Estimation of critical forest structure metrics through the spatial analysis of airborne laser scanner data

The effective management of complex forest ecosystems depends on quantification of critical forest structure components. Three important structural components include vertical foliage distribution, tree size distribution, and horizontal spatial pattern. Active remote sensing technologies, such as LIght Detection And Ranging (LIDAR), are well-suited for analysis of three-dimensional forest structure. In this research, a methodology was developed to relate the spatial distribution and pattern of LIDAR data to forest structure metrics, through implementation of stochastic modeling and image analysis techniques.; An original approach to estimating the vertical distribution of canopy foliage using multiple return LIDAR is presented. A probabilistically transformed estimate of the canopy foliage profile is derived to approximate model-based profiles developed from field data. Plot-wise goodness-of-fit tests showed the transformed LIDAR-based profile provided an improved estimate of the model-based profile. A methodology is presented for estimating canopy cover and LAI using LIDAR.; Two machine vision algorithms, based upon mathematical morphology and Bayesian object recognition, were developed for the spatially-explicit analysis of tree size distributions using high-density LIDAR. The mathematical morphological analysis of the canopy surface model yielded estimates of tree height that were correlated with field-based measurements (r = 0.80). The simulation-based Bayesian object recognition algorithm provided inferences on plot-wise functionals, including Lorey's height, basal area, stem number and volume. A comparison of the maximum a posteriori estimate with field-based measurements showed mean errors (±1 st.dev.) for: Height 3.9 ± 8.9 ft; DBH 0.4 ± 3.2 in; stem volume 6.9 ± 38.8 ft3 (n = 17). A methodology was developed to quantify the error budget in automated individual tree-based forest surveys.; To investigate horizontal spatial patterns, a novel approach to the automated measurement of forest canopy gaps through the analysis of a LIDAR-based canopy height model was developed. This methodology utilized the theory of binary image analysis to detect gaps, filter out noise, and measure canopy gaps efficiently over a large area. The results were correlated with photogrammetric gap measurements in terms of gap size (r = 0.94) and shape measures ( r = 0.88).; Research presented in this dissertation provides a significant contribution to the understanding of how quantitative measures relating to critical components of forest structure can be obtained through the analysis of LIDAR.