Improved techniques for measuring and estimating scaling factors used to aggregate forest transpiration
MetadataShow full item record
AbstractThis research deals with transpiration scaling issues, and its aim is to improve five boreal species canopy transpiration estimates that are computed by scaling up single tree transpiration to the canopy scale. The improvement of canopy transpiration estimates is made by developing a robust scaling approach. The robustness of the scaling approach rests on fine input scaling parameter data and allometric regression models developed at two different scales, tree and plot (i.e. canopy scale). Moreover, the scaling approach integrates the habitat's vegetation heterogeneity by developing regression models for each species, and by adapting the scaling process to the particular allometric characteristics of each species. The scaling approach has three spatial scales: microscopic, tree, and plot. The microscopic scale was used to accurately measure tree sapwood depth by means of microscopical wood tissue analysis. Individual sapwood depth variations around the tree trunk were also observed and quantified. There were interspecific allometric differences showing that a tree's sapwood area does not always grow as the tree grows. At the plot scale, pure and mixed vascular vegetation plots of 60x60m and lOxlOm were delimited. Plot's tree quantity and outside bark circumference at the breast height were recorded. LAI was measured using the Tracing Radiation and Architecture of Canopies (TRAC) and the LAI-2000 optical devices. The results helped to generate the robust regression models. At the tree scale, regression models were fitted between sapwood depth and outside bark diameter at the breast height (DB HOB) to later estimate tree and plot sapwood area. However, not all the models developed were linear relationships. Results for Pinus banksiana, Pinus contorta, and Picea mariana did not lead to linear relationships, while results for Poplllulus tremuloides and Picea glauca did provide strong linear relationships. These results prove that not all vascular species sapwood depth is directly proportional to their respective DB HOB. Thus, two approaches to aggregate sapwood area to the plot scale were combined. The new combined approach drew strong linear correlations at the plot scale between sapwood area and leaf area. This last outcome conclusively proves the theory claiming a linear correlation between sapwood area and leaf area at different scales, where a lack of conclusive proof existed before. The heat dissipation technique was used to collect diurnal tree sap flow and estimate transpiration using sapwood area as the scaling parameter. Single tree sap flow was aggregated to the plot scale using the plot's sapwood area estimates. Since vegetation transpiration rates vary among species, mixed forest transpiration is therefore influenced by vegetation heterogeneity. Thus, the internal plot's vegetation heterogeneity was included in the scaling approach. Additionally, tree sap flow radial variations were computed to provide a correction to in situ measurements. The final canopy transpiration estimates were compared with the canopy actual evapotranspiration which was estimated using the Penman-Monteith equation. Canopy transpiration was found to be a large proportion of the canopy's actual evapotranspiration, normally greater than the 50%. This improved scaling approach includes the error propagation estimation, and showed that the error associated with a plot's leaf area estimate increases with the plot size. The error associated with tree and plot scale sapwood area estimates is practically null. These demonstrate that the error associated with the biometrics can be significantly minimized by using the most robust mensuration methods that currently exist. Overall, the dissertation outcomes demonstrate that the use of robust methods and the careful formulation of the scaling approach were fundamental in obtaining reliable transpiration estimates. It is recommended that prior characterization of the intraspecific biometrics variations be made in order to develop an adequate scaling approach.
Bibliography: p. 217-233