Factor Analytic Models of Bioclimate for Canadian Forest Regions
Relational models of bioclimate were formulated for 90 Canadian forest sections defined by J. S. Rowe in 1972. Models were based on component solutions for correlations among climatic attributes believed to be important in tree growth and reproduction. In addition, computer experiments were attempte...
Published in: | Canadian Journal of Forest Research |
---|---|
Main Authors: | , |
Format: | Article in Journal/Newspaper |
Language: | English |
Published: |
Canadian Science Publishing
1974
|
Subjects: | |
Online Access: | http://dx.doi.org/10.1139/x74-078 http://www.nrcresearchpress.com/doi/pdf/10.1139/x74-078 |
Summary: | Relational models of bioclimate were formulated for 90 Canadian forest sections defined by J. S. Rowe in 1972. Models were based on component solutions for correlations among climatic attributes believed to be important in tree growth and reproduction. In addition, computer experiments were attempted to find remedial solutions to problems of model resolution and R/Q-mode equivalence.An attribute model based on physiographic and climatic variables was characterized by mean annual temperature, mean annual precipitation, and July average daily maximum temperature. These factors accounted for 57, 18, and 12% of the total variation on components I, II, and III, respectively.A station model based on weighted factor scores of climatic attributes alone gave a consistent and realistic separation of major forest regions. The first component distinguished Boreal forest from Pacific Coastal, Acadian, and to a lesser degree Great Lake – St. Lawrence forest regions. The second component differentiated Columbian, Grassland, and Montane regions from the Boreal maritime and Pacific Coastal forests. In addition to this generalized model, analysis of a qualitative dataset derived to help overcome problems of nonlinearity in the original data was able to identify the mean summer position of the arctic polar front and a regional low pressure locus over central Alberta.Cluster analysis of forest stations was employed to illustrate the utility of factor models. Limitations and forest applications of our results are discussed. |
---|