| In order to effectively manage forested ecosystems in a sustainable manner, their condition must be characterized and monitored across multiple spatial extents (i.e., stand, watershed, region, etc.). In the last twenty years light detection and ranging (LiDAR) has emerged as an effective tool for measuring many biophysical properties of forested ecosystems including: biomass, basal area, and individual tree dimensions, among others. This dissertation is focused upon further developing LiDAR applications that can fulfill the ever-expanding information needs of sustainable forest management.;Chapters 1 and 2 develop and evaluate a novel object-oriented technique, spatial wavelet analysis (SWA), to automatically estimate the location, height, and crown diameter of individual trees within complex, mixed conifer forests from LiDAR data. In open canopy conditions SWA derived estimates were well correlated with field measures of tree height (r = 0.97) and crown diameter (r = 0.86), and display small errors. As canopy complexity increases SWA tree measurements significantly decrease in accuracy. Specifically, an equivalence test indicates that SWA tree measurements are not within 25% of field-based tree measurements when canopy cover is greater than 50%.;Chapter 3 evaluates the efficacy of k-nearest neighbor (k-NN) imputation models incorporating LiDAR data to predict and map tree-level forest inventory data (individual tree height, diameter at breast height, and species). When compared to an independent forest inventory dataset, the accuracy of stand-level forest inventory metrics was quite high; the root mean square difference of imputed basal area and stem volume estimates were 5.14 m2 ha -1 and 15.99 m3, respectively. However, the accuracy of imputed forest inventory metrics that incorporated small trees ( e.g., total tree density and quadratic mean diameter) was much lower.;Chapter 4 evaluates the use of LiDAR data for characterizing forest successional stages, across a structurally diverse, mixed-species forest in northern Idaho. A variety of LiDAR-derived metrics were used in conjunction with an algorithmic modeling procedure (Random Forests) to classify six stages of three-dimensional forest development and achieved an overall accuracy greater than 95%. The algorithmic model developed ecologically meaningful decision rules based upon LiDAR metrics quantifying mean vegetation height and canopy cover, among others. |
