Forward modeling of tree-climate relations across the Northern Hemisphere

Abstract

This thesis uses the Vaganov-Shashkin model of tree ring formation, a multivariate, nonlinear, mechanistic model that directly predicts tree-ring growth using climate data, to simulate tree-ring formation across the Northern Hemisphere. Previous research has shown the model has skill in reproducing ring-width variability and climate sensitivity at local and regional scales, but its ability to simulate the major geographical differences in tree-climate relationships at a hemispheric scale has not yet been tested. In this study, we ran the model at over 7,000 locations across the Northern Hemisphere, and compared the seasonal climate responses of the simulations against a network of nearly 2,200 real tree-ring width records. We also calculated the predicted dominant factor at each location and used relative growth rates to explain these patterns. Simulated tree-ring chronologies are consistent with the real ones in the seasonality and relative strength of the encoded climate signals, demonstrating that the model has skill in reproducing tree-ring growth response to climate variability across the Northern Hemisphere. Because the simulations were produced using only climate records and the same set of parameters, the fact that the model was able to reproduce major geographical differences in the observations suggests that climate is the primary factor in determining large-scale tree-climate relationships. We also used relative growth rates to show the sequence of events during the growing season and the possible mechanism of the climate response of tree rings. We found that temperature dominates growth at temperature-sensitive sites during most of the growing season and that at stations where temperature dominates growth at the end of growing season, summer precipitation generally has a strong positive influence on tree-ring formations, while at locations where soil moisture limits growth at the end of growing season, ring widths usually have a positive correlation with winter precipitation. Because the model has skill in reproducing ring widths and tree-climate relationships at local, regional and hemispheric scales, we suggest VSM can potentially be used as a low-cost estimator to predict tree-ring response to climate prior to sampling and to forecast long-term changes in tree-climate relationships.