Predicting Live and Dead Tree Basal Area in Bark Beetle-AffectedForests from Discrete-Return LiDAR

Publication Year:
2012
Usage 460
Downloads 271
Abstract Views 189
Repository URL:
http://digitalcommons.unl.edu/usdafsfacpub/218
Author(s):
Hudak, Andrew T.; Bright, Benjamin C.; Negron, Jose; McGaughey, Robert; Andersen, Hans-Erik; Hicke, Jeffrey A.
article description
Recent bark beetle outbreaks in western North America have been widespread and severe. High tree mortality due to bark beetles affects the fundamental ecosystem processes of primary production and decomposition that largely determine carbon balance (Kurz et al. 2008, Pfeifer et al. 2011, Hicke et al. 2012). Forest managers need accurate data on beetle-induced tree mortality to make better decisions on how best to remediate beetle-affected forests and restore healthy ecosystem services (Negron et al. 2008). Discrete-return LiDAR measures canopy height and density, and LiDAR intensity provides some indication of the spectral reflectance and condition of canopy elements (foliage and branches) (Kim et al. 2009). LiDAR has been successfully applied to estimate biomass and carbon stocks in healthy forest (Hudak et al. 2012) and beetle-affected forest (Bright et al. 2012). A challenge in beetle-affected forests is that most airborne LiDAR has a single near infrared wavelength; i.e., LiDAR lacks the multispectral information useful for distinguishing between green, red, and grey trees. However, LiDAR intensity values may help distinguish between live green and dead red or grey trees. Moreover, mountain pine beetles (the most widespread bark beetle currently) and spruce beetles preferentially attack larger trees, so beetles impart a canopy structural signature that may be exploited (Coops et al. 2009).Our objective is to predict Live and Dead Basal Area (BA) in beetle-affected areas in five states in the USA using canopy height, density, intensity, and topographic metrics derived from discrete-return airborne LiDAR data, tree measurements collected in field plots and summarized into plot-level estimates of Live BA and Dead BA using the Forest Vegetation Simulator (FVS), and the nonparametric Random Forests (RF) machine learning algorithm (Breiman 2001). Predicting both Live and Dead BA in bark beetle-affected forest, where live and dead trees are typically thoroughly mixed, has not been attempted before and should provide insight into the sensitivity of LiDAR to bark beetle effects on coniferous forest canopies.