Applications of Terrestrial LiDAR for Reducing Uncertainty in Forest Carbon Mapping: A Scale-Driven Approach
Stovall, Atticus, Environmental Sciences - Graduate School of Arts and Sciences, University of Virginia
Shugart, Herman, Department of Environmental Sciences, University of Virginia
Epstein, Howard, Department of Environmental Sciences, University of Virginia
Scanlon, Todd, Department of Environmental Sciences, University of Virginia
Forest biomass accounts for the vast majority of aboveground terrestrial carbon storage, but biomass magnitude and spatial distribution is highly uncertain. Global biomass estimates are inextricably linked to tree-level estimates of biomass by way of allometry - an indirect relationship relating tree stem diameter and/or height to destructively harvested dry weight. The difficulty and cost of creating allometric equations has led to spatially biased, low-certainty relationships that limit confidence in global carbon estimates. Efficient, non-destructive biomass estimation with terrestrial laser scanning (TLS) or terrestrial LiDAR can potentially improve single-tree, plot-level, and global biomass estimates through three-dimensional modeling, but little is known of how this approach impacts uncertainty at these spatial scales. With a scale-driven analysis, this work explores the potential for TLS to reduce uncertainty in biomass estimates from tree to landscape. At the single-tree scale, a novel algorithm - the Outer Hull Model (OHM) - was developed and validated with 21 destructively harvested Pinus contorta trees in the Colorado State Forest. The OHM accurately estimated component (e.g. trunk, branch, foliage) and whole-tree biomass, outperforming other approaches. In a broadleaf deciduous Virginia forest, TLS was used to model over 300 trees for developing species-specific allometry and quantifying errors in commonly used national allometry. TLS-derived non-destructive allometry had lower uncertainty than the national equations. A allometric sample-size-based sensitivity analysis was conducted with and without trees above 50 cm diameter, revealing a strong dependency on large trees for accurate biomass prediction with allometry. The dependency of airborne and spaceborne LiDAR biomass estimates on plot-level calibration provided an avenue for landscape-scale improvements in uncertainty with TLS. TLS-based LiDAR calibration was compared to traditional methods with three different models based on mean canopy height and return intensity. TLS reduced uncertainty primarily by accurate direct estimates of standing biomass, improvements in allometric uncertainty, and reduced RMSE LiDAR calibration. The scale-driven approach of this work emphasizes the need for improved allometry in forest ecosystems, especially when high quality calibration and validation data is needed - a goal that can be realized with the strategic deployment of TLS in high uncertainty environments.
PHD (Doctor of Philosophy)
Forest, Carbon, Uncertainty, LiDAR, Biomass, Allometry, Allometric Scaling Theory, Radar, Calibration, Validation, Satellite , Tree, Algorithm, Volume, Ecology, Remote Sensing, Terrestrial Laser Scanning, Wood density
English
All rights reserved (no additional license for public reuse)
2017/11/30