Developing Particulate Layers for a Light Scattering Model as Potential Means of Interpreting Remote Sensing Data of Planetary Surfaces

Abstract

The study of terrestrial surfaces (regolith) of celestial bodies on a large scale is only realistic if conducted remotely. One method of achieving this is through investigating the directional properties of light scattered from them. This light contains information that can be used to interpret their origin, structure and taken as a whole, the evolution of the solar system. Previous studies into photometric properties of light scattered as a function of phase angle, with particular attention to the feature known as surge is reviewed. Surge is where there is a sudden increase in brightness of an object when the backscattering angle approaches the source of the illumination. For celestial objects this is only seen when the celestial body in question is in opposition, i.e. the Earth is directly between the celestial object being studies and the Sun. It is therefore also known as the opposition effect. Features such as shadow hiding and coherent backscatter are explored and explained as are their individual contributions to surge. There is some discussion given to the merits and inherent flaws taken by previous studies of this effect, namely the empirical best fit approach compared with various computational methods in determining probable surface features. This thesis explores a method of simulating light scattering by particulate layers. This differs mostly from previous computational work concerning coherent backscatter in that this uses non‐spherical particles as opposed to spheres (Mishchenko et al.) or aggregates of spheres (Litvinov et al.). The method is tested against backscattering measurements on snow layers by Kaasalainen et al.. For this purpose model layers have been constructed according to the snow layer descriptions (dominating crystal shape, mean snow/grain size, layer density). According to private communication with the authors, ice crystals are assumed to be randomly aligned. The method employed here is similar to the one described by Shkuratov at al.: Particles are randomly distributed in a cuboid, the lateral and bottom sides of which are cyclically closed. Light scattering by these model layers and the resulting phase functions are computed by a ray tracing model in which incident rays have a given finite diameter and are associated with the electric field (not irradiance). Incident rays are arranged in a regular array covering the projected cross section of the layer. The sum of the cross sections of the normally incident rays is equal to the projected cross section of the layer. Diffraction is calculated for individual rays leaving the layer and the complex electric field values are added to the respective angular bins of the electric field distribution. In this way, the diffraction integral of a facet or sub‐facet area is approximated by the sum of the diffraction integrals of the individual rays leaving the facet/ sub‐facet area. When raytracing has finished the phase function is calculated. Comparisons of modelling results with measurements have been carried out for two samples: one of compact hexagonal columns and one of thin plates. For comparison with the experimental results, the same four‐parameter empirical fitting function as in Kaasalainen et al. was used. The phase functions of the compact particles can be compared directly. Similarities were found in the case of the improved seeding technique for needles and compact columns, though analysis of the field measurements and images presented in the Kaasalainen papers has called into question the similarities in the similarities between their samples and the simulations. For non‐compact particles such as hexagonal plates and needles the layer densities measured in the experiment could not be achieved for the model layer. This also enhances the problem of incomplete randomisation due to the limited layer cross section area (which is probably also the reason for the strong oscillations in the measured phase function of the compact hexagonal columns). Since particle size, shape and alignment distributions affect the width of the backscattering peak, it would have been helpful, if more information about these properties had been available.