621 
Figure 1 : Polarised reflectance measured at the 
surface in the principal plane over bare soil 
(Fig. la), and over vegetation (Fig. 2b). The 
lines indicate the results of the models 
defined by eq. (2) and (3). 
Pp(e,. 6 V , cp) = 
My) < 2 > 
4(cos(0j) + cos(0 v )) 
where y is the incidence angle (equal to half 
the phase angle), and Fp is the polarised 
fraction of the specular reflectance as given by 
Fresnel laws. The model is valid for covering 
canopies (i.e. large Leaf Area Indices). 
We developed a different model for bare 
ground surfaces since those have a structure 
different than vegetation: There is no 
attenuation on the incident and outgoing path 
as there is for the vegetation. The model 
assumes that the ground is composed of 
isotropically distributed facets (rough surface). 
One representation for such a distribution is 
obtained considering the surface to be entirely 
covered with hemispheres of varying radii. 
The model writes (Br6on et al. 1994): 
Fi gure 2 : Same as Fig. 1 but for the 
perpendicular plane. 
p P (0 J ,0 v ,(p v ) 
My) 
4 cos Q s cos 0 V 
(3) 
This formulation is clearly not satisfying for 
limb viewing or illumination as it diverges for 
those angles. This results from our 
approximation of neglecting mutual 
shadowing of the facets. For smaller zenith 
angles, however, it agrees with the 
measurements as shown below. 
4 RESULTS AND DISCUSSION 
4.a Polarised reflectance at surface level 
Fig. 1 show typical angular signatures of the 
polarised reflectance in the principal plane for 
bare soil (Fig. la) and vegetation (Fig. lb). We 
recall that the reflectances are positive (resp. 
negative) when the polarisation is 
perpendicular (resp. parallel) to the plane of 
scattering. On each diagram, the solid line 
shows the model results and the diamonds 
indicate the measurements. We use one of the