574 
Scattering by small particles like aerosols is generally modeled by assuming the particles are spherical and applying 
Mie theory. Mie theory calculations provide polarization signatures which exhibit large sensitivity to the aerosol 
dimension (more exactly, to the [aerosol radius]/[observation wavelength] ratio) and to the aerosol refractive index. As 
the aerosol scattering efficiency varies typically according to A.' n , with n depending on the aerosol dimension but 
generally smaller than about 2, the aerosol influence in the atmospheric polarized radiance predominates at near infrared 
wavelengths. The sensitivity of polarization to the aerosol properties is especially large in backward scattering 
directions, which are accessible from space. These characteristics make, therefore, polarization a promising tool for 
aerosol monitoring, particularly if polarization is observed across a range of both directions and near infrared 
wavelengths. 
In the case of terrestrial surfaces, polarization modeling at the macroscopic scale may be difficult, especially concerning 
the shadowing effects linked to the surface structure. By contrast, when the single scattering properties of the aerosol 
are known, polarization modeling of the atmospheric effect is quite tractable. Exact calculations of the atmospheric 
effect may be obtained from elaborate radiative transfer calculations, but such treatments are by no means necessary for 
a first approach. Because the Earth's atmospheric optical thickness is rather small, polarized light is due primarily to 
single scattering. Realistic estimates of the polarized light may be derived from single scattering approximations. 
Moreover, the polarized vibration resulting from single scattering processes within the atmosphere or from single 
reflection processes on canopies is linear and parallel or perpendicular to the scattering plane. Then, provided that 
polarized radiances (or reflectances) are defined as algebraic quantities (here positive for the electric field perpendicular to 
the scattering plane and negative when parallel), the incoherent contributions from aerosols, molecules and surfaces are 
just additive, except for appropriate transmission terms. In a first approximation we will therefore write for the 
polarization of the light scattered by the aerosols, as observed from ground, 
5p(0)P(0) 
R _ — 
p 4cos0 j cos0 u 
where 8 is the optical thickness, p(0) the phase function and P(0) the degree of polarization for aerosol scattering at 
the scattering angle 0; 0s and 0v are the solar and viewing zenith angles. 
The problem concerning the atmospheric polarization is, therefore, whether the polarization contributed by aerosols 
may be modeled satisfactorily using Mie theory. In effect, as a counterpart of the polarization sensitivity to the 
particle properties, it may be that departure from sphericity could make polarization by aerosols quite erratic and 
incomprehensible. Theoretical results comparing scattering from spherical and ellipsoidal aerosols show that 
polarization, rather than the phase function, is more affected by departures from sphericity. This is an important 
problem potentially limiting the usefulness of polarization measurements for remote sensing purposes. 
III.B. Aerosol observations 
Clearly, the only answer to this question may come from experiments. As a possible first approach, ground based 
observations could be used to survey the sky polarization comprehensively within the framework of Mie theory. To 
determine whether terrestrial aerosols exhibit the predicted features, we would observe the skylight polarization in clear 
sky conditions, correcting for the known contribution due to Rayleigh scattering and, eventually, for the indirect 
contamination due to the surface. 
Fig. 4. Polarization by different kinds of aerosols. Fig.A: Degree of polarization P, at 1650 nm, 
observed on 3 different days (curves are translated by AP=10) in a maritime site; the sun direction is 
indicated by a vertical line and measurements are reported vs. the zenith viewing angle. Fig.B: Polarized 
reflectance at 850 nm, observed over La Crau; measurements are corrected from molecular scattering, 
then normalized to unit aerosol optical thickness, and reported vs. the scattering angle. Fig.C: Polarized 
reflectance at 850 nm observed during Hapex Sahel; the measurements are reported vs. the zenith 
viewing angle; the aerosol optical thickness is indicated; the continuous curve shows the molecular 
contribution; parasitic light explains the dispersion of the measurements in forward scattering directions.