The study of trends in the dynamics of radial tree growth by the method of generalized canonical correlation analysis (GCCA)

The favorable conditions of the middle taiga for the growth of conifers make the annual tree ring-width variability less informative for dendroclimatology. The creation of generalized dendrochronologies is hampered by the substantial heterogeneity of mean chronologies, the cause of which is the instability of the results obtained from the ring-width time series from different sites. A large number of the time series in these studies hinders proving of the proposed hypotheses by the methods of univariate statistics. At the same time, the alternative multivariate linear methods of data analysis, which reduce the multiplicity problem, generate linear filters through which the time series are skipped. The combination of the linear filters characteristics and the features of the analyzed signals can lead to nonlinear effects and incorrect conclusions.
In this study, the total variability of the dendrochronological series of the Siberian pine (Pinus sibirica Du Tour), Siberian fir (Abies sibirica Ledeb.), and Siberian spruce (Picea obovata Ledeb.) was established by means of Generalized Canonical Correlation Analysis (GCCA). This type of analysis makes it possible to discover correlations of standardized ring-width chronologies with the eigen components $z^s$ as complex natural conditions. The meaning of the principal $z^s$ components was explained by analyzing the coefficients of correlation with environmental parameters: the time series of air temperature and the amount of precipitation previously passed through a low-pass filter in order to prevent non-linear effects. The eigen components $z^2$ and $z^3$ have statistically significant correlations with low frequency fluctuations in air temperature. The principal eigen components $z^1$ and $z^2$ in the conditions of the middle taiga landscapes of Western Siberia have geophysical origin. The components $z^1$ and $z^2$ demonstrated correlations with the anomalies of the average monthly surface air temperatures of the Northern Hemisphere.