Product Description:

  Experimental Bi-Weekly Global Normalized Difference Vegetation Index

                                   by 
                             Kevin P. Gallo
            National Oceanic and Atmospheric Administration
    National Environmental Satellite, Data, and Information Service
                  Office of Research and Applications

An experimental normalized difference vegetation index (NDVI) was de-
veloped and produced during 1988 through 1990, from weekly visible and 
near-infrared AVHRR channel data available from NOAA's Global Vegetation 
Index product (Kidwell, 1991) distributed by the:

NOAA/NESDIS National Climatic Data Center
Satellite Data Services Division (E/CC6)
Princeton Executive Square, Suite 100
Washington, DC 20233

Tel: 301-763-8400
FAX: 301-763-8443.

NOAA's Mercator-projected product was utilized.  The data are produced 
for the region between 75 degrees North latitude and 55 degrees South 
latitude.  Data resolution in the Mercator projection varies from 19.6 
km pixel size at the equator to 15 km at 40 degrees (North or South).  
The reflectance values of the visible and near-IR data were computed 
from pre-launch calibration coefficients.  The NDVI was computed as:

     NDVI = (nearIR - visible)/(nearIR + visible).

The calibrated visible and near-IR data, and solar zenith angle data in-
cluded on the NOAA GVI product were utilized to screen the NDVI data for 
cloud contamination and low (less than 15 degrees) solar elevation at 
the time of data acquisition.  Data were also screened for data drops.  
Two successive weeks of the screened NDVI data were then composited 
based on the maximum NDVI value of the two weeks.  The bi-weekly data 
were processed for April 1985 through 1989.  The start date of the bi-
weekly composite intervals was 099 (9 April) in 1985.  The start dates 
in 1986, 1987, 1988 were 001 (1 January).  Processing intervals changed 
in 1988 on 11 April to a Monday through Sunday weekly cycle.  The start 
date in 1989 was 002 (2 January), in 1990 was 001 (1 January) and for 
1991 was 007 (7 January).

The bi-weekly NDVI data have been scaled to a byte format from the 
original ND value (a real number with a range from -1.00 to 1.00) com-
puted with the above equation, using the following conversion:

     byteNDVI = (realNDVI x 100) + 100.

Thus, a byte NDVI value of 151 in the data set is equivalent to a com-
puted real NDVI value of 0.51.  A byte NDVI value of 100 is equivalent 
to a computed real NDVI of 0.0.  Data tagged by the cloud, data drop, or 
solar elevation algorithms will include values of 0 through 2, respec-
tively.

There are 26 bi-weekly files for each year but 1985, for which the 19 
last bi-weekly periods have files.  The file naming convention is YYWW.  
For example, file 8602 covers the first two weeks of 1986.  Each bi-
weekly data file includes 1038 lines and 2048 samples of data in a byte 
format.  The value that occupies the first line and sample of each file 
is located at 75 N and 180 W.  A FORTRAN program that computes line and 
sample location from latitude and longitude is appended to this documen-
tation.

Additional information about the data can be provided by Kevin Gallo, 
NOAA/NESDIS, National Climatic Data Center, Federal Building, Asheville, 
North Carolina 28801, (704) 259-0878, or from the NGDC Global Change 
Data Base help line at (303) 497-6125.

Reference:  Kidwell, K. B., (ed), 1991. Global Vegetation Index Users 
Guide.  NOAA/NESDIS, NCDC Satellite Data Services Division (E/CC6), 
Princeton Executive Square, Suite 100, Washington, DC 20233. 

c-----------------------------------------------------------
c
c This program converts input latitude and longitude values
c into MI and MJ coordinates associated with the NOAA Vegetation
c Index products with Mercator projections described in Kidwell
c (1991).  Written by K. P. Gallo, 11 November 1987.
c
c-------------------------------------------------------------
c
c
      real lat,long, mi, mj, in,lonc, x,y,i,j,reply
      data pi/3.1416/,in/2500.0/,lonc/o.o/
c
  20  write(*,701) ' enter latitude and longitude in degrees '
      read(*,'(BN,2f7.2)') lat, long
      write(*,*) lat, long
  701 format(a\)
c
      x=in*(long-lonc)/360.
      y=in*(lat)/360.
c
      i=x+1250.0
      j=-y+522.0
c
      mi=(i*.8192)
      mj=662.0-(log(tan(-0.00126*j+1.44136))*325.95)
c
      write(*,2001) ' mj(line#)= ', mj, ' mi(sample#)= ', mi
 2001 format(a,f7.2,a,f7.2)
      write(*,701) ' enter a "1" for another lat,long '
      read(*,'(BN,f1.0)') reply
      if (reply .eq. 1) goto 20
      end

Note: this program has been compiled and run successfully on several
computers.  In contrast, compilation and running have produced erroneous
results on other computers.  We can only speculate that the reason for
this problem may lie with the way different compilers handle the code,
and/or in different types/levels of error propagation within computers.