A Prediction Model for Atlantic Named Storm Frequency Using a Year-by-Year Increment Approach

Abstract

Abstract This paper presents a year-by-year incremental approach to forecasting the Atlantic named storm frequency (ATSF) for the hurricane season (1 June–30 November). The year-by-year increase or decrease in the ATSF is first forecasted to yield a net ATSF prediction. Six key predictors for the year-by-year increment in the number of Atlantic named tropical storms have been identified that are available before 1 May. The forecast model for the year-by-year increment of the ATSF is first established using a multilinear regression method based on data taken from the years 1965–99, and the forecast model of the ATSF is then derived. The prediction model for the ATSF shows good prediction skill. Compared to the climatological average mean absolute error (MAE) of 4.1, the percentage improvement in the MAE is 29% for the hindcast period of 2004–09 and 46% for the cross-validation test from 1985 to 2009 (26 yr). This work demonstrates that the year-by-year incremental approach has the potential to improve the operational forecasting skill for the ATSF.