| Summary: | Dissertation (Ph.D.) University of Alaska Fairbanks, 2021 Perennial snowfields, such as those found in the Brooks Range of Alaska, are a critical component of the cryosphere. They serve as habitat for an array of wildlife, some of which are crucial for rural subsistence hunters. Snowfields also influence hydrology, vegetation, permafrost, and have the potential to preserve valuable archaeological artifacts. In this study, perennial snowfield extents in the Brooks Range are derived from satellite remote sensing, field acquired data, and snowmelt modeling. The remote sensing data are used to map and quantify snow cover area changes across multiple temporal scales, spatial resolutions, and geographic sub-domains. Perennial snowfield classification techniques were developed using optical multi-spectral imagery from NASA Landsat and European Space Agency Sentinel-2 satellites. A Synthetic Aperture Radar change detection algorithm was also developed to quantify snow cover area using Sentinel-1 data. Results of the remote sensing analyses were compared to helicopter and manually collected field data. Also, a snowfield melt model was developed using an adaptation of the temperature index method to determine probability of melt via binary logistic regression in two dimensions. The logistic temperature melt model considers summer season snow cover area changes per pixel in remotely sensed products and relationships to several independent variables, including elevation-lapse-adjusted air temperature and terrain-adjusted solar radiation. Evaluations of the Synthetic Aperture Radar change detection algorithm via comparison with results from optical imagery analysis, as well as via comparison with field acquired data, indicate that the radar algorithm performs best in small, focused geographic sub-domains. The multi-spectral approach appears to perform similarly well within multiple geographic domain sizes. This may be the result of synthetic aperture radar algorithm dependency on backscatter thresholding techniques and ...
|
|---|