Long‐term trend and decadal variability of the southward penetration of the East Australian Current

Abstract

[1] Observations from a long‐term ocean station off eastern Tasmania show that the southward penetration of the East Australian Current (EAC) has increased over the past 60 years. Changes in temperature and salinity are highly correlated at timescales greater than seasonal, with long‐term trends which differ markedly from global ocean values. The data show that the region has become both warmer and saltier with mean trends of 2.28°C/century and 0.34 psu/century over the 1944–2002 period which corresponds to a poleward advance of the EAC of ∼350‐km. These trends are not directly forced by global surface fluxes but primarily result from changes in the EAC. The summertime trends in temperature and salinity are greater than in winter – there is an augmented summer pulse of warm, high salinity subtropical water associated with the EAC.

1. Introduction

[2] In the waters off eastern Tasmania in the southern Tasman Sea, mean surface temperature maps show a broad, warm tongue of East Australian Current (EAC) water extending southwards [Walker and Wilkin, 1998]. From December to April, a further pulse of warm, saline EAC water is injected into the region along the continental slope [Cresswell, 2000]. This seasonal infusion of warm, salty, nutrient–poor subtropical water has an important influence on the regional climate, affecting both air temperature and rainfall [Langford, 1965], as well as providing a significant control on marine living systems [Edgar, 1984]. These consistent temporal and spatial patterns have fostered stable habitats, and the development of a distinctive range of ecosystems varying from north to south [Edgar, 1997].

[3] Observations from a long‐term hydrological station off the Tasmanian east coast show both a long term warming trend and a quasi‐decadal signal in ocean temperatures [Harris et al., 1987; Thresher et al., 2004]. These studies propose that the trend is evidence of a southward extension of the EAC. Other indirect evidence for EAC changes comes from biological sources. Several species previously only found in northern regions (sea urchins, shore crab) have steadily ranged further southward over recent decades [Edgar et al., 1997; Thresher et al., 2003; Pittock, 2003]. These changes have been attributed to enhanced EAC flow [Edyvane, 2003]. In this study we carefully examine the observations from the long‐term station to determine the true source of the property trends and their connection with the flow of the EAC.

2. Data

[4] A station off the east coast of Tasmania, (148.23°E, 42.6°S) at the 50‐m isobath near Maria Island (see Figure 1), has been regularly occupied since 1944 [Rochford, 1988]. Properties collected are temperature, salinity, oxygen, nitrate, phosphate, and silicate. The station sampling has varied between 2 to 6 weeks and samples have been collected at 10‐m depths between the surface and 50‐m. The samples have been processed at the main CSIRO Laboratory (from 1944–1984, Cronulla, Sydney and 1984–present, Hobart). The sampling and analysis methodology has changed incrementally over the 60‐year period. The station profiles were initially sampled with Nansen bottles, until the improved Niskin design became available. Temperature (T) is measured with deep‐sea reversing thermometers with an estimated accuracy of ±0.05°C. At the outset, salinity (S) was measured by the traditional Knudsen titration method, replaced with on board inductive salinometer processing during the 1950s [Brown and Hamon, 1961] and successively upgraded as newer systems have been implemented [Cowley, 1999]. Errors associated with these methods have steadily been reduced from 0.005 to 0.001 psu.

Figure 1

The February climatological sea surface temperature anomaly around Tasmania (°C) derived from composite SST imagery over the 1992–2006 period (Ridgway, submitted manuscript, 2007). The warm tongue of EAC water is evident off the east coast. The Maria Island station is located at 148.23°E, 42.6°S on the 50‐m isobath (black dot).

3. Maria Island Time Series

[5] Data from the Maria Island station provide a rare long‐term record of a suite of coastal ocean properties and their relationship to changes in offshore conditions in the region. The station is located on the inshore edge of the warm, saline tongue of EAC water that spreads southwards along the coastal boundary (Figure 1). Comparisons with satellite observations confirm that the data are representative of the core of the EAC flow (Figure 2). The station also samples the northward flow of cool, fresh subantarctic water that occurs over winter but is too far north to be influenced by the winter Zeehan Current inflows which originate on the west coast of Tasmania (K. R. Ridgway, Seasonal circulation around Tasmania: An interface between eastern and western boundary dynamics, submitted to Journal of Geophysical Research, 2007, hereinafter referred to as Ridgway, submitted manuscript, 2007).

Figure 2

(a) Surface T obtained at the Maria Island station (°C, 2–4 week sampling). T within the core of the EAC (148.6°E, 42.6°S) from a satellite composite is plotted in red. (b) The surface salinity (psu). (c) The decadal patterns of T and S at Maria Island (blue, cyan) and a satellite composite (red) after application of a Butterworth filter (5‐year time‐window) and the trend of the temperature (black). A mean has been removed from each property and it has been normalized by its standard deviation.

[6] Both T and S time series show a range of signals; seasonal, interannual, decadal and a long‐term trend (Figure 2). These signals either derive from direct mechanisms such as seasonal heating and cooling, river run‐off, precipitation (P) and evaporation (E), or indirectly by advection associated with the EAC. The raw temperature series has a very strong seasonal signal, this is dominated by atmospheric forcing – there is an annual range of ∼4°C (Figure 2a), however, the salinity record shows that there is also a seasonal component that has a purely advective origin associated with the EAC (Figure 1) (Ridgway, submitted manuscript, 2007). In this seasonal case the E‐P and run‐off contributions are only minor.

4. Long‐Term Trend

[7] The low‐passed T and S time series show a high level of coherence with a clear long‐term trend superimposed over a quasi‐decadal pattern (Figure 2c). These data show that the surface waters to the east of Tasmania have become both warmer and saltier with mean trends of 2.28 ± 0.35°C and 0.34 ± 0.03 psu/century over the 1944–2002 period (the trends are some 15–20% less at 50‐m depth). Are these property trends evidence of an enhanced poleward flow of the EAC or are other mechanisms responsible such as direct forcing by changes in surface fluxes? Unfortunately, even at the climatological mean level [Schmitt, 1995], surface fluxes of heat and E‐P are uncertain and hence we are unable to determine their long‐term trends for a localized region such as the southwest Tasman Sea.

[8] Our approach is therefore to compare this local signal with surface property changes over the South Pacific basin and the global ocean. Some idea of the spatial extent of the warming signal comes from global sea surface temperature (SST) analyses and summaries [Smith and Reynolds, 2003]. There is a warming trend over most of the global ocean but it is most pronounced in the southern hemisphere – in fact the North Pacific and North Atlantic show some cooling tendencies (Figure 3). The Maria Island station is now seen to be sampling a long‐term signal that has a limited spatial extent ‐ it encloses the waters surrounding Tasmania, extending north to the southern mainland coast and west into the Indian Ocean. From historical in situ observations, Holbrook and Bindoff [1997] identified a similar localized warming mode east of Tasmania in the upper 100‐m, obtaining a depth‐averaged trend of 1.5°C/century. These comparisons indicate that the Maria trend is larger than values observed in most other regions of the global oceans. The southwestern Tasman Sea trend of 2.28°C/century compares with a quite robust estimate of the global ocean trend of 0.6 ± 0.2°C/century which is distributed reasonably uniformly throughout the century [Smith and Reynolds, 2003] ‐ the Maria warming clearly does not simply represent a global background heating signal. If the global trend pattern is calculated over a longer period (starting in 1900) the values are reduced but there is still a maximum around Tasmania (1.36°C/century [Cane et al., 1997]). In fact, in the first half of the 20th century the trend off Tasmania is only 0.53°C/century with a rapid acceleration beyond 1950.

Figure 3

The trend in SST (°Ccentury⁻¹, 1944–2005) computed at each grid‐point of the ERSST (SST reconstruction) dataset [Smith and Reynolds, 2003]. Similar results are obtained if other global SST products are used.

[9] The comparison of the regional versus global pattern is even more stark for salinity. While no surface datasets exist to resolve the spatial patterns, zonally‐averaged salinity trends for individual ocean basins are available [Boyer et al., 2005]. In the latitude band of Maria Island, there is a negative trend of 0.05 century⁻¹ (from 1955–1998) within a general Pacific‐wide freshening pattern. Further evidence of surface freshening at high southern latitudes comes from trends in intermediate waters sampled between 1930 and 1980 [Wong et al., 1999]. More recent studies indicate a major freshening has occurred within the South Pacific gyre over the past decade [Roemmich et al., 2007]. The Maria salinity increase is therefore moving against the background salinity pattern. For example between Tasmania and New Zealand at 43°S, over a 25‐year period after 1967, the surface layer freshened apart from the waters adjacent to eastern Tasmania [Bindoff and Church, 1992].

[10] The Maria station appears to be sampling trends in T and S that are intensified around Tasmania, with magnitudes well above global values (and opposite in sign for salinity). The fact that this pattern is so localized and that both salinity and temperature change in concert, implies that the trends signify a change in circulation, rather than the direct influence of atmospheric forcing. In fact, model studies obtain similar property patterns forced by the component of the EAC that continues southward after the main separation of the boundary flow occurs north of Sydney [Cai, 2006].

5. Interannual to Decadal Variability

[11] The raw salinity series shows that there is significant variability at greater than seasonal timescales. For example there is intermittent energy within the 2–5 year band in both T and S (Figure 4). Harris et al. [1987] reported an ENSO signal in the Maria data based on a purported linkage between cooler summer maxima and El Nino events prior to 1975. Indeed, from the long time series a small but significant correlation is found between the T and S series and the Southern Oscillation Index (R = 0.17, at 9‐month lag), however, this explains less than 3% of the non‐seasonal, detrended variance. This small ENSO forcing is likely to come from the west along the coastal waveguide [Pariwano et al., 1986] with greatly diminished energy by the time it reaches Tasmania, although additional contributions may arise from the southwest Pacific [Holbrook and Bindoff, 1997].

Figure 4

(a) The hi‐passed (<10‐years) and detrended surface salinity anomaly at the Maria Island station. (b) The wavelet power spectrum of the time series in Figure 4a. The portion of the wavelet spectrum above the black line is significant at the 95% level.

[12] A much more robust signal is associated with the quasi‐decadal variability observed previously [Harris et al., 1987, 1988]. There are maxima in 1953, 1962, 1972 and 1990 and minima in 1957, 1965, 1984 and 1996 (Figure 2c). The property minimum observed in 1996 is expressed in SST patterns as a large cool anomaly surrounding Tasmania from February 1995 to August 1997 (results not shown). The mechanism underlying this decadal pattern is unknown, but the spatial structure of the signal may be extracted from the ERSST dataset – at decadal timescales the Maria data are highly correlated with the waters in an 800‐km radius around Tasmania (not shown).

6. Seasonal Cycle

[13] The seasonal pulse of warm, saline water associated with the EAC is clearly visible in the February SST field (Figure 1). However, here we focus on the salinity time series (Figure 4) to provide an understanding of changes to the seasonal expression of the EAC.

[14] The salinity series shows a distinct peak at the annual frequency (Figure 4b) which is indicative of the EAC high‐salinity summer pulse and a freshwater influx of subantarctic water from the south in winter [Harris et al., 1987]. However, over the 60‐year period, there are distinct changes in the magnitude of this annual amplitude. Prior to 1960, it is questionable whether a seasonal cycle exists, with only sporadic summer peaks in evidence. From 1970 to the mid 1990s the annual peak steadily increases (Figures 4a and 4b). Over the entire record there are 3 periods exhibiting a collapse in the annual cycle, 1957, and 1966 and most recently in 1996. While these 3 events coincide with minima in the low‐passed T and S patterns (Figure 2c) there was no corresponding seasonal collapse during the 1984 minimum (Figure 2). The 1996 event clearly involved a non‐appearance of the annual pulse of warm, saline EAC water – it was not due to an anomalously saline winter period.

[15] Is the increase in the salinity annual cycle also due to the EAC alone or are we able to identify changes during winter? There is no evidence of increased freshwater inflows in winter – in fact we observe a positive trend in salinity in all seasons. There is clearly an enhanced summer (warm, saline) flow of the EAC. The summer trends (Dec–Mar) are far greater, 0.48 ± 0.05 psu/century and 2.63 ± 0.38°C/century, compared to values of 0.20 ± 0.04 psu/century and 1.65 ± 0.22°C/century in winter (Jun–Sep). The dominant factor influencing the seasonal increase is thus the EAC water extending southwards. The seasonal expression of the alongshore geostrophic flow is driven by a cross‐shelf pressure gradient. The gradient is set up by the difference between the seasonal anomaly of the coastal sea level, established by the large‐scale wind stress pattern and the net offshore surface height, which is controlled by the eddy field (Ridgway, submitted manuscript, 2007). We are not able to unambiguously determine which of the two processes are behind the pressure gradient trend as there are no high quality time series of these components for the same period. Since the EAC seasonal cycle appears to be related to seasonal variations in the eddy field [Ridgway and Godfrey, 1997] we postulate that the trends observed here have a similar origin.

7. Discussion

[16] The intensification of the EAC flow past Tasmania is also seen in recent model studies describing both a spin‐up and southward shift of the Southern Hemisphere subtropical ocean circulation [Oke and England, 2003; Cai et al., 2005; Cai, 2006]. The oceanic changes are forced by an intensification of the wind stress curl arising from a poleward shift in the circumpolar westerly winds [Gillett and Thompson, 2003]. Both models predict that the EAC strengthens in the south while it weakens to the north. Cai [2006] used a linear Sverdrup model driven by trend‐fitted winds, to determine an EAC increase of 9 Sv south of 30°S from 1978 to 2002.

[17] The Maria series strongly implies that the EAC has strengthened in the southern Tasman Sea. It is interesting that the change in EAC surface salinity from winter to summer (0.25–0.30 psu from 1989 to 1990 [Thresher et al., 2004]) is of the same order as that observed over the 60‐year period at Maria Island. This corresponds to a poleward extension of some 350‐km in T and S properties. If there is a similar long‐term change in EAC transport it would suggest an increase of 10–15 Sv over the 60‐year period. This is about the value proposed by Cai [2006]. Note also that in the northern EAC region a significant thermocline cooling has been observed from 1975–1990 [Ridgway and Godfrey, 1996] which matches the weakening of the EAC in this region observed in the models [Oke and England, 2003; Cai et al., 2005].

[18] The results presented here strongly suggest that the Maria Island station has captured a real change in the EAC circulation since 1944. Further independent but more indirect evidence from deepwater corals implies that the 60‐year trend observed in the Maria data may form a relatively short segment of a variation in EAC behaviour that has persisted for at least 300‐years [Thresher et al., 2004]. Such a long‐term adjustment of the EAC would certainly predate the anthropogenic forcing mechanisms suggested by recent studies [Gillett and Thompson, 2003; Cai, 2006] and additional natural varying processes would have to be involved.

Acknowledgments

[19] This unique climate record at Maria Island is due to the efforts of many technicians who have faithfully collected, processed and archived the data since 1944. I especially acknowledge the late David Rochford who had the great foresight to establish a network of ocean monitoring stations. Thanks to Jeff Dunn and Frederic Saint‐Cast for their helpful comments. This work is part of the CSIRO Wealth from Oceans Flagship.