Due to the heterogeneous nature of alpine snow distribution, advances in hydrological monitoring and forecasting for water resource management require an increase in the frequency, spatial resolution and coverage of field observations. Such detailed snow information is also needed to foster advances in our understanding of how snowpack affects local ecology and geomorphology. Although recent use of structure-from-motion multi-view stereo (SFM-MVS) 3D reconstruction techniques combined with aerial image collection using drones has shown promising potential to provide higher spatial and temporal resolution snow depth data for snowpack monitoring, there still remain challenges to produce high-quality data with this approach. These challenges, which include differentiating observations from noise and overcoming biases in the elevation data, are inherent in digital elevation model (DEM) differencing. A key issue to address these challenges is our ability to quantify measurement uncertainties in the SFM-MVS snow depths which can vary in space and time. The purpose of this thesis was to develop data-driven approaches for spatially quantifying, characterizing and reducing uncertainties in SFM-MVS snow depth mapping in alpine areas. Overall, this thesis provides a general framework for performing a detailed analysis of the spatial pattern of SFM-MVS snow depth uncertainties, as well as provides an approach for correction of snow depth errors due to changes in the sub-snow topography occurring between survey acquisition dates. It also contributes to the growing support of SFM-MVS combined with imagery acquired from drones as a suitable surveying technique for local scale snow distribution monitoring in alpine areas.