Southern Australia is in the mid-latitude belt that is projected to get drier due to enhanced greenhouse effect and a warming climate. Persistent shifts to the mean circulation are the primary driver of this projected drying, including a more positive southern annular mode, an enhanced subtropical ridge, a poleward shift in the mean storm track and changes to blocking in the Tasman Sea. Climate models have biases in the simulation of the current climate, and a range of projected change to these features, restricting the confidence in the magnitude of likely rainfall change.

Here we examine a few approaches to better constrain and define the projected change signal for southern Australian mean July rainfall based on circulation indices in the CMIP5 models. First we cover the four emergent constraints of model bias and projection, and what this means for rainfall projections, found previously by Grose et al. (2017). These constraints were for the strength of the subtropical jet, frequency of blocked days, the longitude of peak blocking and the latitude of the storm track on the polar front jet. We follow this up with work on using a statistical model of rainfall based on circulation indices not as a prediction model but as a method of evaluating and weighting CMIP5 models. We then also explore an analogue year approach to examining what a projected change in circulation means for rainfall change in southern Australia.