A Forecast Model for Great Lakes Water Levels

Abstract

Spectral analysis of Great Lakes water level elevations reveals the presence of prominent cycles with periods of approximately 1, 11, 22 and 36 years. These frequency components can be isolated and described with respect to their amplitudes and phases. Because these fluctuations are natural phenomena that have recurred throughout recorded history and are expected to continue into the future, they can be extrapolated and combined to provide a forecast predicting future stages of high and low lake levels. Extreme water levels are quite rare because these can only occur when peaks or troughs of all components are aligned. However, since most cycles have peaks and troughs extending over a number of years they need only be approximately in phase to create marked changes in lake levels. Forecasts appear sufficiently reliable to be useful for anticipating episodes of higher or lower than average water levels and should be utilized for future long-term shoreline planning policies.