393 
Where P! and p 2 are visible and near infrared reflectance obtained after applying atmospheric correction based 
on the SMAC model [15]. This index reduces soil influence on the signal through the factor L, and is suitable to 
describe vegetation cover on semi-arid areas. To reduce the number of NOAA/AVHRR data over the 50*50Km 
area, for each day we use the MSAVI value which is representative of the site. In the rainy season, the cloud 
cover alters the signal whose temporal evolution appears noisy. To have a smoothed temporal evolution we 
filtered the MSAVI data using a sliding temporal window (fig.l). 
42.2. ERS1 WSC : For each pixel this sensor measures c 0 with its 3 antennas under viewing angles 
ranging from 18° to 59°. The fore and aft beams have similar incidence angles due to instrument geometry. Its 
frequency (5.4Ghz) and its temporal resolution (3-4 days at maximum) is a motivation to assess soil features in 
spite of a poor spatial resolution <50Km). Due to its lower incidence angles, mid beam antenna (18 to 46°) will 
be more used. Data used correspond to the period April to December 1992. On fig.2 we can see the temporal 
evolution and angular variation of a 0 (mid beam). We easily identify the vegetation growth and some responses 
(doy 132 and 153) resulting from rainfall events. 
Figure 2 : o 0 (mid beam) temporal evolution and angular variation. 
For three different periods, we observed the angular variation of a 0 , fig.3. 
Figure 3 : Quasi linear relationship between o 0 and incidence angles (x (April May), * (August), o (September)).