This article presents verification results for the current sea-surface temperature based seasonal forecasting system of the Australian Bureau of Meteorology’s National Climate Centre. Publicly issued probability forecasts are verified using a range of scoring techniques, including the Linear Error in Probability Space skill-score, correct forecast rates and reliability curves. Of the three seasonal forecast variables (maximum temperature, minimum temperature and rainfall) routinely forecast, the maximum temperature outlooks have been the most successful, while the rainfall forecasts have almost everywhere performed better than climatology. Minimum temperature forecasts have shown mixed success, largely as a result of a sequence of poorly verifying forecasts during 2001 and early 2002, although the most recent forecasts have performed considerably better.
Fawcett et al. (Tue,) studied this question.