A wrapped-surface reconstruction method of lidar points to identify tree crown attributes

Identifying tree attributes from aerial photos or high resolution satellite images is difficult due to shading effects and image distortion. This research employed small footprint Light Detection and Ranging (LIDAR) data to derive precise tree crown structural parameters. LIDAR provides three-dimensional point distributions which captures structural measurements of trees and other objects. The conventional approach fitted assumed geometric shapes over a set of discrete LIDAR points using a regression approach. Differences between observed and geometrically fitted values can be significant, especially with asymmetrical or irregular crown structures. In this study, we took a graphical approach and created a wrapped surface over LIDAR points through implicit surface reconstruction to identify irregular tree crown shapes and corresponding shape parameters. We applied this method in Washington Park Arboretum in Seattle, WA, for which two different point densities of LIDAR data (from 1 pt up to 20 pts per square meter) have been collected. The tree species in the arboretum are common to the Pacific Northwest region. The value of high and low density LIDAR point data were compared as inputs to the method. The wrapped surface method results in better crown shape parameter estimation for tree height and crown volume. Using high point density LIDAR data as an input to the wrapped-surface method identifies unique crown structural attributes that are unrecognized using standard geometrical shape models.