Tolson, Bryan A

Understanding snowpack variability is important as it plays an imperative role in environmental, hydrologic, and atmospheric systems. Research questions related to three linked areas were investigated in this thesis: 1) scaling issues in snow hydrology, 2) forest-snowpack relationships, and 3) methods of integrating snow water equivalent (SWE) into a hydrologic model for a large, forested drainage basin in central Ontario. The first study evaluated differences in SWE across process, measurement, and model scales. Point scale snowpack measurements could be bias corrected using scaling factors derived from a limited number of transect measurements and appropriately stratified point scale measurements may be suitable for replacing transect scale mean SWE when transect data are not possible to collect. Comparison of modelled products to measurements highlighted the importance of understanding the spatial representativeness of in-situ measurements and the processes those measurements represent when validating snow products or assimilating data into models.The second study investigated the efficacy of field-based, and remotely sensed datasets to describe forest structure and resolve forest-snowpack relationships. Canopy cover was highly correlated with melt rate and timing at the site scale however, significant correlations were present in 2016 but not 2017, which was attributed to interannual differences in climate. Peak SWE metrics did not correlate well with forest metrics. This was likely due to mid-winter melt events throughout both study years, where a mix of accumulation and melt processes confounded forest-snowpack relationships. The third study evaluated the accuracy of the Copernicus SWE product and assessed the impact of calibrating and assimilating SWE data on model performance. The bias corrected Copernicus product agreed with measured data and provided a good estimate of mean basin SWE. Calibration of a hydrologic model to subbasin SWE substantially improved modelled SWE performance. Modelled SWE skill was not improved by assimilating SWE into the calibrated model. All models evaluated had similar streamflow performance, indicating streamflow in the study basin can be accurately estimated using a model with a poor representation of SWE. The findings from this work improved knowledge and understanding of snow processes in the hydrologically significant Great Lakes-St Lawrence Forest region of central Ontario.

Author Keywords: data assimilation, hydrologic model, multi-objective calibration, remote sensing, scale, snow