403 
TROPICAL FOREST SAHARA DESERT 
Figure 4 : Temporal evolution of the backscattering coefficient for different vegetation types over Africa 
(year 1992-1993). 
IIL COMPARISON WITH GLOBAL VEGETATION INDEX DATA 
The NOAA-Global Vegetation Index (GVI) weekly data set used in this study comes from the Global Resource 
Information Database (UNEP/GEMS/GRID, Geneva). On a daily basis, GVI maps are computed from AVHRR 
GAC 10 bit as the normalized difference between channel 2 (NIR) and channel 1 (VIS), truncated to 8 bit and 
rescaled. The weekly GVI maps are then derived from the daily maps following the 7-day maximum value 
compositing (MVC) procedure in order to minimize cloud contamination and off-nadir viewing effects. At 
GRID, each weekly file is archived in a 904x2500 byte array format representing the earth's surface, from 75°N 
to 55°S throughout the 360° of longitude, in the standard GRID Plate Carrée projection . The nominal pixel 
resolution is 16 Km at the equator and the time period covered by the archive is from April 82 to present. Due 
to their large temporal and spatial coverage, GVI data have been used in numerous studies and have shown 
great potential in the monitoring of vegetation activities at a global scale (Goward et al. 1985, Justice et al. 
1985). Figure 5 shows the variation of the mean Global Vegetation Index along the transect previously defined 
in section I. Compared with the corresponding figure in the microwave domain (figure 3), similar observations 
can be made. Differences come from the desertic areas where the vegetation index exhibits very low values. The 
backscattering coefficient o° and the GVI are plotted together in figure 6, showing important similarities. 
However, as previously mentioned care must be taken when interpreting such results since the radar wave and 
the optical wave are not sensitive to the same surface parameters.