Estimating Canopy Height of Deciduous Forests at a Regional Scale with Leaf-Off, Low Point Density LiDAR

Abstract. As we strive toward a more accurate understanding and quantification of carbon pools in forested ecosystems, the development of regional-scale maps of forest characteristics is essential in order to establish baselines and monitor change. Light Detection and Ranging (LiDAR) is increasingly being used to improve our understanding of forested ecosystems on a broad spatial scale, although obtaining data can be expensive and time consuming. We evaluated the effectiveness of using freely available low point density, leaf-off LiDAR collected for the entire state of Pennsylvania, in the United States, to create an accurate regional-scale dominant/codominant canopy height model for state forests in Pennsylvania. We evaluated several methodologies using an inventory dataset with over 1400 sample plots. The developed canopy height model was accurate to about 10% of the field-measured dominant/codominant tree heights for each plot, although it underestimated the field values. Root mean square error relative to the mean field height ranged between 3.5% and 12.5% across all site and forest variables such as forest community type, age, and height class. Factors that affected the accuracy of the canopy height model included tree density, slope, and percent evergreen cover.