575 
Ground based or balloon borne observations of the polarization of the skylight were conducted by different groups, 
from different locations (e.g.: B.Herman, Deuze et al. 1988, Deuze et al. 1989). Analysis of these observations 
provides good indication that the aerosol polarization signatures are meaningfull and generally quite comprehendible 
within the freamework of Mie theory. Figures 4A to 4C show some examples. 
In order to observe maritime aerosols, measurements were conducted from Tromelin Island (15°52'S, 54°25'E) in 1989 
during a two week experiment. Figure 4A shows the degree of polarization observed at nearly the same solar 
elevation (6°-10° ) on different days at a wavelength A=1650nm where Rayleigh scattering is negligible. The aerosol 
optical thickness, derived from solar transmission measurements, and the humidity are indicated. The polarization 
clearly exhibits characteristic permanent features, specifically the increase of positive polarized light in the rainbow 
direction (near from scattering angle 0=140°), an increase which is characteristic of scattering from aerosols having 
refractive indices ranging from 1.30 to 1.37. These polarizations, and the correlative radiance measurements, are 
explained well These polarization, and the correlative radiance measurements, are well explained within the model of 
Shettle and Fenn (1979) for hydrated maritime aerosols, using Mie theory calculations (Devaux et al. 1994). 
As a second example, Figure 4B shows polarization measurements obtained on different occasions during a 5 years 
period over La Crau in the southeast of France, during experimental campaigns devoted to SPOT calibration (Santer et 
al. 1992). The observation wavelength is 850nm. The measurements were corrected for the ground contamination, 
molecular scattering and the quite variable aerosol loading. Figure 4B shows the polarized reflectances normalized to 
unit aerosol optical thickness. The measurements were obtained at nearly the same solar elevation and are shown as a 
function of the scattering angle. The gross polarization features are similar, especially the change in the polarization 
direction in backscattering directions. The polarization and radiance measurements, when analyzed within the 
framework of Mie theory, allow estimation of the aerosol refractive index, m. The best fit, except for the curves 
labelled 1 to 3 in Fig. 5, corresponds to 1.40 < m < 1.44, values indicative of sulfate aerosols. By contrast, 
observations 1 to 3 which depart from the general behavior, correspond, for curves 1 and 2, to measurements 
conducted, respectively, 1 and 2 years after the Pinatubo eruption, and for curve 3, to unusual atmospheric conditions 
with an intrusion of a maritime air mass over the observation area. Inversion of these measurements is more difficult; 
after the Pinatubo eruption, retrieval of the tropospheric aerosol properties requires that the measurements should first 
be decontaminated by removing the contribution, derived from SAGEII measurements, due to stratospheric aerosols. 
Polarization measurements conducted quite routinely on stratospheric aerosols for about ten years, from the balloon- 
borne radiometer RADIBAL, provide other examples of polarization measurements comprehensive within the known 
properties of the particles, i.e. hydrated sulfuric acid aerosols in this case (Herman et al. 1986, Santer et al. 1988). 
As a last example, let us consider the difficult case of desertic dusts. Very detailed light scattering measurements, 
performed on such particles by Nakajima et al. (1989), showed that modeling of the polarized light scattered by desertic 
sands raised difficult problems, probably because of the departure of these particles from sphericity. As light scattering 
by nonspherical particles is an active research area, such measurements may make comprehensive sense in the future. 
The question, therefore, is whether the diversity of possible signatures, when the sphericity assumption is no longer 
valid, is not so large that tractable information can not be derived from light scattering measurements. Ground based 
measurements performed during the 1993 Hapex Sahel experiment indicate that polarization by desertic aerosols should 
be significant and should provide useful information. Figure 4C shows again polarized reflectances as a function of the 
scattering angle, measured during Hapex (Tanre et al.. 1994). Data collection, spread over about 6 weeks, corresponded 
to a variety of atmospheric aerosol loadings, with aerosol optical thicknesses ranging from 0.2 to 1.2. Despite the 
large variability in the experimental conditions, polarization properties exhibit fairly constant behavior: the aerosol 
neutral point (i.e. the direction where the polarized light just corresponds to the molecular contribution, in Fig. 4C) is 
quite stable and the polarized light corresponding to aerosols is nearly proportional to their optical thickness. 
III. C. Conclusions 
Repeated observations confirm that polarization by aerosols provides significant and usable information. The 
sensitivity of polarization to the particle characteristics - which is its interesting point - is not so large that it might 
render analysis of polarization data quite intractable. For many kinds of aerosols, polarization is understandable within 
the context of Mie theory. Joined with conventionnal light scattering measurements (aureole, solar transmission), 
polarization measurements allow derivation of the particle refractive index. For some aerosol types like desertic dusts, 
it may be difficult probably to use polarization for retrieving aerosol characteristics; in such case, however, 
polarization seems able to provide a characteristic signature allowing the identification of these particles. 
IV-POLARIZATION MEASUREMENTS FROM SPACE 
IV. A. Introduction 
To test the possibility of obtaining information from polarization measurements conducted from space, we can use 
results obtained during field campaigns of the airborne POLDER instrument. POLDER is a radiometer designed to