Abstract

During austral winter and spring, the El Niño–Southern Oscillation (ENSO) and the Indian Ocean dipole (IOD), individually or in combination, induce equivalent-barotropic Rossby wave trains, affecting midlatitude Australian rainfall. In autumn, ENSO is at its annual minimum, and the IOD has usually not developed. However, there is still a strong equivalent-barotropic Rossby wave train associated with tropical Indian Ocean sea surface temperature (SST) variability, with a pressure anomaly to the south of Australia. This wave train is similar in position, but opposite in sign, to the IOD-induced wave train in winter and spring and has little effect on Australian rainfall. This study shows that the SST in the southeastern tropical Indian Ocean (SETIO) displays a high variance during austral autumn, with a strong influence on southeast and eastern Australian rainfall. However, this influence is slightly weaker than that associated with SST to the north of Australia, which shares fluctuations with SST in the SETIO region. The SST north of Australia is coherent with a convective dipole in the tropical Pacific Ocean, which is the source of a wave train to the east of Australia influencing rainfall in eastern Australia. ENSO Modoki is a contributor to the convective dipole and as a result it exerts a weak influence on eastern Australian rainfall through the connecting north Australian SST relationship. Thus, SST to the north of Australia acts as the main agent for delivering the impact of tropical Indo-Pacific variability to eastern Australia.

1. Introduction

The influence of tropical Indo-Pacific variability has been known to affect the Australian climate in most seasons (McBride and Nicholls 1983; Ashok et al. 2003; Cai et al. 2011; Cai and van Rensch 2013). The main drivers of this interseasonal variability are the El Niño–Southern Oscillation (ENSO) and the Indian Ocean dipole (IOD). An ENSO event typically initiates during austral winter, peaking around December and decaying during the late summer to early autumn (Fig. 1a). As a result of this, ENSO has the strongest influence on Australian climate during the austral spring season (McBride and Nicholls 1983). During this season, ENSO affects Australian climate in two ways: First, through the localized convection associated with tropically trapped and baroclinic circulation features, which are in response to anomalous tropical sea surface heating (Matsuno 1966; Gill 1980). The effects on Australian rainfall also depend on the phase of the ENSO event. During La Niña the convection anomalies are more westward and in close proximity to northeastern Australia; however, during El Niño or neutral phases the convection anomalies are farther east and have little effect on Australia (Cai et al. 2010). Another type of ENSO, referred to as El Niño Modoki, or date line ENSO, has a sea surface temperature (SST) anomaly located near the central Pacific, near the date line. ENSO Modoki has been shown to have a greater rainfall response in Australia due to its closer proximity (Ashok et al. 2007b; Taschetto and England 2009; Ashok et al. 2009), particularly in autumn (Cai and Cowan 2009).

Fig. 1.

(a) Monthly standard deviations of DMI (blue), Niño-3.4 (red), and EMI (green) for the period 1979–2010. (b) Percentage of annual rainfall that falls in MAM for the period 1979–2010. Inconsistencies in central-west Australia are because of sparse data.

Fig. 1.

(a) Monthly standard deviations of DMI (blue), Niño-3.4 (red), and EMI (green) for the period 1979–2010. (b) Percentage of annual rainfall that falls in MAM for the period 1979–2010. Inconsistencies in central-west Australia are because of sparse data.

Second, ENSO affects southern Australian climate through Rossby wave trains. The ENSO signal in the tropical SST induces deep convection and divergence in the upper troposphere, inducing Rossby wave generation that propagates into higher latitudes (Hoskins and Karoly 1981). Over the North Pacific this is known as the Pacific–North American pattern and in the south it is known as the Pacific–South American (PSA) pattern. However, the PSA pattern is typically too far east to have an influence in the Australian region. Only through ENSO’s coherence with the IOD during austral spring can ENSO have an effect on southern Australian rainfall (Cai et al. 2011).

The IOD induces Rossby waves in a similar manner to ENSO, producing an equivalent-barotropic pressure center just south of Australia, which modifies the general circulation in that area (Saji and Yamagata 2003; Cai et al. 2011; Ashok et al. 2007a). The IOD typically develops during austral winter, peaking in spring, but it cannot compete with the commencement of the Australian monsoon in early December, resulting in its termination (Fig. 1a). During austral summer the Indian Ocean has a basinwide mode that is highly coherent with ENSO (Reason et al. 2000; Saji et al. 2006; Xie et al. 2009; Du et al. 2009; Taschetto et al. 2011). This Indian Ocean basinwide mode (IOBM) induces convection anomalies over northern Australia, which in turn induce Rossby wave trains with a center over the Tasman Sea, extending the PSA pattern farther west (Cai and van Rensch 2013).

During the austral autumn from March to May (MAM), Australia receives its second highest seasonal average rainfall, with a large portion occurring in Western Australia and along the eastern coast (Fig. 1b). Yet, both the IOD and conventional ENSO are experiencing or transitioning to their lowest state of variability (Fig. 1a). This decrease in variance would likely have a diminishing effect on their teleconnections, with a reduced influence on Australian rainfall observed during this season (Fig. 2). Numerous studies have examined southern Australian autumn rainfall, mainly in the context of the observed rainfall decline. A drought in southern Australia from 1997 to 2009 was characterized by a strong rainfall decline during the cool season, particularly in May (Purich et al. 2013; Cai et al. 2014). In an attempt to explain the late autumn southeast Australian rainfall decline, Cai and Cowan (2008) show that wave trains propagating from the west of Australia can affect southeast Australian rainfall. They also found that SST in the Indonesian Throughflow region can affect rainfall in southeast Australia though the northwest cloud band phenomenon. Another study, focusing on mid-to-late autumn, described a link between blocking in the Australian region and the IOD (Cowan et al. 2013), even though it is not typically known to be active during that time of year. These studies were restricted to a portion of austral autumn; however, Nicholls (2010) used an extended season of March–August to examine the rainfall decline in the whole of southern Australia. The study largely attributed the 1958–2007 decline in rainfall to trends in the southern annular mode (SAM), but also reinforced a link between southern Australian rainfall and SSTs north of Australia that was established in an earlier study (Nicholls 1984). Other studies confined to the austral autumn months also suggest tropical links influencing Australian rainfall. Focusing on an alternative form of ENSO, studies have found a link between ENSO Modoki and Australian rainfall during this season (Taschetto and England 2009; Cai and Cowan 2009).

Fig. 2.

Regressions of (left) 200-hPa geopotential height (m), (center) SST (°C), and (right) rainfall (mm) onto (a)–(c) DMI, (d)–(f) IOBM, (g)–(i) Niño-3.4, and (j)–(l) EMI during MAM for the period 1979–2010. The analysis was performed using detrended data, and the regressions have been normalized by the standard deviations of their respective indices. The green (purple) contours indicate correlations significant at a p value of 0.05 (0.01).

Fig. 2.

Regressions of (left) 200-hPa geopotential height (m), (center) SST (°C), and (right) rainfall (mm) onto (a)–(c) DMI, (d)–(f) IOBM, (g)–(i) Niño-3.4, and (j)–(l) EMI during MAM for the period 1979–2010. The analysis was performed using detrended data, and the regressions have been normalized by the standard deviations of their respective indices. The green (purple) contours indicate correlations significant at a p value of 0.05 (0.01).

There has been considerable research undertaken in determining the influence of tropical climate drivers on Australian rainfall during austral autumn. However, the exact influence of the IOD is still unclear, as is the importance of the IOBM during this season. This has repercussions on the development of wave trains and their influence on Australian climate, with further implications on the ENSO Modoki–southern Australian rainfall relationship, if a mechanism similar to the springtime ENSO–IOD–rainfall pathway is in operation. After describing the data and methods used, this study examines the effects of the known sources of tropical variability during MAM, for example, the IOD and ENSO. Following this, we search for other more robust sources of wave trains influencing Australian climate, in an attempt to fully explain the tropical to midlatitude interactions during the autumn season. Finally, an analysis of their implications on ENSO Modoki will be explored.

2. Data and methods

This study focuses on the 1979–2010 period, ensuring enhanced coverage of datasets through the use of satellites. We use SST data from the Hadley Centre global sea ice and SST analyses (Rayner et al. 2003), from which all our SST-based time series are derived. Mean sea level pressure, geopotential height, divergence, and vertically integrated water vapor flux are obtained from the European Centre for Medium-Range Weather Forecasts Interim Re-Analysis (ERA-Interim) (Dee et al. 2011). All pressure and geopotential height results in this study match those calculated using the National Centers for Environmental Prediction–National Center for Atmospheric Research reanalysis (Kalnay et al. 1996). Outgoing longwave radiation (OLR) is used as a measure of convection and provided by the National Oceanic and Atmospheric Administration (Liebmann and Smith 1996). Finally, a gridded rainfall dataset over the Australian region is provided by the Australian Bureau of Meteorology (Jones et al. 2009).

Throughout this paper new time series will be introduced, calculated as a spatial average of a gridded dataset over a region of interest; these indices will be described in appropriate sections. We also make use of several known climate indices based on SST, such as the Niño-4 (5°S–5°N, 160°E–150°W), Niño-3.4 (5°S–5°N, 170°–120°W), and Niño-3 (5°S–5°N, 150°–90°W) indices. The El Niño Modoki index (EMI) is defined using the method described in Ashok et al. (2007b): EMI = BoxA − 0.5(BoxB + BoxC), where BoxA (10°S–10°N, 165°E–140°W), BoxB (15°S–5°N, 110°–70°W), and BoxC (10°S–20°N, 125°–145°E) are spatially averaged boxes. An alternative version of the EMI was also examined where EMIalt = 3.0BoxA − 2.0BoxB − BoxC (Li et al. 2010); however, with the correlation between EMIalt and EMI at 0.97 during this season, our results were essentially the same. To depict the IOD, the dipole mode index (DMI; Saji et al. 1999) is used: DMI = BoxY − BoxZ, where BoxY is the area average of SST in the region 10°S–10°N, 50°–70°E and BoxZ is the region 10°S–0°, 90°–110°E. Finally the IOBM is taken from the first principal component of detrended MAM SST in the region 40°S–30°N, 30°–140°E (Cai and van Rensch 2013).

Regression and correlation analysis are performed on the datasets in order to discern regions of high coherence. The importance of the regressions is aided with the outline of significant correlation values, where significant is defined as the regions where the correlations have a p value ≤0.05 as determined by a two-sided Student’s t test (Cohen and Cohen 1983). A more statistically significant p value of 0.01 is also indicated. In instances where regression and correlation are performed, the datasets have been detrended to minimize any incidental coherence resulting from underlying trends. To isolate one mode of variability from another in the circulation field, various methods could be used. One could examine years in which the modes occur independently (Meyers et al. 2007) or use statistical methods, such as removing the lagged regression (Werner et al. 2012), or a partial correlation technique. Here, we use partial regression/correlation techniques as this method has previously been successfully used in similar studies (Ashok et al. 2007a; Cai et al. 2011), and it provides us with greater data points than selecting years of independent events. Partial regression/correlation determines the influence of one mode of variability A on the general circulation B if another mode X was no longer acting on both the original mode and the general circulation. This is performed by determining the linear relationship between A and X and subtracting the resulting linear equation with respect to A from A, producing A|X. The same is done for B, producing B|X. Then we perform a regression/correlation on A|X and B|X, producing (B versus A)|X. Because of the various types of indices used and to assist with any comparisons, the magnitudes of each regression is adjusted by multiplying by the index standard deviation.

3. Known tropical variability during austral autumn

During MAM tropical influences on Australian rainfall are known to be weak (McBride and Nicholls 1983). ENSO is typically moving to its decayed phase or about to develop, and the IOD does not show its characteristic structure (Fig. 1a). The IOBM is still quite active and represents 37.4% of the total variance observed in the Indian Ocean. The known modes of tropical climate variability have already been shown to have a strong wave train response in other seasons (Cai et al. 2011; Cai and van Rensch 2013), and we will show that there is utility in examining these modes in MAM when investigating any wave train responses to tropical SST.

A typical IOD event matures in austral spring; however, there is evidence that the IOD can develop in April and May and peak during austral winter (Du et al. 2013). The mid-to-late autumn IOD has previously been shown to have an effect on blocking in the Australian region (Cowan et al. 2013). Regressing gridpoint 200-hPa geopotential height anomalies onto the DMI shows an anomaly center just south of Australia as part of a wave train (Fig. 2a), as shown in Cowan et al. (2013). This wave train is opposite in sign and roughly 5°–10° south than that in winter and spring (cf. Fig. 2a and Figs. 3k and 5k of Cai et al. 2011; Ashok et al. 2007a), resulting in the autumn DMI showing little influence on southern Australia rainfall, as opposed to the following two seasons (cf. Fig. 2c and Figs. 3l and 5l of Cai et al. 2011). The significant difference seen during autumn is likely a result of the relatively weak IOD amplitude during this season (Fig. 1a).

During austral autumn the IOBM also has the opposite sign of the wave train response when compared to the sign in summer, with the pressure center shifted westward, south of Western Australia (cf. Figs. 2d and 3k of Cai and van Rensch 2013). The IOBM–rainfall response shows a significant difference when comparing autumn to summer (cf. Figs. 2f and 3i of Cai and van Rensch 2013), with little rainfall response in the northern, western, and central Australia during the autumn season. There is, however, a significant rainfall response in the southeast Queensland region associated with the IOBM, with a slight wave train signal to the east of Australia over northern New Zealand. Overall, considering the different response in rainfall and wave train sign, there are different teleconnective dynamics originating from the Indian Ocean during autumn when compared to other seasons.

The Pacific Ocean shows the largest tropical variability during autumn, even though most ENSO indices are moving into their weakest phase (Fig. 1a). Focusing on an index of ENSO (Niño-3.4), there is a strong wave train response; this is the PSA pattern [Fig. 2; the Niño-4 index also produces similar wave train results, yet with a stronger rainfall response in the northwest of Australia (not shown)]. The rainfall response over Australia appears to be attributable to tropical effects rather than interactions with the PSA. This is clear when observing a range of indices, such as the Niño-3.4, Niño-4 (not shown), and the EMI, where an index with a closer proximity to Australia shows a stronger eastern Australia rainfall response. Also, the rainfall response on Australia is quite limited to the northern and northwestern regions, with only the EMI showing an additional rainfall response over New South Wales (Figs. 2i,l). The PSA pattern associated with the EMI possesses a more zonal structure and is situated farther to the north when compared to other ENSO indices (Figs. 2g,j), with the pressure anomaly pattern sitting closer to Australia inducing a slight rainfall response in the east of Australia (Figs. 2j,l). The difference in eastern Australian rainfall response to the EMI compared to other ENSO indices, such as the Niño-3.4 index, could be a result of the use of gradients in the index definition. The EMI is a measure of the zonal gradients between three poles, with one pole located just north of Australia extending into the Northern Hemisphere, which could be a source of greater coherence with Australian climate.

4. The influence of southeastern tropical Indian Ocean on Australian rainfall

In the previous section we examined the wave train and rainfall responses associated with well-known climate driver indices. In many respects these did not prove to have a significant contribution to Australian rainfall, particularly when comparing their relative effects to other seasons. Is there another mechanism that can provide a more significant response to Australian rainfall during MAM? A map of SST standard deviations for MAM shows a strong variance in the tropical Pacific Ocean extending from the east to well over 170°E (Fig. 3a). This is a typical ENSO pattern and highlights the significance of the spatial extent of ENSO corresponding to the temporal variability, shown in Fig. 1a. The Indian Ocean lacks any large variance comparable to the Pacific, yet there are still pockets of higher variance that may be a source of variability. There is a region of high variance south of Java and to the west of northern Australia, which extends south along the Western Australian coast then out toward the southwest. This matches quite well with the second principal component of detrended Indian Ocean SSTs, where the first principal component is defined as the IOBM (not shown).

Fig. 3.

(a) Standard deviation of SST during MAM for the period 1979–2010. (b) Time series of the average SST during MAM for the box shown in (a).

Fig. 3.

(a) Standard deviation of SST during MAM for the period 1979–2010. (b) Time series of the average SST during MAM for the box shown in (a).

When choosing a suitable time series, we note that the atmosphere interacts directly with total SSTs, so we use an average of SST in the region of highest variance. This region in the southeastern tropical Indian Ocean (SETIO) is defined as 20°–10°S, 100°–110°E and highlighted by the box in Fig. 3a. The time series, shown in Fig. 3b, is characterized by two large warming events in 1983 and 1998, corresponding to the extended warming of the two largest El Niño events during our chosen period of 1979–2010. There also appears to be a warming trend experienced in this region; however, this trend is not statistically significant (below the 0.01 p value) unless the two large El Niño years are removed.

Regressing various gridpoint circulation fields onto this new SETIO index we see that this Indian Ocean variability has a significant effect on Australian climate. There is a large rainfall response to this index along the east of Australia and along the south-central and southeastern coast (Fig. 4a). Convection anomalies are observed over the east coast of Australia, with a suggestion that the anomalies could link up to convection over the SETIO region and continue west into the Indian Ocean (Fig. 4b). The northwest cloud band phenomenon could be the main driver of this relationship (Tapp and Barrell 1984); however, over central Australia there is a region of low coherence that is inconsistent with this theory. Similar to the IOBM case, the SST in the SETIO region is associated with a pressure center to the south of Australia (Figs. 4c,d). However, this pressure center is considerably weaker than those associated with both the IOBM and the IOD, although it is still statistically significant. A dramatic difference is seen in the sea level pressure plot, which shows the center to be almost nonexistent (Fig. 4c). With the signal mainly confined to the upper levels, this suggests that this Rossby wave train is not quite equivalent barotropic, as described by Hoskins and Karoly (1981). This result is unexpected, as this shows that with a reduced wave train signal in the south of Australia, an increase in southern Australian rainfall is observed. This suggests that other rainfall mechanisms or wave trains may be responsible for this rainfall–SETIO SST coherence, a point further explored in section 5.

Fig. 4.

(a) Rainfall (mm), (b) OLR (W m−2), (c) mean sea level pressure (hPa), (d) 200-hPa geopotential height (m), and (e) SST (°C) regressed onto SETIO SST time series during MAM. The analysis was performed using detrended data, and the regressions have been normalized by the standard deviation of SETIO SST. The green (purple) contours indicate correlations significant at a p value of 0.05 (0.01).

Fig. 4.

(a) Rainfall (mm), (b) OLR (W m−2), (c) mean sea level pressure (hPa), (d) 200-hPa geopotential height (m), and (e) SST (°C) regressed onto SETIO SST time series during MAM. The analysis was performed using detrended data, and the regressions have been normalized by the standard deviation of SETIO SST. The green (purple) contours indicate correlations significant at a p value of 0.05 (0.01).

The upper-level pressure field exhibits another wave train pattern to the east of Australia, with the anomaly stretching zonally east over northern New Zealand and into the central Pacific (Fig. 4d). This may be responsible for the eastern Australia rainfall coherence. The anomaly appears to be tilted in the vertical axis, as the sea level pressure anomaly is located northeast of New Zealand (Fig. 4c). The wave train is extended by a stronger companion anomaly to its south, which is barotropic, highlighting that this anomaly is possibly reinforced by other means compared to the northern anomaly. Overall the appearance of the wave train to the east of Australia is reminiscent to that of the PSA pattern, which is associated with ENSO. The SETIO SST shows coherence with SST in the east Pacific (Fig. 4e), with a correlation of 0.42 to the Niño-3 index, which is statistically significant above the 98% confidence interval. However, the wave train associated with SETIO SST is situated in the western Pacific. If correlating with an ENSO index from this region, such as the Niño-4 index, the SETIO index is unrelated (Table 1), yet the EMI shows a significant negative correlation. The EMI better represents an ENSO Modoki event rather than the Niño-4 index because the EMI incorporates the observed zonal gradient, which involves SST over the north of Australia. Therefore, the wave train pattern is more likely a result of ENSO Modoki–type events rather than canonical ENSO events. A large region of SST exhibits coherence with the SETIO region, which includes much of eastern Indian Ocean and extends along the north of Australia into the Coral Sea (Fig. 4e). This study cannot determine any causal links between these regions; however, the SST to the north and northeast of Australia could provide a more obvious pathway for SSTs over the SETIO region to exert an impact on eastern Australian rainfall, which is examined in the following section.

Table 1.

Correlations between all detrended indices during MAM. Correlations with a p value ≤0.05 are marked in boldface. Values in parentheses indicate p values as determined by a two-sided t test.

Correlations between all detrended indices during MAM. Correlations with a p value ≤0.05 are marked in boldface. Values in parentheses indicate p values as determined by a two-sided t test.
Correlations between all detrended indices during MAM. Correlations with a p value ≤0.05 are marked in boldface. Values in parentheses indicate p values as determined by a two-sided t test.

5. The significance of north Australian SST

The SST to the north and northeast of Australia show a strong coherence to Australian rainfall (Figs. 5a,b). However, this small difference in proximity between these two regions is capable of having a significantly different rainfall relationship. The SST to the northeast of Australia in the northwest Coral Sea (the region within 20°–6°S, 142°–160°E, indicated by the gray box in Fig. 5d) shows a weaker east Australian rainfall response when compared to the SETIO region (cf. Figs. 4a and 5a). Variations of SST over the northwest Coral Sea region does, however, induce a strong rainfall response in the north of Australia (Fig. 5a), which is reminiscent of the Niño-3.4 rainfall response (Fig. 2i), yet opposite in sign. The SST region to the north of Australia (15°S–0°, 110°–150°E, indicated by the black box in Fig. 5d), as used in Nicholls (2010), shows a rainfall response slightly stronger than that associated with the SETIO SST. Thus, it appears that the more dominant driver of eastern Australian rainfall during MAM is through SST to the north of Australia rather than the SETIO region. Nicholls (1984) also examined the relationship between Australian rainfall and SST to the north of Australia during MAM. Using the period 1964–82, the study shows some marked differences from our results, with only a significant relationship in the southeast and central west coast of Australia. Reproducing our analysis for the period 1964–82 produced similar results to Nicholls (1984) (not shown), suggesting a distinct multidecadal variability in this rainfall relationship.

Fig. 5.

Detrended rainfall (mm) regressed onto detrended (a) Coral Sea and (b) SST north of Australia. (c) SST time series north of Australia. (d) Detrended SST regressed onto detrended SST north of Australia (°C). All results are for the MAM season. The regressions have been normalized by the standard deviations of their respective indices. The green (purple) contours indicate correlations significant at a p value of 0.05 (0.01). The black box in (d) indicates the region north of Australia, and the gray box indicates the northwest Coral Sea region.

Fig. 5.

Detrended rainfall (mm) regressed onto detrended (a) Coral Sea and (b) SST north of Australia. (c) SST time series north of Australia. (d) Detrended SST regressed onto detrended SST north of Australia (°C). All results are for the MAM season. The regressions have been normalized by the standard deviations of their respective indices. The green (purple) contours indicate correlations significant at a p value of 0.05 (0.01). The black box in (d) indicates the region north of Australia, and the gray box indicates the northwest Coral Sea region.

The time series of the north Australian SST shows a similar signal as the SETIO, but with less coherence from about 2000 onward (cf. Figs. 3b and 5c). The effect of the prolonged warming from the two strong El Niño years, 1982/83 and 1997/98, is not as prominent when compared to the SETIO time series, and this results in a warming trend in Fig. 5c that is statistically significant at the 0.01 p value even with those years included. The SST pattern associated with the north Australian SST shows a very broad area of coherence that extends south along the east and west Australian coasts (Fig. 5d). The SST pattern off the west coast of Australia shows a very similar structure to that associated with the SETIO SST. Similarly, the SST in the Coral Sea shows a strong statistically significant coherence with north Australian SST (Table 1). The cause of this widespread region of coherence is not fully understood and is outside the scope of this study. There are no statistically significant SST anomalies in the eastern Pacific coherent with north Australian SSTs, which is likely a result of the diminished impact of the strong El Niños in this region.

The strong coherence with SST along the east coast of Australia suggests that the rainfall response associated with north Australian SST could be attributable to a localized response to SST. However, this is not the case for the northwest Coral Sea, as there is not a widespread rainfall response, as seen in Fig. 5b, in close proximity to this region (Fig. 5a). Likewise, if the north Australian SST induces east Australian rainfall solely through a trapped, baroclinic tropical response, it would not extend as far south as that observed (Fig. 5b). Thus, we can assume that SST to the north of Australia induces eastern Australian rainfall (at the very least in the southern Australia region) through remote effects.

As the north Australian SST is strongly related to the SETIO SST, many of their teleconnective patterns are strikingly similar. The lower- and upper-layer geopotential height anomalies associated with the north Australian SST shows a wave train to the south of Australia in the same location as that of the SETIO case (Figs. 6a,b). However, in the north Australian SST case the strength is noticeably stronger and the shape is better defined. This is likely a result of the SST to the north of Australia having a greater coherence with the SST in the Indian Ocean region, from which this pressure center originates (Table 1). Associated with the stronger wave train to the south of Australia, there is less of an impact to rainfall in the south, indicating that the SETIO region has a greater influence in this region. At these pressure levels the wave train to the east of Australia that is associated with north Australian SST is similar in location and strength to the equivalent SETIO SST wave train. The SETIO wave train shows a greater extension toward the west, yet does not produce a greater extension of rainfall response than that of the north Australian SSTs. This inconsistency could be attributable to the better positioning of the associated winds for the north Australian SST case, whereas with the SETIO SST–induced wave train the winds could be weaker. It appears the SST to the north of Australia is better equipped to influence east Australian rainfall through the wave train to the east of Australia. Comparing Figs. 6a and 6b, a distinct eastward tilt is observed as one moves toward the surface. Figure 6c shows the profile of this tilt at the 30°S latitude. Focusing on the strongest anomaly over 165°W, the barotropic structure associated with Rossby wave trains upholds. Near the surface the anomaly is shifted eastward, which could be a result of interactions with the Australian continent.

Fig. 6.

(a) Mean sea level pressure (hPa), (b) 200-hPa geopotential height (m), and (c) geopotential height at 30°S regressed onto SST north of Australia during MAM. The analysis was performed using detrended data, and the regressions have been normalized by the standard deviation of the SST time series north of Australia. The green (purple) contours indicate correlations significant at a p value of 0.05 (0.01).

Fig. 6.

(a) Mean sea level pressure (hPa), (b) 200-hPa geopotential height (m), and (c) geopotential height at 30°S regressed onto SST north of Australia during MAM. The analysis was performed using detrended data, and the regressions have been normalized by the standard deviation of the SST time series north of Australia. The green (purple) contours indicate correlations significant at a p value of 0.05 (0.01).

As the wave train to the east of Australia is relatively remote to the continent, particularly at the lower levels, would this wave train still have a noticeable effect on east Australian rainfall? To examine this, we first examine the vertically integrated water vapor flux associated with the SST to the north of Australia. The orange and red vectors in Fig. 7a show statistically significant transportation of water vapor south and southeast from the northwest of Australia. The wave train to the east shows a significant anticyclonic flow associated with its pressure center. The two flows from the north of Australia and the anticyclonic flow converge in the Tasman Sea. Using an index that describes the east of Australia wave train (defined as the spatial average of the 200-hPa geopotential height in the region 40°–30°S, 160°E–150°W minus 65°–55°S, 170°E–140°W), the effect of the wave train can be removed from the SST north of Australia. Upon removal, the convergence in the Tasman Sea is no longer observed, nor is the water vapor transport toward southeastern Australia (Fig. 7b). This highlights the importance of the wave train in maintaining the relationship between SST north of Australia and east Australian rainfall.

Fig. 7.

Regression of vertically integrated water vapor flux onto (a) SST north of Australia (kg m−1 s−1), and (b) SST north of Australia with the effect of the east Australian wave train removed (kg m−1 s−1) during MAM. The analysis was performed using detrended data, and the regressions have been normalized by the standard deviations of their respective indices. The orange arrows indicate correlations significant between p values of 0.05 and 0.01, with red arrows indicating significance below 0.01.

Fig. 7.

Regression of vertically integrated water vapor flux onto (a) SST north of Australia (kg m−1 s−1), and (b) SST north of Australia with the effect of the east Australian wave train removed (kg m−1 s−1) during MAM. The analysis was performed using detrended data, and the regressions have been normalized by the standard deviations of their respective indices. The orange arrows indicate correlations significant between p values of 0.05 and 0.01, with red arrows indicating significance below 0.01.

6. Source of the wave train to the east of Australia

The wave train to the east of Australia is, at least in part, a result of influences in the tropics, yet the exact cause is not clear. The SST to the north of Australia shows a high coherence with this wave train, yet this SST does not seem to influence a significant convective response in the same region conducive to Rossby wave development (Fig. 8a). Consistent with Fig. 5b, there is a significant and widespread convection response in the east of Australia that extends eastward into the Tasman Sea. This matches quite well with the convergence of water vapor in Fig. 7a. The SST north of Australia is also coherent with two large convective responses in the tropical western and central Pacific. Associated with these convection anomalies are upper-level divergence anomalies (black and blue contours in Fig. 8b), which are sources of Rossby waves. The shading in Fig. 8b shows the correlation between north Australian SST and 200-hPa geopotential height. Tracing south-southwest from the blue divergence anomaly in the central Pacific, a weak wave train can be seen tracing south, increasing in strength and area. The increasing strength as the wave train extends poleward could be an artifact of the meridional distribution of pressure variance, where low variance is experienced in the tropics compared to higher in the midlatitudes. With the wave train originating from the central Pacific Ocean, the relevance of north Australian SSTs comes into question. The large area of convection associated with north Australian SSTs appears to be part of a basinwide structure. It would be intuitive to assume that other Pacific Ocean basinwide processes, like ENSO, would likely have an influence on this wave train and also affect rainfall. However, we know that the north Australian SST is not related to typical ENSO indices (Table 1), and ENSO does not have a relationship with east Australian rainfall during this season (Fig. 2). There is, however, a statistically significant relationship between the EMI and north Australian SST (Table 1), with the EMI producing a rainfall response in the east of Australia, although they are weak and not widely distributed (Fig. 2).

Fig. 8.

(a) Detrended OLR regressed onto detrended SST north of Australia (W m−2) during MAM. The regression has been normalized by the standard deviation of the time series north of Australia, with the green (purple) contours indicating correlations significant at a p value of 0.05 (0.01). (b) Shaded (line) contours show detrended 200-hPa geopotential (divergence) correlated with the detrended time series north of Australia during MAM. Positive (negative) divergence correlations are indicated by the blue (black) color. The thickened contour lines show the correlations significant at a p value of 0.05 (0.349), with the regular contours showing a p value of 0.01 (0.449).

Fig. 8.

(a) Detrended OLR regressed onto detrended SST north of Australia (W m−2) during MAM. The regression has been normalized by the standard deviation of the time series north of Australia, with the green (purple) contours indicating correlations significant at a p value of 0.05 (0.01). (b) Shaded (line) contours show detrended 200-hPa geopotential (divergence) correlated with the detrended time series north of Australia during MAM. Positive (negative) divergence correlations are indicated by the blue (black) color. The thickened contour lines show the correlations significant at a p value of 0.05 (0.349), with the regular contours showing a p value of 0.01 (0.449).

The convection response to the EMI shows a similar structure to that of the north Australian SST OLR response, although opposite in sign and increased in strength (cf. Figs. 8a and 9a). There are some slight differences, for example, the central Pacific OLR anomaly is located slightly farther east; the thin OLR anomaly band, extending from the north of Australia to the southeast into the Pacific, is not evident in the EMI case; and also, there is a greater OLR anomaly over the north and northwest of Australia, extending into the Indian Ocean. Removing the influence of the north Australian SST shows little difference in the OLR anomaly, with the convective dipole remaining intact (Fig. 9b). This shows that the ENSO Modoki is a major contributor to the convective dipole, as opposed to the north Australian SST. Figure 2j shows the wave train response associated with the EMI. There is a clear wave train to the east of Australia, although the pressure anomaly with the closest proximity to eastern Australia is roughly 5° south and zonally narrower compared to that associated with the north Australian SST (Fig. 5b), which is likely attributable to slight differences in OLR between the two cases. The effect of these slight differences is seen in the eastern Australian rainfall response, with patchy anomalies localized to New South Wales and along the northeastern coast (Fig. 2l). Removing the effect of the north Australian SST shows that a large portion of the rainfall response to EMI in the east of Australia is related to north Australian SSTs, whereas rainfall in the northwest is relatively unchanged (Fig. 9c). This highlights the importance of the north Australian SST in teleconnecting some influence of the Pacific Ocean to eastern Australian rainfall. Even with the convective dipole largely driven by ENSO Modoki, the north Australian SST is still required for there to be a rainfall response.

Fig. 9.

(a) OLR regressed onto the EMI (W m−2) during MAM. Partial regression of (b) OLR and (c) rainfall regressed onto the EMI with the effect of SST north of Australia removed (W m−2 and mm, respectively) during MAM. The analysis was performed using detrended data, and the regressions have been normalized by the standard deviations of their respective indices. The green (purple) contours indicate correlations significant at a p value of 0.05 (0.01).

Fig. 9.

(a) OLR regressed onto the EMI (W m−2) during MAM. Partial regression of (b) OLR and (c) rainfall regressed onto the EMI with the effect of SST north of Australia removed (W m−2 and mm, respectively) during MAM. The analysis was performed using detrended data, and the regressions have been normalized by the standard deviations of their respective indices. The green (purple) contours indicate correlations significant at a p value of 0.05 (0.01).

7. Conclusions

During austral autumn, many of the tropical SST climate drivers that typically affect Australian rainfall during other seasons are experiencing their weakest stage. The canonical ENSO is typically decaying during this season, and the IOD is slowly beginning its intensification. These two drivers are known to induce wave trains during their active seasons, and we have shown that they are still able to induce wave trains in the austral autumn season. However these wave trains do not seem to have an influence on Australian rainfall. During austral autumn, the IOD and IOBM produce strong wave trains that have a pressure center to the south of Australia. These pressure centers are similar in location to austral winter and spring; however, the anomalies have the opposite sign to those in winter and spring. During spring, the IOD acts as a pathway by which ENSO can influence southeast Australian rainfall, but with the IOD having virtually no effect on southeast Australian rainfall during austral autumn, ENSO also does not have an effect in this region. The only strong rainfall anomalies associated with ENSO are located to the north and northwest of Australia, which are likely attributable to the influence of tropically trapped dynamics.

This study shows that there is still a remote influence of tropical SSTs on Australian rainfall during the austral autumn season, as summarized in Fig. 10. There is a comparatively strong SST variance in the SETIO region that is capable of producing a significant rainfall response in the southern and eastern portions of Australia. The wave train response associated with this SST is in a similar position to the IOD pressure center to the south of Australia; however, it is noticeably weaker and does not influence east Australian rainfall. There is another pressure center, or wave train, associated with this SST that is located to the east of Australia. It is unlikely that this wave train is directly induced by the SST in the SETIO region because of their relative position and distance. However, the SETIO SST is also coherent with SST to the north of Australia.

Fig. 10.

Schematic illustrating the influences of the north Australian SST, SETIO, and ENSO Modoki affecting rainfall over the eastern Australian region during MAM. Shading indicates SST anomalies, clouds indicate convection anomalies, oval outlines indicate upper-level pressure anomalies, and stippling indicates rainfall anomalies. Red and blue colors distinguish signals of opposite anomalous response, with blue indicating cooler SSTs, stronger convection, weaker pressure, or increased rainfall. Eastern Australian rainfall is enhanced by wave trains originating from the central Pacific Ocean convection dipole, but only in the presence of moisture transport from the north (indicated by the dashed arrow) associated with SST in the north Australian region. Additionally, influence of SETIO to eastern Australia rainfall is exerted through the north Australian SSTs.

Fig. 10.

Schematic illustrating the influences of the north Australian SST, SETIO, and ENSO Modoki affecting rainfall over the eastern Australian region during MAM. Shading indicates SST anomalies, clouds indicate convection anomalies, oval outlines indicate upper-level pressure anomalies, and stippling indicates rainfall anomalies. Red and blue colors distinguish signals of opposite anomalous response, with blue indicating cooler SSTs, stronger convection, weaker pressure, or increased rainfall. Eastern Australian rainfall is enhanced by wave trains originating from the central Pacific Ocean convection dipole, but only in the presence of moisture transport from the north (indicated by the dashed arrow) associated with SST in the north Australian region. Additionally, influence of SETIO to eastern Australia rainfall is exerted through the north Australian SSTs.

North Australian SSTs show a stronger relationship to east Australian rainfall than the SST in the SETIO region, and is related to the wave train to the east of Australia. The source of the wave train is not located over the region north of Australia, as there is no convection or divergence response in this region. This wave train is likely a result of a strong convective response in the central tropical Pacific. The central Pacific convection is the eastern pole of a large tropical convection dipole coherent with north Australian SSTs. This dipole is reminiscent with that of the El Niño Modoki, and it is also shown that the influence of ENSO Modoki on the eastern Australian rainfall is enhanced by the SSTs in the north Australian region.

In conclusion, SST variability to the north of Australia has a predominant influence on eastern Australia autumn rainfall. It conveys not only the impact of its own variability, but also that of variability coherent with ENSO Modoki and with SST in the SETIO region.

Acknowledgments

This research is supported by the Australian Climate Change Science Program and the Goyder Institute. The authors thank Ariaan Purich and Evan Weller for their help in improving the manuscript before submission. Furthermore, the authors acknowledge the valuable comments provided by three anonymous reviewers, which greatly improved the study.

REFERENCES

REFERENCES
Ashok
,
K.
,
Z.
Guan
, and
T.
Yamagata
,
2003
:
Influence of the Indian Ocean dipole on the Australian winter rainfall
.
Geophys. Res. Lett.
,
30
, 1821,
doi:10.1029/2003GL017926
.
Ashok
,
K.
,
H.
Nakamura
, and
T.
Yamagata
,
2007a
:
Impacts of ENSO and Indian Ocean dipole events on the Southern Hemisphere storm-track activity during austral winter
.
J. Climate
,
20
,
3147
3163
,
doi:10.1175/JCLI4155.1
.
Ashok
,
K.
,
H.
Weng
,
S.
Behera
,
S. A.
Rao
, and
T.
Yamagata
,
2007b
:
El Niño Modoki and its teleconnection
.
J. Geophys. Res.
,
112
, C11007,
doi:10.1029/2006JC003798
.
Ashok
,
K.
,
C.-Y.
Tam
, and
W.-J.
Lee
,
2009
:
ENSO Modoki impact on the Southern Hemisphere storm track activity during extended austral winter
.
Geophys. Res. Lett.
,
36
, L12705,
doi:10.1029/2009GL038847
.
Cai
,
W.
, and
T.
Cowan
,
2008
:
Dynamics of late autumn rainfall reduction over southeastern Australia
.
Geophys. Res. Lett.
,
35
, L09708,
doi:10.1029/2008GL033390
.
Cai
,
W.
, and
T.
Cowan
,
2009
:
La Niña Modoki impacts Australia autumn rainfall variability
.
Geophys. Res. Lett.
,
36
, L12805,
doi:10.1029/2009GL037885
.
Cai
,
W.
, and
P.
van Rensch
,
2013
:
Austral summer teleconnections of Indo-Pacific variability: Their nonlinearity and impacts on Australian climate
.
J. Climate
,
26
,
2796
2810
,
doi:10.1175/JCLI-D-12-00458.1
.
Cai
,
W.
,
P.
van Rensch
,
T.
Cowan
, and
A.
Sullivan
,
2010
:
Asymmetry in ENSO teleconnection with regional rainfall, its multidecadal variability, and impact
.
J. Climate
,
23
,
4944
4955
,
doi:10.1175/2010JCLI3501.1
.
Cai
,
W.
,
P.
van Rensch
,
T.
Cowan
, and
H. H.
Hendon
,
2011
:
Teleconnection pathways of ENSO and the IOD and the mechanisms for impacts on Australian rainfall
.
J. Climate
,
24
,
3910
3923
,
doi:10.1175/2011JCLI4129.1
.
Cai
,
W.
,
A.
Purich
,
T.
Cowan
,
P.
van Rensch
, and
E.
Weller
,
2014
:
Did climate change induced rainfall trends contribute to the Australian Millennium Drought?
J. Climate
,
in press
.
Cohen
,
J.
, and
P.
Cohen
,
1983
: Applied Multiple Regression/Correlation Analysis for the Behavioral Sciences. Lawrence Erlbaum Associates, 545 pp.
Cowan
,
T.
,
P.
van Rensch
,
A.
Purich
, and
W.
Cai
,
2013
:
The association of tropical and extratropical climate modes to atmospheric blocking across southeastern Australia
.
J. Climate
,
26
,
7555
7569
,
doi:10.1175/JCLI-D-12-00781.1
.
Dee
,
D. P.
, and Coauthors
,
2011
:
The ERA-Interim reanalysis: Configuration and performance of the data assimilation system
.
Quart. J. Roy. Meteor. Soc.
,
137
,
553
597
,
doi:10.1002/qj.828
.
Du
,
Y.
,
S.-P.
Xie
,
G.
Huang
, and
K.
Hu
,
2009
:
Role of air–sea interaction in the long persistence of El Niño–induced north Indian Ocean warming
.
J. Climate
,
22
,
2023
2038
,
doi:10.1175/2008JCLI2590.1
.
Du
,
Y.
,
W.
Cai
, and
Y.
Wu
,
2013
:
A new type of the Indian Ocean dipole since the mid-1970s
.
J. Climate
,
26
,
959
972
,
doi:10.1175/JCLI-D-12-00047.1
.
Gill
,
A. E.
,
1980
:
Some simple solutions for heat-induced tropical circulation
.
Quart. J. Roy. Meteor. Soc.
,
106
,
447
462
,
doi:10.1002/qj.49710644905
.
Hoskins
,
B. J.
, and
D. J.
Karoly
,
1981
:
The steady linear response of a spherical atmosphere to thermal and orographic forcing
.
J. Atmos. Sci.
,
38
,
1179
1196
,
doi:10.1175/1520-0469(1981)038<1179:TSLROA>2.0.CO;2
.
Jones
,
D. A.
,
W.
Wang
, and
R.
Fawcett
,
2009
:
High-quality spatial climate data-sets for Australia
.
Aust. Meteor. Oceanogr. J.
,
58
,
233
248
.
Kalnay
,
E.
, and Coauthors
,
1996
:
The NCEP/NCAR 40-Year Reanalysis Project
.
Bull. Amer. Meteor. Soc.
,
77
,
437
471
,
doi:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2
.
Li
,
G.
,
B. H.
Ren
,
C. Y.
Yang
, and
J. Q.
Zheng
,
2010
:
Indices of El Niño and El Niño Modoki: An improved El Niño Modoki index
.
Adv. Atmos. Sci.
,
27
,
1210
1220
,
doi:10.1007/s00376-010-9173-5
.
Liebmann
,
B.
, and
C. A.
Smith
,
1996
:
Description of a complete (interpolated) outgoing longwave radiation dataset
.
Bull. Amer. Meteor. Soc.
,
77
,
1275
1277
.
Matsuno
,
T.
,
1966
:
Quasi-geostrophic motions in the equatorial area
.
J. Meteor. Soc. Japan
,
44
,
25
43
.
McBride
,
J. L.
, and
N.
Nicholls
,
1983
:
Seasonal relationships between Australian rainfall and the Southern Oscillation
.
Mon. Wea. Rev.
,
111
,
1998
2004
,
doi:10.1175/1520-0493(1983)111<1998:SRBARA>2.0.CO;2
.
Meyers
,
G.
,
P.
McIntosh
,
L.
Pigot
, and
M.
Pook
,
2007
:
The years of El Niño, La Niña, and interactions with the tropical Indian Ocean
.
J. Climate
,
20
,
2872
2880
,
doi:10.1175/JCLI4152.1
.
Nicholls
,
N.
,
1984
: Seasonal relationships between Australian rainfall and north Australian sea surface temperature. Extended Abstracts, Conf. on Australian Rainfall Variability, Arkaroola, South Australia, Australia, Australian Academy of Science and Bureau of Meteorology, 71–73.
Nicholls
,
N.
,
2010
:
Local and remote causes of the southern Australian autumn-winter rainfall decline, 1958–2007
.
Climate Dyn.
,
34
,
835
845
,
doi:10.1007/s00382-009-0527-6
.
Purich
,
A.
,
T.
Cowan
,
S.-K.
Min
, and
W.
Cai
,
2013
: Autumn precipitation trends over Southern Hemisphere midlatitudes as simulated by CMIP5 models. J. Climate,26, 8341–8356,
doi:10.1175/JCLI-D-13-00007.1
.
Rayner
,
N. A.
,
D. E.
Parker
,
E. B.
Horton
,
C. K.
Folland
,
L. V.
Alexander
,
D. P.
Rowell
,
E. C.
Kent
, and A. Kaplan,
2003
:
Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century
.
J. Geophys. Res.
,
108
, 4407,
doi:10.1029/2002JD002670
.
Reason
,
C. J. C.
,
R. J.
Allan
,
J. A.
Lindesay
, and
T. J.
Ansell
,
2000
:
ENSO and climatic signals across the Indian Ocean basin in the global context: Part I, interannual composite patterns
.
Int. J. Climatol.
,
20
,
1285
1327
,
doi:10.1002/1097-0088(200009)20:11<1285::AID-JOC536>3.0.CO;2-R
.
Saji
,
N. H.
, and
T.
Yamagata
,
2003
:
Possible impacts of Indian Ocean dipole mode events on global climate
.
Climate Res.
,
25
,
151
169
,
doi:10.3354/cr025151
.
Saji
,
N. H.
,
B. N.
Goswami
,
P. N.
Vinayachandran
, and
T.
Yamagata
,
1999
:
A dipole mode in the tropical Indian Ocean
.
Nature
,
401
,
360
363
.
Saji
,
N. H.
,
S.-P.
Xie
, and
T.
Yamagata
,
2006
:
Tropical Indian Ocean variability in the IPCC twentieth-century climate simulations
.
J. Climate
,
19
,
4397
4417
,
doi:10.1175/JCLI3847.1
.
Tapp
,
R. G.
, and
S. L.
Barrell
,
1984
:
The north-west Australian cloud band: Climatology, characteristics and factors associated with development
.
J. Climatol.
,
4
,
411
424
,
doi:10.1002/joc.3370040406
.
Taschetto
,
A. S.
, and
M. H.
England
,
2009
:
El Niño Modoki impacts on Australian rainfall
.
J. Climate
,
22
,
3167
3174
,
doi:10.1175/2008JCLI2589.1
.
Taschetto
,
A. S.
,
A.
Sen Gupta
,
H. H.
Hendon
,
C. C.
Ummenhofer
, and
M. H.
England
,
2011
:
The contribution of Indian Ocean sea surface temperature anomalies on Australian summer rainfall during El Niño events
.
J. Climate
,
24
,
3734
3747
,
doi:10.1175/2011JCLI3885.1
.
Werner
,
A.
,
A. M.
Maharaj
, and
N. J.
Holbrook
,
2012
:
A new method for extracting the ENSO-independent Indian Ocean dipole: Application to Australian region tropical cyclone counts
.
Climate Dyn.
,
38
,
2503
2511
,
doi:10.1007/s00382-011-1133-y
.
Xie
,
S.-P.
,
K.
Hu
,
J.
Hafner
,
H.
Tokinaga
,
Y.
Du
,
G.
Huang
, and
T.
Sampe
,
2009
:
Indian Ocean capacitor effect on Indo–western Pacific climate during the summer following El Niño
.
J. Climate
,
22
,
730
747
,
doi:10.1175/2008JCLI2544.1
.