Application of Multiple Mean Generational Function Method to Typhoon Prediction over the Western North Pacific

摘要

[Objective]The multiple mean generational function (MMGF) method was applied to forecast the annual number of typhoons (TYs) over the Western North Pacific (WNP). [Method]The method yields a number of predictors by mean generational function based on the rolling 50-year data of TYs frequency and sunspot number, and was repeated to generate forecasts year after year by optimal subset regression. [Result] The results showed a reasonably high predictive ability during period 2000-2010, with an average root mean square (RMSE) value of 1.92 and a mean absolute error (MAE) value of 1.64. [Conclusion] Although the MMGF method needs further validation in the practical operation, it already has strong potential for the improvement of skill at forecasting annual frequency of TYs in the WNP.