624 
Figure 5 : Polarised reflectance at aircraft level 
measured by the airborne POLDER in the 
principal plane. Full circles are for 450 nm 
measurements whereas open diamonds are 
for 865 nm. In both figures the plain lines 
show the polarised reflectance generated by 
molecular scattering considering a clear 
atmosphere (no aerosols) and no surface 
(upper one for 450 nm, lower one for 865 
nm). The dashed lines correspond to the 
surface models of Eq. (2) and (3) (upper one 
bare soil model, lower one for vegetation). 
The aerosol optical thickness is 0.28 at 450 
nm and 0.22 at 865 nm. 
surface vegetation cover. However, those lines 
do not consider signal reduction resulting 
from atmospheric diffusion. This reduction 
increases as the atmospheric optical thickness 
increases. At 450 nm, the polarised radiance 
generated by the surface is largely reduced by 
atmospheric scattering. The polarised 
radiance measured at the aircraft level is 
generated mostly by molecules and aerosol 
single scattering. At 865 nm, the optical 
thickness is smaller and, especially during 
"clear" days, a large fraction of polarised 
radiance generated by the surface does reach 
the instrument. The angular signature of the 
polarisation is not as smooth at 850 nm as it is 
at 450 nm. The total reflectance image at this 
wavelength shows a large spatial 
heterogeneity which may introduce some 
noise on the polarised reflectance because of 
misregistration (we recall that the polarised 
reflectance is derived from a combination of 
three measurements which need to be 
precisely registered). However, we note that 
the deviation from smoothness is compatible 
with the result derived from Fig. 4: The 
normalised value of 0.15 found above (contrast 
between bare soil and vegetation) is equivalent 
to a polarised reflectance of 0.22 at nadir 
Figure 6 . Same as Fig. 5 but for a hazy 
atmosphere: The optical thickness is 0.77 at 
450 nm and 0.68 at 865 nm. 
viewing, and 0.29 at 40° zenith viewing angle. 
Those values have the same order of 
magnitude as the high frequency variability in 
Fig. 5. We can then interpret the "noise" on 
the 865 nm data points as a variable signal 
from the surface, which depends on the 
surface vegetation coverage. 
An interesting result yields from the 
comparison of Fig. 5 and Fig. 6. Although 
there is a large increase in aerosol optical 
thickness between the two days, the signal on 
the polarised reflectance is rather small. At 
450 nm, the polarised reflectance is even 
smaller for the larger optical thickness. This 
implies that the aerosols in the instrument 
field of view show little polarisation. Other 
similar measurements over southern France 
have shown a polarised reflectance much 
larger than that theoretically produced by 
molecular scattering (Deuzé et al., 1993) which 
suggested large polarisation by aerosols. 
However our results are consistent with other 
measurements by Nakajima et al (1989) which 
indicated a small polarisation of desert aerosol. 
If aerosol particles show little polarisation, 
their effect on the total polarised radiance can 
be negative. This is because aerosol diffusion 
reduces the polarised radiance generated by 
other processes such as molecular scattering 
and surface specular reflection. 
5 CONCLUSION 
In this paper, we investigate the polarised 
reflectance of natural targets. We analyse the 
measurements acquired over bare soils and 
vegetation, both from the surface and from an 
aircraft, and we compared these observations 
to the predictions of two analytical models.