Australia: The Land Where Time Began

A biography of the Australian continent 

Australian Summer Rainfall – Breakdown of Relationship with ENSO Resulting from Warming of Tropical Indian Ocean SST

There was an Interdecadal change around the mid-1980s during the period from 1960 to 2015 to the relationship between the Australian Summer Rainfall (ASR) and El Niño-Southern Oscillation (ENSO). The Australian Summer Rainfall was significantly correlated with the Sea Surface Temperature (SST) in the tropical central pacific (TCP), but after that time it was not. ASR had a close relationship with La Niña, though El Niño was always independent from ASR. This relationship was weakened, however, following the mid-1980s. The warming of the Indian Ocean SST might contribute to the weakening of the relationship between ASR and El Niño. For La Niña events prior to the mid-1980s, the negative SSTA over TCP and the southern tropical Indian Ocean induced a large-scale lower-level cyclonic anomaly over Australia, which led to precipitation over Australia that was almost uniformly positive. A significant relationship between ASR and La Niña was established in this summer. Contrary to this, after the mid-1980s, as a result of warming of the SST in the Indian Ocean, the equatorial eastern Indian Ocean and maritime continent presented positive SSTAs and increased levels of moisture, which favoured increased rainfall anomalies over the Equatorial Maritime Continent. A lower level cyclonic anomaly was induced to the west of Australia by this increased rainfall condensation heating. At the eastern flank of this cyclonic anomaly the northerly anomalies counteracted the southerly anomalies at the western flank of the cyclonic anomaly over eastern Australia that was induced by the negative TCP SSTA, which led to insignificant circulation  and rainfall anomalies over Australia. As such, being interfered with by the equatorial Maritime Continent the relationship between ASR and La Niña was weakened.

Relationships between tropical SSTs and global and/or regional precipitation are widely detected, as ENSO is the most important source of predictability for seasonal climate prediction (Ropelewski & Halpert, 1987, 1989; Mason & Goddard, 2001; Chang et al. 2000). The relationship between ENSO and regional rainfall and/or circulation is, however, not stationary (Parthasarathy et al., 1991; Gershunov & Barnett, 1998; Gao et al., 2006; Wang et al., 2008; Ding et al., 2010; Yoon & Yeh, 2010). Based on a 142 year historical record, 1856-1997, e.g., (Kumar et al., 1999) it is indicated that the inverse relationship between ENSO and the Indian summer monsoon, a weak monsoon that arises from warm ENSO events, breaks down after the mid-1980s. It was suggested (Kwon et al., 2005) that the most dominant rainfall mode is associated with ENSO over the period 1979-1993, though for the period 1994-2004 it is associated with the western North Pacific summer monsoon. The relationship between the winter rainfall in southern China and the western Pacific warm pool heat content also marks an incredible Interdecadal change in the early 1990s (Ren et al., 2017). It was suggested (Chen et al., 2014) that the relationship between the winter-spring precipitation in southern China and ENSO weakened over the period 1973-1974, and the SST anomalies in the south Indian Ocean contributed to the winter-spring precipitation variability in southern China independently. Also, it was suggested that the relationship between the summer rainfall over the contiguous US and the tropical-eastern Pacific SST underwent an Interdecadal weakening at about the early 1990s (Zhu & Li, 2016,2018).

Many studies have been carried out on the relationship between precipitation in Australia and ENSO. It has been suggested, for instance, that during La Niña events abnormally high levels of precipitation occur often, whereas there is a tendency for abnormally low precipitation to occur during El Niño episodes (McBride & Nicholls, 1983; Murphy & Ribbe, 2004). The seasonal forecast skill of Australian summer rainfall (ASR) turned out to be quite low (Hendon et al., 2012), despite the fact that the ENSO signal is strongest during the austral summer (Nicholls et al., 1982).

Zhu suggests there are 3 hypotheses that might be responsible for the low forecast skill of ASR:

1)    The first is related to the negative feedback of the local air-sea interaction during the monsoon season (Hendon et al., 2012). Prior to the onset of the Australian summer monsoon, positive feedback between surface winds, SST, and rainfall, that results in stronger and more persistent SSTAs to the north of Australia that complement the remote impact on the Australian rainfall from ENSO in the Pacific Ocean, could result from the presence of trade easterlies. After the onset of the monsoon, however, this local feedback cannot be maintained in the monsoonal westerly regime, which would result in SSTAs to the north of Australia that are weaker and are not persistent. The negative feedback of local air-sea interaction would, therefore, result in an unreliable ENSO-ASR relationship, and low forecasts of skill of ASR.

2)    The second hypothesis emphasises the symmetry (nonlinearity) of the relationship between ENSO and ASR (Wang & Hendon, 2007; King et al., 2013). It was indicated (Power et al., 2006) that the relationship between ENSO (as measured by Niño-4 or the Southern Oscillation Index, and all-Australian rainfall is asymmetric in observations and in simulations in a coupled general circulation model  during June-December:  a large La Niña SSTA is closely linked to a large rainfall response in Australia, i.e. usually Australia becomes much wetter, whereas the magnitude of an El Niño event is not a good precursor to the degree to which Australia will actually become dry. There is a tendency for Australia to be dry during El Niño events, though the extent of this dryness is not associated as tightly with the SSTA of El Niño. The La Niña rainfall relationship, as indicated (Cai et al., 2012), is statistically significant, as there is an increase in summer rainfall in southeast Queensland increases with the amplitude of La Niña; contrasting with this, the rainfall reductions that are induced by El Niño do not have a  statistically significant relationship with the amplitude of El Niño. The asymmetry of the ENSO-ASR relationship means that it is hard to use ENSO in seasonally predicting ASR.

3)    The third hypothesis argues that the Pacific decadal oscillation (PDO); (Arblaster et al., 2002; Power et al., 2006; Risbey et al., 2009; Cai & van Rensch, 2012; King et al., 2013) or the Indian Ocean SSTAs (Risbey et al., 2009; Ummenhofer et al., 2011) modulates the ENSO-ASR relationship. It was found (Power et al., 2006), for instance, that the negative PDO phase favours La Niña conditions over those of El Niño, and La Niña is correlated more closely with rainfall in Australia than is El Niño. The correlations are, therefore, stronger during negative phases of the PDO. It is indicated (Taschetto et al., 2011) that the basinwide warming of the Indian Ocean leads to a Gill-Matsuno-type  response (Matsuno, 1966; Gill, 1980) that reinforces the anomalies induced by changes in the SST in the Pacific Ocean. This warming, in particular, drives strong subsidence over Australia, and during January-March prolongs the dry conditions, when SSTAs related to El Niño begin to decay. As well as the anomalous circulation in the tropics, a basinwide warming excites a pair of barotropic anomalies in the Indian Ocean extratropics that induces an anomalous cyclone in the Great Australian Bight. It was found (Wu & Kirtman, 2007) that in the Indian Ocean air-sea coupling plays an important role in maintaining the negative ENSO-ASR relationship. The ENSO-ASR relationship is weakened significantly, or even masked by the internal variability, when the Indian Ocean is decoupled from the atmosphere. In summary, the modulation of a 3^rd factor could lead to failure of prediction by use of ENSO as the only predictor.

It is necessary to revisit the ENSO-ASR relationship, given the low prediction skill of ASR and the controversy in explaining why. This paper illustrates the role played by ENSO in forcing ASR, and explains how the SST warming in the Indian Ocean breaks down the ENSO-ASR relationship, by comparing the differences between the air-sea interaction before and after the decadal change in the ENSO-ASR relation.

Conclusions and discussion

An Interdecadal change between the leading mode of the ASR and ENSO is revealed by the present study. The PC of the first ASR mode had a close relationship with the TCP SST, prior to the mid-1980s, whereas after the mid-1980s this relationship broke down. It was suggested by further correlation analysis that the relationship between ASR and TCP SST is significant only in cases of negative TCP SSTAs. Also, the correlation between negative TCP SSTAs and ASR declines remarkably, from ID1 to ID2, which leads to the weakened ENSO-ASR relationship.

Mechanisms are proposed to explain the insignificant ASR-El Niño relationship and the weakening of the ASR-La Niña relationship, based on composite analyses for different scenarios. A Rossby wave lower-level anticyclonic anomaly appears to the southeast of the Australian continent, specifically, when the SSTA over the TCP is positive. The local air-sea interaction the WES feedback mechanism induces the anticyclonic anomaly. Another anticyclonic anomaly is induced over the northwest of continental Australia, meanwhile, as a result of the low level divergence and condensation cooling over the equatorial Maritime Continent. Weak circulation and precipitation anomalies over continental Australia results from the northerly and southerly anomalies of 2 separate anticyclonic anomalies competing with each other. As such, ASR does not have any kind of significant relationship with positive TCP SSTAs (or El Niño) for both ID1 and ID2.

Prior to the mid-1980s, condensation is induced over the southwestern tropical Pacific Ocean, when the SSTA over the TCP is negative. Together with the condensation cooling over the tropical Indian Ocean, a large-scale cyclonic anomaly is induced, which dominates the whole of continental Australia, leading to a strong relationship between La Niña and the leading mode of the ASR. After the mid-1980s, however, significant rainfall and condensation heating appear over the equatorial eastern Indian Ocean and Maritime Continent, as a result of the SST warming in the Indian Ocean. A cyclonic (low pressure) anomaly over the west of Australia, which competes with the cyclonic anomalies over eastern Australia, is induced by the heating over the equatorial Indian Ocean and Maritime Continent, and this leads to significant circulation and rainfall over Australia. The relationship between La Niña and the leading mode of the ASR, therefore, breaks down after the mid-1980s.

The complexity of the ENSO-ASR relationship has been unravelled by the study presented in this paper. The present study emphasises the role of the SSTA in the Indian Ocean in the modulation of the relationship, which differs from previous studies (Arblaster et al., 2002; Power et al., 2006; Risbey et al., 2009; Cai & Rensch, 2012; King et al., 2013) which supported the modulation of the PDO on the ENSO-ASR relationship. The tropical Indian Ocean SST is not tied to the PDO. A simple trend is shown in the TOD index during the period, with negative anomalies prior to the mid-1980s and positive anomalies after the mid-1980s, whereas there are multidecadal variability that is present in the PDO index, and its phase transition is not related to the variation of the tropical Indian Ocean SST. It is caused mainly by the warming trend in the Indian Ocean SST, though the correlation coefficient between PDO and TID index is 0.36, which is significant at the 99% confidence level. The correlation coefficient between the PDO and TID index is just 0.2, which is not significant, after both indexes have been detrended.

The heating of the equatorial Maritime Continent induces the northerly anomaly over the eastern flank of the lower level cyclonic anomaly (low pressure), and this counteracts the southerly anomaly over the western flank of the cyclonic anomaly (low pressure) that is induced by the cold TCP SSTA. The competition of the anomalous winds may be a partial explanation of the variability the ASR having larger amplitude prior to the mid-1980s when there was insignificant heating of the Maritime Continent, therefore no northerly anomaly is induced. Prior to the mid-1980s the standard deviation of PC1 is 1.30 mm/day, though after the mid-1980s it reduces to 0.61 mm/day.

According to Zhu the ENSO-ASR relationship obviously cannot be applied in a practical sense for the seasonal prediction of ASR, as the convection anomalies over the equatorial Maritime Continent and over the TCP both have impacts that are critical on the rainfall in the austral summer over continental Australia. They also suggest that attention should be paid to the precursors of condensation heating over the equatorial Maritime Continent as well as the tropical Pacific Ocean when empirical seasonal predictions of the ASR are made. Considering the effects from the Pacific and Indian Ocean SSTAs would improve the forecasting skill for ASR, given the warming trend over the Indian Ocean.

Note that summer rainfall in Australia, and ENSO, also present a weak relationship during the 1930s-1940s if data with a longer period is used. It has been found that a positive tropical Indian Ocean SSTA appears during that period. It was found that during that period tropical Indian Ocean SSTA appeared. According to Zhu it is also noteworthy that the mechanism proposed for the breakdown of the relationship between ASR and ENSO doesn’t apply in other seasons. The response of the atmosphere to identical tropical forcing of SSTA can be quite different during distinct seasons (Zhu & Li, 2016; Zhu et al., 2014).

The effect of the Atlantic Ocean SST is not discussed in the present study, as it is a long distance from Australia. It has been observed that the tropical Atlantic SST has a similar warming trend as that of the tropical Indian Ocean SST. The circulation over the Maritime Continent could be impacted by the tropical Atlantic SST by a “replying effect” of the Indian Ocean (Yu et al., 2016). The cross-basin effect on Atlantic on ASR is, as Zhu says, interesting topic that merits further investigation.

Sources & Further reading

1. Zhu, Z. (2018). "Breakdown of the Relationship between Australian Summer Rainfall and ENSO Caused by Tropical Indian Ocean SST Warming." Journal of Climate 31(6): 2321-2336.