Computational light scattering by atmospheric dust particles

Mineral dust is one of the most common aerosols in the atmosphere, and it can have substantial impacts on both the weather and the climate. Many of these impacts arise from the interaction of visible light and other radiation with the dust particles, which causes the incident light energy to be scattered and absorbed by the particle. Computing these interactions via single-scattering simulations is needed for applications and models to be able to account for the radiative effects of dust. A large portion of dust particles are in the same size scale as visible light wavelength, which means that their single scattering properties cannot be calculated by either Rayleigh scattering, which is applicable to much smaller particles, or optical models, which are applicable to much larger particles. Instead, one must use models applicable to wavelength-scale particles, which typically either assume a specific mathematical shape, or resolve the electrodynamic interactions between discretized volume elements of the particle shape model. In this work I have studied the effects of detailed dust particle shape and composition features on light scattering by computationally generated or modified dust particle proxy models. The features studied were the surface roughness and the varying types of internal structures. Surface roughness was shown to have a clear impact on light scattering even when the particle is irregularly shaped. Internal structures, in particular hematite nodes, were also shown to have substantial effects on scattering. In addition to investigating the effects of these particle features, I also tested the validity of using effective-medium approximations to homogenize atmospheric dust, and the validity of one particular mathematical shape model, ellipsoids, for retrieving particle refractive index. The use of effective-medium approximations seems to produce erroneous results when applied to particles with macroscopic, localized inhomogeneity, especially when the refractive index of the inhomogeneity differs notably from the bulk material. Finally, using ellipsoidal model particles in dust refractive index retrievals were shown to produce wrong results in virtually all of the tested scenarios, and the magnitude of these errors was enough to cause severe inaccuracies in the case where the retrieval results were applied to computing the dust particle radiative effects.