Patterns of forest structural complexity studied with airborne LiDAR data

The three-dimensional arrangement of trees and their crowns is a fundamental attribute of forest ecosystems correlated with many ecological processes and services. I studied forest canopy structure using LiDAR data for a large Pacific Northwest watershed. LiDAR metrics for 95th percentile height, rumple, and canopy density were found to be correlated (Procrustes R=0.72-0.78) with structure measured using common field metrics. Classification of structure for sites representing a range of age classes and forest zones was done based on the variation within the LiDAR data Using the metrics that could be calculated from the LiDAR data, forest structure could not be easily related to specific forest ages for many stands in 70-90 and 220-350 year old age ranges as existing models of forest development would predict. These results would be consistent with forests following multiple pathways of development and call into question the assumption of linear development of forest structure used in many modeling and remote sensing studies.;I calibrated a widely used satellite measure of forest structure, canopy self shadowing, using the same LiDAR data. Canopy self-shadowing was correlated strongly with the canopy's geometric complexity (R2=0.94-0.87), but was less correlated with other common stand measurements: e.g., canopy height R2=0.68 and basal area ha-1 R 2=0.18. These results may explain the moderate correlations between satellite measurements and forest conditions generally reported in the literature. Modeling based on the LiDAR data was used to develop a new empirical adaptive shade compensation (ASC) topographic compensation model that provided more accurate corrections than existing methods. ASC uses measured shadow as an estimate of canopy complexity and a topographic term to describe the illumination geometry as independent variables to determine topographic correction.;Recommendations for future work are made to (I) identify LiDAR metrics that are ecologically meaningful and that can be used across a wide variety of forest types and variations in LiDAR acquisitions, (2) explore hierarchies of scale and to examine forests as nested hierarchies of patches, and (3) ensure that the field sites used to build and test models represent the full diversity of structure present in their study areas.