Summary

Tropical forcing of Australian extreme low minimum temperatures in September 2019.

Scientific journal article

Researchers from the Bureau of Meteorology and the University of Melbourne recently investigated why the seasonal forecasting model, ACCESS-S1, failed to predict extreme cold in September 2019. Minimum temperatures were below average over much of south-eastern Australia and frosts led to severe crop damage. The common climate drivers were in phases that would be expected for this event. This includes El Niño conditions in the central Pacific and a near record strength positive Indian Ocean Dipole (IOD). Given historical impacts on Australian weather under these conditions, the low temperatures should have been predicable 1 to 2 months in advance.

The results suggest the model underestimates the impact of the positive IOD (and to a lesser extent El Niño) on low temperatures in south-eastern Australia. The IOD reflects sustained changes between sea surface temperatures of the tropical western and eastern Indian Ocean and has three phases: positive, neutral, and negative. A positive IOD is associated with cooler waters in the eastern portion of the Indian Ocean and weakened westerly, in some cases even a switch to easterly winds allowing cool deep ocean water to rise in the east. These conditions lead to reduced moisture in the northwest of Australia which changes the path of weather systems from Australia’s west and typically produces clear, dry conditions in southeast Australia. Reduced cloud cover at night results in increased outgoing longwave radiation and colder minimum temperatures.

The model’s weak connections between the IOD and these weather patterns were the main reason for its failure to predict the low minimum temperatures in September 2019. These weak internal relationships are likely the result of biases in average state of the tropical Indian Ocean in the model. This study helps identify ways to improve predictions of these types of events. Given these events generally coincide with the winter crop growing season the risk of significant impact on agriculture is greatly increased. Thus, for agriculture, the importance of accurate forecasts at this time of year make model improvements in this area a high priority.