Simulation and Forecasting of Lake Erie Storm Surges

Abstract

A numerical model based on the impulse response function method is used to hindcast and forecast storm surges on Lake Erie. The impulse response function method is more efficient then numerical integration of the dynamic equations when results are required at only a few grid points. Hindcasts use wind observations from seven weather stations around Lake Erie. The surge phenomenon depends on the two-dimensional structure of the wind field and on the stability of the atmospheric boundary layer over the lake. The overall correlation coefficient between computed and observed water level deviations for 15 five-day hindcast cases is 0.83 at eight water level recording stations. Operational Great Lakes wind forecasts are used to drive the model for water level forecasts at Buffalo, NY, and Toledo, OH. The accuracy of the water level forecasts is currently limited by the accuracy of the forecast winds.