Lehning, MichaelLöwe, HenningKrol, Quirien Eibhilín2017-10-272017-10-272017-10-27201710.5075/epfl-thesis-7996https://infoscience.epfl.ch/handle/20.500.14299/141649urn:nbn:ch:bel-epfl-thesis7996-3Snow microstructure and its evolution play an important role for various applications of snow physics in cryospheric sciences. The main modes of microstructure evolution in snow are referred to as isothermal and temperature gradient metamorphism. The former describes the coarsening driven by interfacial energy while the latter is dominated by recrystallization processes induced by temperature gradients. An accurate description of these processes in snowpack models is of key importance. However common snowpack models are still based on traditional grain metrics, originally tailored to field observations, and empirical evolution laws. This treatment of snow microstructure is essentially unrelated to recent advances of snow observations by microcomputed tomography (ÎŒCT). The present thesis contributes to the solution of this problem by i) identifying suitable microstructure parameters ii) deriving evolution equations for these from first principles and iii) developing methods that allow to utilize 4D ÎŒCT measurements of snow as a link between local ice crystal growth and upscaled microstructure as relevant on the scales of interest for common snowpack models. To this end, three studies have been conducted. The first study focuses on estimating local ice-crystal growth rates from interface tracking by analyzing 4D ÎŒCT data of in-situ snow metamorphism experiments under isothermal and temperature gradient conditions. For temperature gradient metamorphism diffusion-limited growth is considered, while for isothermal metamorphism the data is compared to kinetics and diffusion limited growth. Despite considerable scatter, in both cases the significance of underlying growth laws could be statistically confirmed. The second study uses ÎŒCT images from a variety of snow samples to investigate the role of grain shape in the context of microwave and optical properties of snow. Grain shape can be objectively defined via size-dispersity of structure from the second moment of either the mean curvature distribution or the chord-length distribution. In addition, a quantitative link between these quantities and the exponential correlation length is shown. The latter is relevant for parameterizing macroscopic properties such as microwave scattering coefficients, dielectric permittivity and thermal conductivity. Finally, a rigorous, upscaled microstructure scheme is developed by deriving mathematically exact evolution equations for the density, specific surface area, the mean and Gaussian curvature and the second moment of mean curvature. The microstructural evolution is driven by local ice crystal growth. All parameters are upscaled by volume averaging and the correctness of the model is confirmed for the time evolution of idealized grains. The model can be compared to 4D ÎŒCT data without any a-priori assumptions. This benchmarking reveals the uncertainties of the interface tracking method which are largely caused by limited temporal and spatial resolution. The model allows to statistically assess the validity of ice crystal growth laws during snow metamorphism. For a temperature gradient experiment it is shown that a diffusion limited growth law is not consistent with the observed decay of the specific surface area. The developed model is a powerful and rigorous tool that is tailored to 4D ÎŒCT data. It connects microscale ice-crystal growth thermodynamics with the macroscale snowpack modeling.ensnowcrystal growthmetamorphismmicrostructure modelinggrain shapeexponential correlation lengthmean curvatureGaussian curvature and micro-computed tomography.thesis::doctoral thesis