A Poisson regression model approach to predicting tropical cyclogenesis in the Australian/southwest Pacific Ocean region using the SOI and saturated equivalent potential temperature gradient as predictors

Abstract

This paper explores the potential of the Southern Oscillation index (SOI) in combination with the saturated equivalent potential temperature gradient (EPT) as predictors of tropical cyclogenesis in the Australian/southwest Pacific Ocean. This is undertaken using a series of Poisson regression models of tropical cyclogenesis developed on a 2° latitude × 5° longitude and monthly grid. Links between tropical cyclogenesis and the predictors are investigated, with the most significant models cross‐validated, and the skill of their hindcasts evaluated. The September lead SOI‐only Poisson regression model provided skillful predictions of the temporal variability of tropical cyclogenesis across the entire region, with a root‐mean‐square error 22% better than climatology. The combination SOI and EPT model adds spatial skill and further improves temporal skill. Temporal skill is best in the Eastern subregion (western tropical Pacific) (significant correlations with observations at ∼99% level), while spatial skill is best elsewhere.