Climatic reconstruction from tree rings at Banff

Abstract

This study describes new techniques of tree-ring data preparation and data analysis for deriving proxy climate data from senescent Douglas-fir (Pseudotsugamenziesii var. glauca (Beissn.) Franco) trees from the Canadian Rockies, near Banff, Alberta. Fifteen annual tree-ring variables were measured by X-ray densitometry for 429 years (1550–1978) for 12 increment cores. Ring variable data were reduced to standard indexes using a 99-year normally weighted digital filter. Missing ring values were estimated using correlation with younger and more vigorous specimens, and each ring variable data set (12 cores × 429 years) was reduced to its first and second principal component score, to be used in the development of response and transfer functions. Factor analysis identified six subsets of ring variable principal component scores. The best multiple regression equations for transferring tree-ring variable principal components into reconstruction of climate were identified by screening all possible combinations of principal component scores between factor groups. Annual climate variables, such as total precipitation, did not transfer as successfully as did the shorter-term climate variables like June–July precipitation (R² = 0.36 compared with 0.51). Verified transfer functions were developed for five climate variables which can now be reconstructed to 1550 a.d. (429 years).