394 
At low incidence angles (6<30°), the signal varies almost linearly with no scatter, which is not the case at high 
angles. In April May (x), the soil is bare and very dry hence the low values of ct 0 . The points are almost 
aligned. In August (*) the ground is first dry and then wetted by the first heavy rain. No significant vegetation 
cover is present. The points are thus still aligned but along two lines of differing offsets corresponding to dry 
and wet. At low incidence angle, ct 0 has high values. Finally in September (o) the vegetation starts to be 
significant. For this reason, at low incidence angles, the backscatiering coefficient (wet with little vegetation) 
has a values intermediate between April May and August, but at high incidence angles ct 0 has highest values. 
Consequently in September the points are again aligned along two lines : the first represents little vegetation 
cover and the second significant vegetation cover (flowering stage) with a larger intercept Figure 3 is thus a 
qualitative description of the concept used for the study presented in this paper. 
5. METHODOLOGY 
If we assume that the relationship ct 0 vs incidence angle (0) is linear, we can write for each pixel: 
CT o (db)=A*0+B (3) 
A and B are two constants. Figure 4.1 represents the values of A during the study period. We can see a 
saturation effect for 0>3O°; thus for wet soils, when there is no vegetation (surface scattering), A may be linked 
to soil roughness. According to (3), B is the signal at nadir. To reduce the parameters which intervene in B we 
made the assumption of constant roughness. This appears for 0>3O° (fig.4.1) where soil moisture has not effect 
on the signal. For the time being, B will be linked to biomass. Figure 4.2 shows the difference between slope in 
May and July; this effect is linked to variation in soil roughness (preparation of soils for planting and flattening 
by the first heavy rains). We can also see, that the intercept varies while the slope remains constant when 
biomass increases. 
Figure 4.1 : Slope A as a function of incidence angle. Figure 4.2: Illustration of our methodology. 
6. PRELIMINARY RESULTS : TEMPORAL EVOLUTION BY ANGULAR RANGE OF a* A, AND B. 
According to previous studies, a 0 response resulting from surface features is linked to incidence angles. In this 
view, the results presented will be done within a angular range. The notation [30°,40°] means that all incidence 
angle (mid beam antenna) are between 30° and 40°. We will analyse jointly the temporal evolution of angular 
signatures of WSC and surface parameters evolution. 
6.1. WSC/Rainfall : In arid and semi arid lands, the vegetation growth is strongly dependent on soil 
moisture which is linked to rain. This last alters the soil roughness. For these reasons, rainfall is an important 
factor in term of signal modification. Figure 5 shows that before July (doy 183, bare soil), even at high