3.7.2 Australia
The Australian monsoon covers the northern third of continental Australia and surrounding seas and, considering its closely coincident location and annual evolution, is often studied in conjunction with the monsoon over the islands of Indonesia and Papua New Guinea. The Australian monsoon exhibits large interannual and intra-seasonal variability, largely associated with the effects of ENSO, the Madden-Julian Oscillation (MJO) and tropical cyclone activity (McBride, 1998; Webster et al., 1998; Wheeler and McBride, 2005). Using rain-gauge data, Hennessy et al. (1999) found an increase in calendar-year total rainfall in the Northern Territory of 18% from 1910 to 1995, attributed mostly to enhanced monsoon rainfall in the 1970s and coincident with an almost 20% increase in the number of rain days. With data updated to 2002, Smith (2004) demonstrated that increased monsoon rainfall has become statistically significant over northern, western and central Australia. Northern Australian wet season rainfall (Jones et al., 2004), updated through 2004–2005 (Figure 3.36), shows the positive trend and the contribution of the anomalously wet period of the mid-1970s as well as the more recent anomalously wet period around 2000 (see also Smith, 2004). Wardle and Smith (2004) argued that the upward rainfall trend is consistent with the upward trend in land surface temperatures that has been observed in the south of the continent, independent of changes over the oceans. Strong decadal variations in Australian precipitation have also been noted (Figure 3.36). Using northeastern Australian rainfall, Latif et al. (1997) showed that rainfall was much increased during decades when the tropical Pacific was anomalously cold (the 1950s and 1970s). This strong relationship does not extend to the Australian monsoon as a whole, however, as the rainfall time series (Figure 3.36) has only a weak negative correlation (approximately –0.2) with the IPO. The fact that the long-term trends in rainfall and Pacific SSTs are both positive, the opposite of their interannual relationship (Power et al., 1998), explains only a portion of why the correlation is reduced at decadal time scales.