These data were developed at CICS/ESSIC.  This file briefly describes the
analysis and tells how to read the data.  Questions or comments about the
data should be addressed to Tom.Smith@noaa.gov.  Constructive feedback
about the analysis, its uses and limitations, and suggestions for
improvements are also welcome.

Data: historical reconstructed precipitation (anomalies + climatology).
Analysis are computed using global EOF analyses of GPCP 1979-2008 data 
to form statistics.  GHCN gauge data are fit to the EOF modes to 
produce the analyses.

Two analyses are performed:  
1. A first guess anomaly analysis uses EOFs of annual-average GPCP, 
and annual GHCN+CCA data are fit to the modes to produce the first 
guess annual anomaly analysis.  For this analysis 20 annual EOF modes 
are used.  Data are required to sample at least 5% of a mode's 
variance or it is excluded.
2. Monthly anomaly increments from the first guess are computed using 
40 monthly increment EOFs and monthly increment GHCN historical data.  
Modes that have less than 5% variance sampled are not used.

The reconstruction anomaly is the sum of the first guess and the 
increment analysis.  The monthly reconstruction anomaly is then adjusted
by statistical reinjection of GHCN anomalies where available.  The 
GPCP climatology is added back on the give the full precipitation 
reconstruction.

Reconstruction full field (mm/mon): recon.eof1.mon.ascii.dat.gz
GPCP 1979-2008 monthly anomaly: gpcp.clim.1979.2008.mon.ascii.dat.gz


Grid: 5-deg latitude by longitude.  The 72 longitudes are centered on
2.5E (i=1) to 357.5E (i=72).  The 36 latitudes are centered on 87.5S (j=1)
to 87.5N (j=36).  The polar latitudes are not analyzed and are filled
with special values (-99999) to mark them.

Data are written as integers.  To extract the data and convert the values 
to mm/mon, use the following code:

c    67  1         2         3         4         5         6         7 2
      parameter(im=72,jm=36,iyr1=1900,iyr2=2008)
      real a2(im,jm)
      integer ia2(jm)
c
c* Open the input data file.
      open(21,file='rec.mon.eof1.c05.ascii.dat')
c
c* Convert the real numbers to ascii text to real numbers (mm/mon).
      do iy=iyr1,iyr2
       do mon=1,12
c* Read in the year and month for each month in the record.
        read(21,90) iyr,monr
c* Read in the data as ascii mm/mon x 100 and convert to real mm/mon.
c* Regions with no data at high latitudes are marked -99999 in ascii,
c* and converted to -999.99 in real numbers.
        do i=1,im
         read(21,91) (ia2(j),j=1,jm)
         do j=1,jm
          a2(i,j)=real(ia2(j))/100.0
         end do
        end do
c* Data are now in array a2(i,j).



c* End loop over months.
       end do
c* End loop over years.
      end do
c
      stop
 90   format(2i7)
 91   format(36i7)
      end