HR: 1340h
AN: B43B-0277    [Abstracts]
TI: A Simple Approach to Global Classification of Tree Plant Functional Types Using Cluster Analysis
AU: * Wang, A
EM: awang@nrcan.gc.ca
AF: Department of Renewable Resources, University of Alberta, 751 General Services Building, Edmonton, AB T6G 2H1 Canada
AU: Price, D T
EM: dprice@nrcan.gc.ca
AF: Natural Resources Canada, Northern Forestry Centre, 5320-122 St., Edmonton, AB T6H 3S5 Canada
AB: Classification of vegetation species diversity into a restricted number of Plant Functional Types (PFT) simplifies modelling of large-scale (global) vegetation dynamics. One purpose of such classification is to quantify relationships between a PFT's geographic distribution and environmental factors to identify structural and functional PFTs at global scale. Here, we aimed at a simple integrated algorithm to relate climate space to global distributions of tree PFTs. Multi-variance clustering was used to analyze the statistical homogeneity of the climate space where individual tree PFTs exist. In this way, primary tree PFTs identified from the satellite-based GLC2000 classification were separated into tropical, temperate and boreal PFTs for use in the Canadian Terrestrial Ecosystem Model (CTEM). Global datasets of monthly minimum temperature, growing degree days, an index of climatic moisture and estimated PFT cover fractions were used as variables in the cluster analysis. The sequences of statistical analysis procedures included a comparison of the K-means and Fuzzy C-means clustering algorithms, followed by cluster validation and merging. Overall, K-means produced clusters with narrower dispersions about their centers. The optimal number of clusters for K-means was determined from the Cubic Clustering Criterion, Pseudo F and R2, combined with examination of geographical distributions. Where the optimal number of clusters exceeded the number of desired PFTs, clusters were merged following a decision rule based on overlapping temperature ranges. The statistical results for individual PFT clusters were found consistent with those from other global scale PFT classifications. As an improvement of the quantification of the climatic limitations on PFT distributions, the results also demonstrated overlapping between PFT cluster boundaries that reflected vegetation transitions, e.g., between tropical and temperate biomes. The resulting global database should provide a better basis for simulating the interaction of climate change and terrestrial ecosystem dynamics including carbon cycling using global vegetation models.
DE: 0400 BIOGEOSCIENCES
DE: 0414 Biogeochemical cycles, processes, and modeling (0412, 0793, 1615, 4805, 4912)
DE: 0430 Computational methods and data processing
DE: 0439 Ecosystems, structure and dynamics (4815)
SC: Biogeosciences [B]
MN: Fall Meeting 2005