https://hdl.handle.net/1880/120245

The distribution of permafrost in mountainous regions is influenced by various factors such as topography, climate, vegetation, and substrate. Alpine landforms, in particular, exhibit unique thermal characteristics, where the presence of coarse sediments creates a distinctive ground thermal regime. These sediments can maintain lower temperatures compared to adjacent fine-grained soils, primarily due to natural air convection and lower thermal conductivity. Existing statistical and physically based models often fail to accurately represent permafrost occurrence in alpine landforms as they do not account for these effects. This research overcomes these limitations by developing tools incorporating these effects, thereby improving the prediction and understanding of permafrost distribution in mountainous regions. The study utilized a combination of drone photogrammetry, machine learning, and image processing to create an efficient and accurate workflow for estimating grain size distributions from orthomosaic images with a ground sampling distance of around 3 cm. Data-driven techniques, such as logistic regression, support vector machines, and random forests, were employed to analyze the spatial distribution of permafrost in six alpine basins within the Canadian Rockies. Additionally, natural convection was integrated into a hydrological model to enhance the representation of the thermal behavior of coarse sediments. The results revealed that localized conditions strongly influence permafrost variability, highlighting the importance of local field data for accurate predictions. Additionally, variables such as enhanced vegetation index, sediment size, and slope angle were identified as the most important factors for predicting permafrost at the patch scale. Testing the developed convection-enhanced hydrological model on a talus slope in the Canadian Rockies demonstrated that coarse-grained sediments can significantly lower ground temperatures when natural convection is considered. The study also identified natural convection in alpine landforms operating under a local thermal non-equilibrium (LTNE) state, and that applying local thermal equilibrium conditions hinders the impact of natural convection. Overall, this research improves the understanding of permafrost dynamics in alpine landforms and enables a more accurate analysis of permafrost extent and its influence on groundwater discharges.

https://hdl.handle.net/1880/120245