Western redcedar on the Olympic Peninsula: Locating this culturally, economically, and ecologically important species using remote sensing methodologies

Abstract

Western redcedar (Thuja plicata Donn ex D. Don) is a culturally, economically, and ecologically important species native to the Pacific Northwest. Abundance of western redcedar on the landscape today is a fraction of what it was historically due to a combination of overharvesting in the 20th century, incompatibility with short-rotation forestry, cedar browse by ungulates, and a documented western redcedar dieback phenomenon linked to climate change. Restoring cedar will rely on our ability to map western redcedar trees on the landscape, particularly using tools such as remote sensing that have larger spatial coverage capabilities. This study used metrics derived from UAV (unmanned aerial vehicle) multispectral imagery and LiDAR to train a model to identify and map western redcedar trees on the west side of the Olympic Peninsula, with a focus on the T3 Watershed Experiment in the Olympic Experimental State Forest. Models were trained based on different combinations of data collection years, study areas, multispectral imagery and LiDAR, and the portion of the crown represented by LiDAR point clouds clipped for each tree (small cylinder vs. only the top 3 m). Generally, models achieved the highest accuracies when subset by study area and when only the top 3 m of the LiDAR trees were used. The best and most robust model achieved an accuracy of 86%. In addition to species classification, indices derived from the multispectral imagery were used as proxies to assess cedar health. However, no clear conclusions could be drawn due to confounding factors related to study area differences, the data limitations of using spectral signatures for this type of work, and the complexities inherent in tree health; all of which were discussed in depth for use in future research. Overall, this study demonstrated that UAV LiDAR and multispectral imagery are powerful tools in mapping western redcedar trees over relatively large areas, including robust methodology on how to achieve this work on an individual tree basis.