Penttilä, Antti
(Helsingin yliopisto, 2009)
The polarization of asteroids and comets, and especially the various models for polarization observations, are studied. The goal of this study is to compare existing models for polarization and their properties. Based on the results presented here a particular model and estimation technique can be found suitable for a certain types of polarization observations and research frames. With the polarization particularly the linear polarization ratio is referred here. The Sun radiates light or more generally electromagnetic radiation - in the framework of this study to an asteroid or to a comet - in where all the linear polarization planes are equally represented, i.e. non-polarized light. Depending on its surface structure the target might scatter the different polarization planes of the radiation with different intensities making the scattered radiation polarized. The ratio between the difference and the sum of the perpendicular and parallel polarized components is called the linear polarization ratio. The linear polarization ratio can tell the researcher something about the target properties, e.g. about the surface material composition, packing density and roughness. Some models have been developed for the linear polarization ratio, and the models can be fitted to the observations as a function of the phase angle. These models are fully empirical and are not based on actual physical modeling of the polarization event. The estimated models are, however, useful when comparing polarization properties between different targets with often quite sparse and limited observational data. The nonlinear regression analysis and also Bayesian nonlinear regression analysis are used in the model estimation. With the Bayesian method the a priori information about the behavior of the polarization curve can be utilized, making the estimation robust also in cases where the number of observations is small or the data is insufficient in other ways. The polarization is also studied as a function of the wavelength of light when the interest is in the effect of the wavelength on the polarization properties. Modeling these effects reliably requires a delicate selection of the proper model and the estimation technique. The possibilities of the multiple response and the hierarchical Bayesian regression analysis in modeling are studied. Suitable models for analyzing the polarization observations are presented and compared in the thesis, and some results concerning the wavelength effect in polarization are studied.