LONG LEAD FORECAST DATA SET FORMAT

DATA SET NAME  cpcllft.dat ,cpcllfp.dat

DESCRIPTION:

   These two datasets contain the official CPC long lead forecasts
interpolated to 102 climate regions (prior to 1999) or 126 locations
(102 forecast divisions and 24 Alaskan stations) after Jan 1 1999.

  Data begin with the forecast
issued in December 1994 and continue through the present.  Each 
forecast is identified by a header record that describes the data 
used for the forecast, and its lead time.  Two groups of data follow;
the first contains the forecast values for the probability of the 
below normal class, and the second contains the forecast probability 
of above normal.  The forecast probability for the near normal class 
is one minus the sum of the two given classes. Each group consists of 
the forecast value for 102 climate regions based on NCDC's climate 
division data.  The data order and location are described in climate 
region dictionary file. The last record is a blank, with year 
(IY)=9999, and a record of  data, all equal to a missing indicator 
of -9.999.  

FORMAT:
   Line type 1:     IY,IM,IH,NSTA

                    FORTRAN FORMAT(5I5)

     KYR, KMO = Year (ie. 1997) and month (1=Jan ... 12=Dec) that the  
                forecast was issued. An official
                forecast is typically made about ten days after the 
                last complete month of data and is issued around the 
                15th of the month  KYR, KMO.
   
     KP    =  Lead time for this forecast (in months).  This is the lead
              time between the latest data used (the month prior to,
              KYR,KMO) and the start of the valid time of the forecast.
  
     NSTA  =  Number of data locations in the each of the two data groups.

     KID =  Data identifier, 950 (951) = Forecast probabilities for 
            seasonal temperatures (precipitation).

     KFL = Data flag.  This location is reserved for data identification
           purposes.  It is used to label any procedural factor that may 
           affect forecasts.  0= interpolation from Intergraph software and
           "borrowing" rules used to define non-favored categories.
           1 - Manual interpolation with "borrowing" rules used for non-
           favored categories
           2 - interpolation from NTRANS software.  Transformed-normal 
           distribution with skill contraction used to integrate distribution
           to obtain non-favored categories.

Line type 2:        (PB(I),I=1,NSTA)
                    FORTRAN FORMAT(9(12(F6.3)/))
     
     PB(I) = The forecast probability that the 3 month season that 
             begins KP months after KYR,KMO will be in the lowest third 
             of the climatological distribution of that element in 
             region I.  The valid time can be computed by adding KP+1 
             to KMO, adjusting for any year change. So, for example,
             if KYR=1994, KMO=11 and KP=2 then the valid time is:
             KMO + 2 + 1=14.  Since 14>12, the year must be adjusted, 
             (+1) and the month becomes 14-12=2   indicating that  the 
             valid period begins on the first day of February (month 2)
             1995, and extends through the last day of April (FMA 1995).   

Line type 3:        (PA(I),I=1,NSTA)
                    FORTRAN FORMAT(9(12(F6.3)/))
     
      PA(I) = The forecast probability that the 3 month season that
              begins KP months after KYR,KMO will be in the highest 
              third of the climatological distribution of that element 
              in region I.