Climatic Variability Around the Great Lakes Specific to the Problem of Ice Forecasting.

Abstract

This dissertation presents a statistical analysis of climatic variability around the Great Lakes region specific to the problem of ice forecasting. The data set used for this research consists of 81 years of mean monthly temperatures--spanning the period from 1897 to 1977, inclusive--at 25 stations around the Great Lakes. Prior to the statistical analysis, these temperature series were subjected to a homogeneity analysis. The homogeneity analysis revealed many inhomogeneities in the temperature series for which correction factors were estimated. These correction factors were applied to the original series in order to approximate homogeneous conditions. Results of the statistical analysis demonstrate that the Great Lakes region is homogeneous with respect to both short- and long-term fluctuations in air temperature. A simple model, which includes only a regionally average annual harmonic accounts for over ninety percent of the variance in the regional climate. Geographic variation in the long-term mean and amplitude of the annual harmonic explains only an additional 4.8 percent of the total variance in these temperature series. Our results indicate that, although there is no statistical evidence for predictable inter-annual climatic fluctuations that would be useful in forecasting ice conditions on the Great Lakes, significant long-term climatic changes have occurred in the 81 year period between 1897 and 1977. These climatic changes have not been of the nature of a monotonic trend or of smooth oscillations. Rather, climatic change in the Great Lakes region over the past 81 years has taken the form of an alternation between distinct climatic regimes with relatively short transitional periods between them. This regime-like behavior is evident in the mean annual temperature series, the winter and summer seasonal mean series, but not in the spring and fall seasonal series. Climatic changes occurring in the winter and summer seasons are strongly correlated. Further study of these regional patterns of climatic change is needed to determine their cause.