Abstract
Correlations between the water supplies to each of the Great Lakes and prewhitened tree-ring chronologies from 16 sites around the Great Lakes suggested some strong associations for the summer months, particularly June and July. Some of these associations were, however, nonlinear and variables had to be log-transformed. Regression models for June–July water supplies totals were developed for each lake. These models explained between 45 and 56% of the variance in the dependent datasets; they also performed satisfactorily on independent data sets. The models were therefore used to estimate past water supplies to each of the lakes as far back as the chronologies in the models permitted (about 200 years). The association between variations in supplies to the individual lakes during this century was compared to that of the reconstructed past supplies and was found to be similar. Variations in the water supplies to Lake Superior are to some degree independent of those to the lower lakes—which makes the... Abstract Correlations between the water supplies to each of the Great Lakes and prewhitened tree-ring chronologies from 16 sites around the Great Lakes suggested some strong associations for the summer months, particularly June and July. Some of these associations were, however, nonlinear and variables had to be log-transformed. Regression models for June–July water supplies totals were developed for each lake. These models explained between 45 and 56% of the variance in the dependent datasets; they also performed satisfactorily on independent data sets. The models were therefore used to estimate past water supplies to each of the lakes as far back as the chronologies in the models permitted (about 200 years). The association between variations in supplies to the individual lakes during this century was compared to that of the reconstructed past supplies and was found to be similar. Variations in the water supplies to Lake Superior are to some degree independent of those to the lower lakes—which makes the...