Seasonal forecasting of tropical cyclone activity over the western North Pacific and the South China Sea

This paper presents the development of operational statistical forecasts of seasonal tropical cyclone (TC) activity over the western North Pacific (WNP) and the South China Sea (SCS) based on 30 yr of data (1965-94). Predictors include monthly values of indices representing (a) the El Nino-Southern Oscillation phenomenon, and (b) the environmental conditions over East Asia and the WNP for the months from April of the previous year to March of the current year. Trends and short-term oscillations of the TC activity are also incorporated. The prediction equations are derived from the predictors of individual parameters using the Projection Pursuit Regression technique, which is a statistical method that reduces high-dimensional data to a lower-dimensional subspace before the regression is performed. This technique is found to provide explanations of certain nonlinear variations of the predictands. The predictions from individual parameters are then tested using the jackknife technique. Those predictions that have correlations (with the observed) significant at the 95% level or higher are retained. The values of the correlation coefficients are then used as weights in combining the predictions to form a single forecast of each predictand. The forecasts obtained this way are found to be superior to those from individual parameters. The combined forecast equations are then used to predict the TC activity over the WNP and the SCS for 1997. The prediction is for a slightly above-normal activity for the entire WNP but slightly below normal for the SCS. The former is found to be correct and the latter has the right trend although the activity over the SCS was far below normal, probably as a result of the El Nino of 1997.