Statistical study of polarization of light scattered by rough grains I. The method With preliminary results

In order to quantify the effects of shape of grains on their optical proper ties, we use a new statistical approach to study the effects of roughness o n the degree of linear polarization. In this approach, we compute the degree of linear polarization of light sca ttered by a rough grain. The roughness is obtained by a random depletion of the surface of an homogeneous sphere. The analysis of the results is achie ved through a specific statistical method: the probability density function of the linear polarization P-l is estimated by a Gaussian kernel method. T his method allows the characterization of effects specifically due to the r oughness. The grain is described by the Discrete Dipole Approximation (DDA) (Draine 1 988). We present the preliminary results for a spherical grain of water ice with radius a(eff) similar or equal to 0.3 mum which is comparable to the incident wavelength lambda = 1.88 mum where water ice is non-absorbing. The effect of grains roughness on polarization can be distinguished from th e effects of volume in several different ways. The shape of the density fun ction of the polarization is found to be non-symmetrical to the mean value of P-l and non-unimodal for several scattering angles. We compare the mean value of the linear polarization deduced from the probability density funct ion to the value obtained from a mass equivalent sphere and from the core s phere. We present the effects of roughness for various scattering angles.