Thu, Aug 18, 2022: 9:15 AM-9:30 AM
518C
Background/Question/MethodsQuaking aspen (Populus tremuloides) is the most widely distributed tree species in North America. In many regions of the western United States, it serves as the dominant deciduous tree species, providing numerous ecosystem service benefits ranging from biodiversity and wildlife habitat to economic and recreational value. Aspen forests have also been associated with wildfire adaptation and expansion of such forest types has been proposed as a method for managing community risk to wildfire. Foundational to understanding aspen ecology and management is the ability to create high-resolution, accurate maps of aspen presence. Current nationwide maps are based on 30-meter spatial resolution imagery and often mislabel aspen stands, with estimated error rates from current databases of 50-65% across the Southern Rocky Mountains ecoregion (U.S. Environmental Protection Agency Level III). With increasing focus on quaking aspen management and ecology in the face of a changing climate and altered disturbance regimes, there is critical need for accurate and reproducible methods to map this important species across its range. Here we present a novel method for classifying quaking aspen trees using seasonal composite imagery and chlorophyll vegetation indices derived from thousands of Sentinel-2 Multispectral Instrument (MSI) tiles in the Google Earth Engine platform.
Results/ConclusionsWith model accuracy ranging from 90-95% across the Southern Rockies and a spatial resolution of 10 meters, these methods achieve significant improvements over existing maps of quaking aspen presence. Further, derived spectral characteristics from multi-seasonal image composites provide context to the seasonal changes in quaking aspen forests and can be applied to other species of interest. The Google Earth Engine platform allows us to scale these methods across broad geographic regions, providing critical maps of quaking aspen presence that will be useful for federal, state, and local land managers and applied research of the ecology, fire effects, and habitat suitability of this important tree species throughout the western United States.
Results/ConclusionsWith model accuracy ranging from 90-95% across the Southern Rockies and a spatial resolution of 10 meters, these methods achieve significant improvements over existing maps of quaking aspen presence. Further, derived spectral characteristics from multi-seasonal image composites provide context to the seasonal changes in quaking aspen forests and can be applied to other species of interest. The Google Earth Engine platform allows us to scale these methods across broad geographic regions, providing critical maps of quaking aspen presence that will be useful for federal, state, and local land managers and applied research of the ecology, fire effects, and habitat suitability of this important tree species throughout the western United States.