Fine-scale, multidimensional spatial patterns of forest canopy structure derived from remotely sensed and simulated datasets

• Main Author: Frazer, Gordon Wilson
• Other Authors: Niemann, K. O.
• Language: English; en
• Published: 2010
• Subjects: Forest canopy; Remote sensing; UVic Subject Index::Sciences and Engineering::Agriculture::Forests and forestry
• Online Access: http://hdl.handle.net/1828/2496

Forests are not simply storehouses of timber or wood fibre for human consumption and economic development. They represent structurally and ecologically rich habitat for an estimated 40 percent of the earth's extant species, and form the functional interface between the biosphere and atmosphere for some 27 percent of the earth's terrestrial surface. Forests, therefore, play a vital role in the maintenance of biodiversity and the regulation of local to global scale ecosystem processes and functions. Present strategies for conserving biodiversity in managed forests are based on the notion that maintaining the full range of structural conditions historically present in natural forests is the best approach for assuring the long-term persistence of a broad range of native species. The overarching goal of this dissertation is to contribute to the development of novel forest measurements that are relevant to organisms and ecosystems, and much needed by forest scientists and managers to recognize and retain the key elements and patterns of forest structure that are crucial for the conservation of forest biodiversity. This study focuses explicitly on fine-spatial-scale, multidimensional patterns of forest canopy structure based on the assumption that the 'canopy' is the primary focal site of complex interactions between vegetation and the physical enviromnent. Two disparate remote sensing technologies-ground-based hemispherical (fisheye) canopy photography and airborne discrete-return LiDAR-are employed to characterize angular, vertical, and horizontal patterns of forest canopy structure. A quantitative technique is developed for precise measurements of gap fraction (P), element clumping (O), mean projection coefficient (G), and leaf area index (L) from sequences (sets) of black and white pixels extracted at specific view angles in digital fisheye photos. Results are compared with three other leading techniques and validated using well-documented simulated and real fisheye photosets. Variables P, O, G, and L control light capture and penetration in forest canopies, and are key input parameters for process-based models of stand productivity, stand dynamics, and material (CO2 and H20) and energy fluxes between the canopy and atmosphere. Findings show that this new technique consistently produced the best estimates of stand LAI in each of the three experimental forest sites. However, further validation work is required to determine the adequacy of these methods in other closed and discontinuous canopies. Finally, a methodological framework is devised for quantifying, classifying, and comparing fine-spatial-scale vertical and horizontal patterns of canopy structure derived from airborne LiDAR data. This methodology is tested with simulated forest canopies and ultimately demonstrated using an airborne LiDAR dataset collected over very young to old, coastal Douglas-fir/western hemlock forests on Vancouver Island, British Columbia. A pseudo 'space-for-time substitution' sampling approach is used to investigate age-related developmental changes in canopy structure at decadal and century time scales. Discrete classes of vertical and horizontal canopy structure are identified by k-means partitioning. The structural differences found among age-classes were consistent with the characteristics, patterns, and dynamics predicted by generalized models of stand development for similar coastal Douglas-fir/western hemlock forests of northwestern North America.