Projected change in Frequency, Intensity and Duration of Atmopsheric Temperature Inversions for Southeast Australia.

Temperature inversions occur when temperature increases with altitude in the lower atmosphere. An inversion can lead to pollution events such as smog being trapped close to the ground, with possible adverse effects on health, and may result in violent thunderstorm and freezing rain in the cold season. The effect of temperature inversions means that any trends in their frequency, intensity (temperature gradient within the inversion layer), and duration under global warming have implications for sectors such as air pollution management or agriculture. In this study, we used outputs of 12 historical and future Regional Climate Model (RCM) simulations (each covering three time periods: 1990-2009, 2020-2039, and 2060-2079) from the NSW/ACT Regional Climate Modelling (NARCliM) project to investigate changes in low level temperature inversions. For each 10km by 10km grid cell within the NARCliM domain, temperature inversions were identified by checking the vertical temperature profile in 3-hourly data. Characteristics of the inversions such as height, temperature at the top and bottom of the inversion layer were recorded. Temperature inversions for the two future periods (2020- 2039 and 2060-2079) are compared to the historic period (1990-2009) to investigate the changes in frequency, intensity, and duration of inversions for each of the 12 simulations. The results show that there is a substantial increase in the frequency and duration of temperature inversions and a decrease in the intensity of the temperature inversion for most simulations for southeast Australia. The largest differences between simulations were associated with the driving GCMs, suggesting that the large scale circulation plays a dominant role in forming and sustaining low level temperature inversions.

Figure 1. Ensemble mean frequency of temperature inversion for 1990-2009, and changes in frequency for 2020-2039 and 2060-2079 relative to 1990-2009 (unit: %).