Arterial ocean circulation of the southeast Indian Ocean



[1] The time mean ocean circulation for the southeast Indian Ocean (SEIO) is analyzed using a colorwheel to reveal an “arterial-like” structure. Although quasi-zonal, the mean circulation in this region is distinctly more complex than previously presented in recent literature. The mean flow is derived from an eddy-resolving ocean reanalysis of the past 14 years. The ocean state is constrained through data assimilation with observed sea level anomaly, sea surface temperature and in-situ profiles. Geostrophic currents derived from the reanalyzed mean dynamic topography (MDT) are validated against two other MDT analyses to show a comparable arterial structure and distribution of scales. In particular, the broad scale eastward flow is regularly interspersed by narrow bands of the near westward flow. Meridional sections from the ocean reanalysis reveal a net eastward transport in the surface layer and a net westward flow at mid-depth. The seasonality of the surface layer flow in the SEIO appears to have increased (decreased) eastward mean transport that coincide with seasons of stronger (weaker) Leeuwin Current (LC). Whether the narrow bands of mean current and mass flux are simply the average of the eddies or represent an underlying weak jet remains a topic of current debate.

1. Introduction

[2] The upper ocean circulation off the west coast of Australia is dominated by the LC system [Cresswell and Golding, 1980] and the West Australian Current [Andrews, 1977]. The LC owes its existence from an anomalously large along shore near surface pressure gradient. A few hundred meters below the surface the LC meridional pressure gradient reverses direction resulting in equator-ward Leeuwin Undercurrent [Godfrey and Ridgway, 1985]. The open ocean circulation of the tropical SEIO also includes the eastward flowing South Java Current [Sprintall et al., 1999], the fast flowing fresher South Equatorial Current (SEC) [Wijffels et al., 2002] and the Eastern Gyral Current [Meyers et al., 1995]. Other regional wind driven circulation features include the equator-ward flowing Ningaloo Current [Taylor and Pearce, 1999] and Capes Current [Pearce and Pattiaratchi, 1999]. Recently Lagrangian studies have revealed pathways of the LC system [Domingues et al., 2007]. Through these pathways the LC connects to the wider SEIO and draws water masses from the tropics and subtropics. Another remarkable aspect of the upper-ocean circulation of the SEIO is the near surface eastward flow [Siedler et al., 2006; Palastanga et al., 2007]. This eastward flow is called South Indian Counter Current (SICC) owing to its ability to flow against the prevailing winds. The SICC originates from the coast of Madagascar and flows eastward between 22–26°S. The SICC is a narrow jet like flow that becomes more spread out east of 90°E. Little is known of the SICC east of 105°E and its influence on LC system.

[3] Until recently, it has been presumed that the geostrophic turbulence in the ocean is unorganised such that any mean patterns are a result of under sampling due to the finite averaging period. More recently, analyses have revealed coherent mean patterns that are under investigation for potential underlying organization, though Schlax and Chelton [2008] have highlighted how challenging this pursuit will be. The recent discovery of multiple long term open ocean zonal striations from high resolution numerical models [Nakano and Hasumi, 2005; Richards et al., 2006] and satellite observation analysis products [Maximenko et al., 2005; Maximenko et al., 2008] have drawn attention and invoked debate as to their origin. Geostrophic turbulence theory indicates that zonal jets form on a beta plane [Rhines, 1975] and the resemblance of these mean ocean zonal striations to those observed on the giant planets [Galperin et al., 2004] has been noted. Studies of eddy induced zonal velocity bandings have shown features with comparable width and magnitude to striation structures found from satellite altimetry [Schlax and Chelton, 2008]. In this paper, we analyze the complex current patterns of the SEIO from an eddy resolving 14-year ocean reanalysis. We first validate the reanalysis MDT against other MDT estimates. We then examine the reanalysis annual and Austral seasonal mean currents to reveal the surface and sub-surface structure of these alternating quasi-zonal bandings to highlight properties in this region.

2. Ocean Model and Reanalysis

[4] The results presented are drawn from a 14 year ocean reanalysis (BLUElink > ReANalysis 2.1; BRAN2.1) providing a “best” ocean state estimate from 1993 to 2006. BRAN2.1 combines the Ocean Forecasting Australia Model (OFAM) with ocean observations using the BLUELink Ocean Data Assimilation System (BODAS) to provide the first high resolution ocean reanalysis for the Asian-Australian region [Schiller et al., 2008]. OFAM is a global model based on MOM4p0d [Griffies et al., 2004], with 1/10° horizontal resolution around Australia (90–180°E, south of 16°N) and 47 vertical levels. The model was initialized by a climatological ocean state and forced by ERA40 (1993 to mid 2002) and ECMWF (mid-2002 to 2006) surface fluxes. BODAS uses an ensemble optimal interpolation scheme based on a stationary ensemble of modeled anomalies from the forced model integration without data assimilation [Oke et al., 2008]. BODAS employs a seven day analysis cycle assimilating all available satellite altimeters (ERS, GFO, Topex/Poseidon, Envisat and Jason), SST observations from Pathfinder and AMSR-E missions and in-situ profiles from Argo floats and XBT. The quality of BRAN2.1 has been objectively compared with other ocean state estimations [Hernandez et al., 2009] demonstrating comparable or superior performance in the SEIO. In general, the performance is expected to decline below the thermocline as the background error covariability of remotely sensed observations reduces, although this is partly compensated by a reduction in ocean state variability. The BRAN2.1 MDT surface geostrophic currents are compared with that from MDT by AVISO [Rio et al., 2009] and Asia-Pacific Data Research Center (APDRC) [Maximenko and Niiler, 2005]. The AVISO MDT estimation represents the period 1993–1999 with 0.25° horizontal resolution, which uses ocean observations, e.g., ocean drifters and hydrographic profiles. The APDRC MDT is for 1992–2002 period and has global horizontal resolution of 0.5°, which combines ocean observations in the context of surface horizontal momentum balance. In this study we analyze data for the SEIO in the domain 50°S to the equator and 90°E and 120°E.

3. Results

[5] The surface geostrophic currents from BRAN2.1, AVISO and APDRC MDTs are shown using a colorwheel for magnitude and direction (Figure 1). Figure 1 shows the westward flowing SEC at 10°S is the dominant feature with mean speeds exceeding 0.1 m s−1. Between the SEC and 30°S the circulation is dominated by a broad eastward flow regime towards the West Australian coast, which is regularly punctuated by narrow fingers of near westward flow. Unlike in other oceans eastward geostrophic flow is stronger and broader in the SEIO (see Figure S1 of the auxiliary material). These broad eastward flows reach current speeds up to 0.1 m s−1 and in general are consistent with the three MDTs. In Figure 1a we identify five eastward flowing jets of around 2–3° meridional width embedded in the broad eastward regime, of which four of them show connection to the West Australian coast (also see Figure S2). The eastward jets in the tropics turn toward the coast passing over the Exmouth Plateau and the Northwest Shelf region. The 27°S eastward flow bifurcates along 105°E and continues to the coast as two branches. Immediately poleward of this broad eastward flow at ∼33°S, there is a narrow westward flow with a small but steady northward component between the Naturaliste and Broken Plateaus. The magnitude of this westward transport at the surface weakens as it leaves the Australian coast and the Naturaliste Plateau. The tail of this flow joins with the westward flowing Flinders current of the Great Australian Bight (GAB); which indicates the dynamics are not entirely local to the SEIO [Middleton and Bye, 2007]. South of this westward feature, the surface circulation of the SEIO becomes complex with no clear mean flow direction. Poleward of 40°S the circulation is dominated by the fast eastward flowing Antarctic Circumpolar Current. Even though the geostrophic currents from the reanalyzed MDT and other MDTs agree fairly well with each other, we note that there are some discrepancies between each plot in Figure 1 especially in regions of SEC and south of 35°S. These variations indicate some sensitivity to the differences in data products, the observation period and method used with each of the analysis product. Similarly, regions with consistent features are an indication that they are robust and relatively insensitive to the variations amongst the products.

Figure 1.

Geostrophic surface currents derived from MDT (a) BRAN2.1, (b) AVISO, (c) APDRC; represented using a colorwheel for both direction and magnitude (m s−1), with bathymetry contours overlaid (500:1000:3500 m).

[6] A meridional depth section of mean reanalysis currents along 100°E, 105°E and 110°E (as indicated in Figure 1a) is shown in Figure 2. This shows that the features at the surface extend coherently towards the abyssal ocean, but decline in magnitude as seen in other high resolution model studies without data assimilation [Richards et al., 2006]. These vertically coherent mean features are similarly quasi-zonal. The eastward surface flow deepens and becomes more dominant as it approaches the West Australian coast. The westward flow dominates at mid-depth indicating that a more dense water mass with a strong flow at 33°S is present closer to the coast, similar to that noted by Domingues et al. [2007]. Eastward flows are largely confined to the upper 250 m of the section with weaker and narrower abyssal extensions at mid-depth. We note that there are coherent westward mean flow structures with velocity magnitudes close to 0.1 m s−1 at depth that weaken and become narrower toward the surface. These depth-intensified structures show limited signatures at the sea surface, making them difficult to verify using present day observations. The northern aspects of each bathymetric feature in Figure 2 consistently show westward flow. Conversely eastward flow consistently aligns with the southern aspects of the bathymetry. We suggest the bathymetry acts to steer the mean currents that contributes to the arterial structure of the region. The alignment of flow direction on bathymetric features and resultant alternating zonal flows around the topography seen in Figure 3 is consistent with anticyclonic vorticity production and subsequent vortex stretching [Stommel, 1982].

Figure 2.

Meridional section of the mean current along (a) 100°E (b) 105°E and (c) 110°E from BRAN2.1 (as shown by meridional lines in Figure 1a); represented using the same colorwheel as in Figure 1, the black line represents the 250 m depth contour. A block representation of colorwheel used is given at the bottom.

Figure 3.

BRAN2.1 seasonal mean depth averaged (25–250 m) currents (a) Annual mean, (b) Austral Summer (DJF), (c) Austral Autumn (MAM), (d) Austral Winter (JJA), and (e) Austral Spring (SON); using colorwheel for both direction and magnitude (m s−1) with bathymetry contours overlaid.

[7] BRAN2.1 annual mean currents for surface layers at different depth levels are represented in Figure S3. Regions of strong wind stress (see Figure S4) show Ekman drift to the left of the direction of the wind as seen in the colorwheel diagram (Figure S3a). In the tropics the surface currents at 5 m depth exceed 0.1 m s−1 with a west-southwest direction. Poleward of 35°S, surface currents are directed north-northeast. Similar to the broad eastward geostrophic flow (Figure 1a) we can identify embedded zonal features of higher magnitude in the broader Ekman drift especially 15–25°S. Regions of weak wind stress reveal the underlying geostrophic zonal flows eg: the westward flow connecting the Naturaliste and Broken Plateaus, eastward flows north of the Broken Plateau and the Northwest Shelf. Below the surface layer at 15 m depth (Figure S3b) the wind driven effects diminish and the geostrophic flow dominates in most parts other than regions of strong wind stress, notably south of 35°S. At 25 m depth (Figure S3c) the wind driven effect becomes negligible and the ocean current is mainly constrained by the quasi-geostrophic flow that includes broad scale eastward flows towards the coast in the tropics (as seen in Figure 1).

[8] BRAN2.1 mean depth averaged currents (25–250 m) for the SEIO is shown in Figure 3. This reveals a more complex view of the current system than the mean surface flow patterns in Figure 1a, or the simpler alternating zonal flow structures seen in other high resolution models [Richards et al., 2006]. This sort of flow structure, which we refer to as the “arteries” of the ocean circulation, show a variety of pathways and a range of directions, with meandering and meridional connections. Zonal features in depth averaged plots show a range of meridional widths with a larger number of smaller filaments <2°. The annual mean pattern (Figure 3a) shows currents with velocity magnitudes close to 0.1 m s−1, particularly nearer to the equator and between 20°S and 30°S. There is a strong westward flow connecting the Naturaliste and Broken Plateaus, which is clearer in Figure 3 than in Figure 1, indicating that the flow strengthens below the surface. The eastward flow becomes more band-like in depth averaged plots, and comparable in width to alternate westward flows. The strongest eastward band is at 28°S, flowing towards the coast through the northern edge of the Naturaliste Plateau.

[9] There is considerable seasonality in the depth averaged (25–250 m) upper layer currents (Figures 3b3e), notably in the coastal region and along the SEC. The Austral Summer and Autumn periods (Figures 3b and 3c) show greater westward flows in the tropics; the SEC is broad and appears to consist of a large number of filaments. During Autumn; when the LC begins to strengthen [Feng et al., 2003], we can see alternating dipole structures throughout the western continental shelf of Australia. This persistent seasonal dipole structure is consistent with the formation of cyclonic and anticyclonic eddy pairs during these periods [Meuleners et al., 2007]. In Spring and Summer months (Figures 3e and 3b, respectively) the mean current originating in the GAB bifurcates south of Naturaliste Plateau. Interestingly this bifurcation is not evident in Winter periods. The equatorial alternating current system develops during Summer months. Another notable feature is the north-westward flowing mean jet that emanates during Spring from the combined effect of currents coming from either side of the Cuvier Plateau. The SEC becomes more consistently zonal and an eastward flow forms to its south during Winter and Spring (Figures 3d and 3e). The mean westward flow weakens and eastward and southeastward flows strengthen in the surface layer circulation during Winter and Spring periods, coinciding with periods of strong LC. These complex eastward flows are believed to contribute the mass flux to help strengthen the LC transport. The reason behind these more zonal eastward flows in the surface and westward flow underneath (see Figure 2) is not known. One cause that has been proposed in the SEIO is subduction. This subduction drives eastward currents at the surface which sinks along the coastline of Western Australia. Through this process their water masses are transformed and return to the open ocean at mid-depths as westward flows [Domingues et al., 2007]. Occurrence of offshore positive steric height anomalies during strong LC are also identified [Godfrey and Ridgway, 1985]; suggesting a pile up of watermass in the SEIO during May–July.

4. Discussion

[10] This study shows relatively broad geostrophic eastward surface flows embedded with a series of eastward jets, and punctuated by narrow fingers of westward flow of the SEIO, using an ocean reanalysis and available high resolution MDT datasets. By applying a colorwheel visualization for total current, we have shown that the current patterns in the high resolution model and MDT are much more complex that are otherwise masked by the conventional visualization of zonal current in the eddy resolving datasets. These circulation patterns in the SEIO are more arterial with considerable meridional variations. The prominent eastward flow patterns at the surface and westward flow patterns at mid-depth give some promising insights into the recently discovered LC source flow regions in the SEIO [Domingues et al., 2007]. The increase and decline of these eastward flow structures closely coincides with the phases of the LC, making an interesting link with the LC dynamics. There is a specific orientation of reanalysis currents near bathymetry, indicating they are on average influenced by and interacting with the bathymetry. Some limitations to the accuracy of the ocean reanalysis in representing the mesoscale variability have been previously mentioned. Altimetry, SST and the sub-surface observations taken independently are sparse relative to the mesoscale. The BRAN2.1 provides a best-estimate of the ocean state combining observations with an ocean model background. Comparisons with independent MDT products have provided some confidence that the ocean reanalysis can explain the mean mesoscale structures found in this region. The dominance and consistency of the mean circulation between the Naturaliste and Broken Plateaus, its observable surface signature and position coinciding with low wind stress is proposed as a suitable region for further analysis and observational campaigns.


[11] The authors would like to thank K. Walsh, N. Maximenko, M. Feng, P. Sandery and BLUElink Ocean Forecasting team for useful comments and access to data; APDRC and AVISO for providing MDT datasets. PD acknowledges the support of the University of Melbourne scholarship program.