Multidecadal Simulations of Australian Rainfall Variability: The Role of SSTs

Abstract

Australian rainfall variability and its relationship with the Southern Oscillation index (SOI) and global sea surface temperature (SST) variability is considered in both observational datasets and ensembles of multidecadal simulations using two different atmospheric general circulation models forced by observed SSTs and sea ice extent. Monthly and seasonal time series have been constructed to examine the observed and modeled relationships. The models show some success in the Australian region, largely reproducing the observed relationships between rainfall, the SOI, and global SSTs, albeit better in some seasons and geographical regions than others. A partition of the rainfall variance into components due to SST forcing and internal variability, suggests that both models have too much internal variability over the central eastern half of the continent, especially during austral winter and spring. Consequently, the strength of the SOI and SST relationships tend to be underestimated in this region. The largest impact of SST forcing is seen over the tropical and western parts of the continent. A principal component analysis reveals two dominant rotated modes of rainfall variability that are very similar in both the observed and modeled cases. One of these modes is significantly correlated with SST anomalies to the north-northwest of Australia (in the case of the models) and the SST gradient between the Indonesian archepelago and the central Indian Ocean (in the observed case). The other mode is significantly correlated with the typical SST anomaly pattern associated with the El Niño?Southern Oscillation. Correlative maps between the principal component time series and the modeled MSLP, 700-hPa, and 300-hPa geopotential heights are used to explore the underlying physical processes associated with these statistical relationships.