Characterizing vegetation and return periods in avalanche paths using lidar and aerial imagery

Abstract

ABSTRACT Snow avalanches are a hazard and ecological disturbance across mountain landscapes worldwide. Understanding how avalanche frequency affects forests and vegetation improves infrastructure planning, risk management, and avalanche forecasting. We implemented a novel approach using lidar, aerial imagery, and a random forest model to classify imagery-observed vegetation within avalanche paths in southern Glacier National Park, Montana, USA. We calculated spatially explicit avalanche return periods using a physically based spatial interpolation method and characterized the vegetation within those return period zones. The automated vegetation classification model differed slightly between avalanche paths, but the combination of lidar and spectral signature metrics provided the best accuracy (88–92 percent) for predicting vegetation classes within complex avalanche terrain rather than lidar or spectral signature metrics alone. The highest frequency avalanche return periods were broadly characterized by grassland and shrubland, but the influence of topography greatly influences the vegetation classes as well as the return periods. Furthermore, statistically significant differences in lidar-derived vegetation canopy height exist between categorical return periods. The ability to characterize vegetation within various avalanche return periods using remote sensing data provides land use planners and avalanche forecasters a tool for assessing the spatial extent of large-magnitude avalanches in individual avalanche paths.