Abstract
Accurate and spatially explicit measurements of forest attributes are critical for sustainable forest management and for ecological and environmental protection. Airborne Light Detection and Ranging (lidar) systems have become the dominant remote sensing technique for forest inventory, mainly because this technology can quickly provide highly accurate and spatially detailed information about forest attributes across entire landscapes. This dissertation is focused on developing and assessing novel and advanced methods for three dimensional (3-D) forest characterization. Specifically, I map canopy structural attributes of individual trees, as well as forests at the plot and landscape levels in both natural and industrial plantation forests using lidar remote sensing data.
Chapter 1 develops a novel framework to automatically detect individual trees and evaluates the efficacy of k-nearest neighbor (k-NN) imputation models for estimating tree attributes in longleaf pine (Pinus palustris Mill.) forests. Although basal area estimation accuracy was poor because of the longleaf pine growth habit, individual tree locations, height and volume were estimated with high accuracy, especially in low-canopy-cover conditions. The root mean square distance (RMSD) for tree-level height, basal area, and volume were 2.96%, 58.62%, and 8.19%, respectively.
Chapter 2 presents a methodology for predicting stem total and assortment volumes in industrial loblolly pine (Pinus taeda L.) forest plantations using lidar data as inputs to random forest models. When compared to reference forest inventory data, the accuracy of plot-level forest total and assortment volumes was high; the root mean square error (RMSE) of total, commercial and pulp volume estimates were 7.83%, 7.71% and 8.63%, respectively.
Chapter 3 evaluates the impacts of airborne lidar pulse density on estimating aboveground biomass (AGB) stocks and changes in a selectively logged tropical forest. Estimates of AGB change at the plot level were only slightly affected by pulse density. However, at the landscape level we observed differences in estimated AGB change of >20 Mg·ha−1 when pulse density decreased from 12 to 0.2 pulses·m−2. The effects of pulse density were more pronounced in areas of steep slope, but when the DTM from high pulse density in 2014 was used to derive the forest height from both years, the effects on forest height and subsequent AGB stocks and change estimates did not exceed 20 Mg·ha−1.
Chapter 4 presents a comparison of airborne small-footprint (SF) and large-footprint (LF) lidar retrievals of ground elevation, vegetation height and biomass across a successional tropical forest gradient in central Gabon. The comparison of the two sensors shows that LF lidar waveforms are equivalent to simulated waveforms from SF lidar for retrieving ground elevation (RMSE=0.5 m, bias=0.29 m) and maximum forest height (RMSE=2.99 m; bias=0.24 m). Comparison of gridded LF lidar height with ground plots showed that an unbiased estimate of aboveground biomass at 1-ha can be achieved with a sufficient number of large footprints (> 3).
Lastly, Appendix A presents an open source R package for airborne lidar visualization and processing for forestry applications.