Year of Award


Document Type


Degree Type

Master of Science (MS)

Degree Name


Department or School/College


Committee Chair

Anna Klene

Commitee Members

Erich Peitzsch, Lloyd Queen


Glacier National Park, Avalanche Paths, Remote Sensing, Vegetation, Random Forests, Disturbance


University of Montana

Subject Categories

Physical and Environmental Geography | Remote Sensing | Spatial Science


Snow avalanches are the common form of mass wasting in the high mountain environments of Glacier National Park (GNP), Montana. These natural disturbances play important roles in mountain ecosystems by regularly disturbing montane systems, providing critical habitat for some species, transporting debris, and influencing vegetation and fire dynamics. Since the 1900s, natural avalanche-related activity recorded along important transportation corridors within the park has frequently disrupted transportation.

While many of the steep slopes of GNP are susceptible to avalanching, formal inventories exist only for small, critical portions of the park and they vary substantially from one another. GNP’s protected status does not allow for avalanche mitigation, allowing this area to serve as a natural mountain environment for studying these processes. A current, high-resolution inventory of avalanche locations in the park is needed for the entirety of the Park.

Imagery and digital elevation models (DEMs) were used to map the distinct biogeographic and topographic patterns left by avalanching using machine learning methods. Mosaics of National Agricultural Imagery Program (NAIP) aerial photographs acquired in 2013 were segmented to map avalanche tracks. Principal components from the imagery and derivatives of the DEM were used as input to a Random Forests algorithm which mapped the most likely class for each segment using a probabilistic approach. Avalanche paths were found to comprise approximately 5-12% of the park, along predominantly south and southeasterly facing slopes between 20° to 40°. While this estimate is similar to previous studies, this work did not map starting or runout zones which would have increased the total area. The paths predicted provide a comprehensive inventory that can be used to monitor shifts in vegetation and climate dynamics within the disturbance regime. Changes were clearly seen in the contraction and expansion of trim lines of some avalanche paths in recent imagery. Future research could use this work as a baseline for time-series analysis.



© Copyright 2019 Morgan Voss