip
200   continue
  
   
      **************************************************
      *      --------->  PLEASE NOTE   <-------------  *
      *  The estimates are valid for the tropics and   *
      *   warm season extratropics, and will indicate  *
      *   rainfall in regions of persistant thick      *
      *   cirrus.  This is because temperature is the  *
      *   only parameter used in the estimation scheme,*
      *   so, except for the cases noted above, when   *
      *   limited to regions where the background temps*
      *   are relatively warm, the estimation technique*
      *   is able to identify regions of deep convec-  *
      *   tion.                                        *
      **************************************************

************************************************************
                       IMPORTANT!!!
From the 19th pentad of 1998 pentad data for 2.5x2.5 are derived
from 1x1 degree daily data.
************************************************************