Challenges to Estimating Tree Height via LiDAR in Closed-Canopy Forests: A Parable from Western Oregon

Abstract

Abstract We examine the accuracy of tree height estimates obtained via light detection and ranging (LiDAR) in a temperate rainforest characterized by complex terrain, steep slopes, and high canopy cover. The evaluation was based on precise top and base locations for > 1,000 trees in 45 plots distributed across three forest types, a dense network of ground elevation recordings obtained with survey grade equipment, and LiDAR data from high return density acquisitions at leaf-on and leaf-off conditions. Overall, LiDAR error exceeded 10% of tree height for 60% of the trees and 43% of the plots at leaf-on and 55% of the trees and 38% of the plots at leaf-off. Total error was decomposed into contributions from errors in the estimates of tree top height, ground elevation model, and tree lean, and the relationships between those errors and stand- and site-level variables were explored. The magnitude of tree height error was much higher than those documented in other studies. These findings, coupled with observations that indicate suboptimal performance of standard algorithms for data preprocessing, suggest that obtaining accurate estimates of tree height via LiDAR in conditions similar to those in the US Pacific Northwest may require substantial investments in laser analysis techniques research and reevaluation of laser data acquisition specifications.