Empirical Orthogonal Function Analysis of the Track of Tropical Storms and Typhoons and Its Application to Forecasting

Abstract

   The statistical technique for analyzing and forecasting the track of tropical storms and typhoons is described. Twenty-five years of data (1951-75) for the western North Pacific were evaluated to determine the track characteristics. The results show that the first two out of possible twenty-four eigenvectors account for 75 per cent of the observed variance. The coefficients of the selected eigenvectors were used as the predictands in the development of the forecasting equations. Climatological parameters and 5-day mean 500 mb heights were employed as the potential predictors. Using these predictands and potential predictors at the formation of a tropical storm, forecasting equations for the whole track were derived by regression technique.