A Bayesian Forecast Model of Australian Region Tropical Cyclone Formation

A new and potentially skilful seasonal forecast model of tropical cyclone formation\ud(genesis, TCG) is developed for the Australian region. The model is based on Poisson\udregression using the Bayesian approach. Predictor combinations are chosen using a stepby-\udstep predictor selection. The three-predictor model based on derived indices of June-\udJuly-August average convective available potential energy, May-June-July average\udmeridional winds at 850 hPa (v850) and July-August-September geopotential height at 500\udhPa produces the smallest standard error (se = 0.36) and root-mean-squared error (RMSE\ud= 5.20) for the leave-one-out cross-validated TCG hindcasts over the 40-year record\udbetween 1968/89-2007/08. The corresponding correlation coefficient between observed\udannual TCG totals and cross-validated model hindcasts is r = 0.73. Using four-fold crossvalidation,\udmodel hindcast skill is robust with 85% of the observed seasonal TCG totals\udhindcast within the model standard deviations. Seasonal TCG totals during ENSO events\udare typically well captured with RMSE = 5.14 during El Niño and RMSE = 6.04 during\udLa Niña years. The model is shown to be valuable in hindcasting seasonal TCG totals in\udthe Eastern Australian subregion (r = 0.73) and also provides some skill for the Western\udAustralian region (r = 42), while it not useful for the Northern region. In summary, we\udfind that the three-predictor Bayesian model provides substantial improvement over\udexisting statistical TCG forecast models, with remarkably skilful hindcasts (forecasts) of\udAustralian region and subregional seasonal TCG totals provided one month ahead of the\udTC season