This study aims to quantify ground ice content and describe the cryostructures and sediment in 15 ice-bearing permafrost cores collected from nine sand-wedge polygons in University Valley. The objectives were reached through laboratory measurements and computed tomodensitometric (CT) scanned image analysis of the permafrost cores. The soils in the valley were predominantly medium-sand. Four types of cryostructures were present in the cores: structureless, suspended, crustal and porphyritic. Excess ice content ranged from 0 to 93%, gravimetric water content ranged from 13 to 1881% and volumetric ice content varied from 28 to 93%. Median excess ice, volumetric ice and gravimetric water contents significantly increased in the top 20 cm of the cores taken from the polygon shoulders with increasing distance from University Glacier. Ground ice was preferentially stored in the centre of the investigated polygons where the ground surface remains cryotic throughout the year. Conversely, higher ground ice contents were measured in the shoulders of the investigated polygon where the ground surface is seasonally non-cryotic. CT-scanned images were shown to reasonably assess the distribution and presence of excess ice in permafrost cores taken from a cold and hyper-arid environment. The results of this thesis provide the first cryostratigraphic study in the McMurdo Dry Valleys of Antarctica.

Ce mémoire a pour objectif de mieux comprendre la tomodensitométrie par rayons-x appliquée à la géocryologie. La majorité des études ayant pour but d’analyser le pergélisol en utilisant la tomodensitométrie sont qualitative. De ce fait, il est principalement question de décrire les cryostructures et la disposition géométrique de la glace dans le pergélisol. Il existe peu d’études quantitatives où le volume de glace est estimé avec la tomodensitométrie. De ce fait les méthodologies quantitatives sont peu développées et leur degré de confiance est mal connu. Des échantillons de pergélisol ont été scannés puis analysés avec le logiciel de visualisation Dragonfly©. Une méthode d’analyse quantitative a été développée et combinée à des mesures en laboratoire afin de déterminer l’erreur entre la proportion en glace estimée sur les Ct-Scans et celle mesurée sur l’échantillon physique. Ensuite, une équation pour essayer de corriger cette erreur a été appliquée. Celle-ci a été déterminée à partir de comparaisons des estimations des volumes de glace sur les Ct-Scans et leurs valeurs mesurées en laboratoire. Ces comparaisons ont également permis d’estimer la quantité de matière organique présente dans un échantillon. Les cryostructures du pergélisol ont été modélisées en 3D. Cela a permis de bonifier la classification Nord-Américaine établie par Murton et French en 1994 et de quantifier les volumes respectifs de glace et de sédiments. / This thesis aims to better understand x-ray computed tomography in geosciences and more particularly its use with frozen ground science. Most studies in permafrost science using x-ray computed tomography were qualitative. Therefore, they mainly describe the permafrost cryostructures and the geometric arrangement of the ice in permafrost. There were few quantitative studies where the volumetric ice content of permafrost was estimated with Ct-Scans. As a result, quantitative methodologies are poorly developed, and their degree of confidence is also poorly assessed. Samples from different locations in the Canadian Arctic were scanned and analyzed using Dragonfly©, a visualization software. A quantitative analysis method has been developed and combined with laboratory measurements to calculate the error between the proportion of ice estimated on the Ct-Scans and the one measured on the physical sample. A correction method was developed and applied to the Ct-scans dataset to reduce the error in volumetric ice content estimations. This equation also allows to give an estimation of the organic matter proportion inside a sample. The permafrost cryostructures were modeled in 3D. This improved the visualization of the cryostructures of the North American classification established by Murton and French in 1994, and it allowed to quantify their volumetric ice and sediment contents.