1. Background
The long-term mean transport between the Pacific and Indian Oceans, known as the Indonesian Throughflow (ITF), is widely recognized as important for global and regional climate as it redistributes ocean heat and affects the atmospheric circulation (Godfrey 1996; Schneider 1998; Wajsowicz and Schneider 2001). However, it has been difficult to quantify the ITF and its heat transport from observations because of the very large temporal and spatial variability of the currents in the Indonesian Throughflow region. In addition, the interbasin exchange is broken into several filaments passing through the complex bathymetry of the Maritime Continent (Fig. 1), making its measurement logistically challenging.
The main upper branch of the ITF passes through Makassar Strait into the Java and Banda Seas, before exiting through the primary outflow passages: Lombok Strait, Ombai Strait, and Timor Passage (Fig. 1). Intermediate and deep waters likely also flow southeast of Sulawesi, Indonesia, through the Malukus (Cresswell and Luick 2001; Gordon and Fine 1996; Talley and Sprintall 2005). The deepest exchange between the Banda Sea and the Indian Ocean is believed to be restricted to above 1250 m (Molcard et al. 1996).
Meyers et al. (1995) calculated the 0–400-m geostrophic flow and documented the main flow pathways through the region using the first 6 yr of repeat expendable bathythermograph (XBT) sections established by Australia’s Commonwealth Scientific and Industrial Research Organisation (CSIRO) and now run by the Joint Australian Facility for Ocean Observations (JAFOOS) of CSIRO and Australia’s Bureau of Meteorology (BOM). A broad regional spatial context is documented in Qu and Meyers (2005) based on heavily averaged scattered historical data. Shallow pressure gauges have been deployed across the outflow passages for several years and ship-based velocity surveys were carried out to calibrate geostrophic transport estimates for the upper 100-m flow (Hautala et al. 2001, Sprintall et al. 2003). The synoptic velocity fields in the straits revealed strong shear and often flow reversals with depth (Hautala et al. 2001).
Past mooring work has also seen several of the major ITF straits instrumented for periods of up to 1 yr (Murray and Arief 1988; Cresswell et al. 1993; Molcard et al. 1996; Gordon et al. 1999; Luick and Cresswell 2001; Cresswell and Luick 2001) but never have all the channels been measured at one time. This is being achieved by the International Nusantara Stratification and Transport program (INSTANT) currently underway (Sprintall et al. 2004). Though spanning all depths, INSTANT will only deliver a 3-yr record and interannual variability in the region is very strong (Meyers 1996; Wijffels and Meyers 2004). Thus, an accurate estimate of the long-term mean total upper transport of the ITF and its heat transport requires a multidecadal dataset, and for the ITF such measurements exist only in the form of the repeat XBT lines that span the major current systems in which the ITF is imbedded. Here we exploit a 20-yr record of repeat track XBT data first reported by Meyers et al. (1995), taking advantage of the increased sampling between 400 and 750 m to present a more refined and quantitative description of the mean flow across the XBT lines, complementing Qu and Meyers’ (2005) broad-brush description of the outflow region.
2. Data and methods
Around 1984 two XBT lines were established by CSIRO and BOM, where temperature recording probes are deployed off volunteer merchant ships in a routine sustained manner (Fig. 2). The IX1 line spans the eastern South Indian Ocean between southwestern Australia and the western tip of Java. The line PX2 cuts across the lower Banda Sea from the shelf break of the Java Sea in the west to the shelf break off northwest Australia (Arafura shelf) in the east. Another line, IX22, was established a few years later, sampling from the Australian Northwest shelf (NWS) northward past the western end of Timor Island, across the Savu Sea and Banda Sea. Temporal sampling along IX1 is roughly twice per month and only once per month along PX2 and IX22. Temperature probes are dropped roughly every 3–4 h, thus sampling every 100–150 km along the line. The ship’s crew made an effort to drop a probe on the 200-m isobath at the start and end of each line, helping constrain transports to the shelf break. Here we analyze the data collected up to the end of 2006.
The throughflow region is characterized by very strong internal tides (Ffield and Robertson 2005; Katsumata and Wijffels 2006), which impose large ageostrophic short vertical scale temperature variability that negates the use of synoptic sections within the internal seas for geostrophic transport estimates. This problem particularly affects the PX2 and IX22 datasets, but we believe that the large amount of data collected over the last 20 yr makes it possible to remove this noise through averaging to reveal the mean flow structures.
The JAFOOS XBT data were quality controlled to the high standards set during the World Ocean Circulation Experiment (WOCE) (Bailey et al. 1994). They were then gridded in space and time using a simple parametric fitting technique similar to that of Ridgway et al. (2002), but the functional fits were achieved using a robust fitting method rather than conventional least squares to avoid biases from outliers [as implemented in MATLAB’s robust-fit routine and based on Holland and Welsch (1977)]. Besides spatial polynomials, a mean seasonal cycle was represented as annual and semiannual sinusoids and was fitted independently at all depths. We describe the annual cycle elsewhere. Here we present the mean fields.
Density was calculated from the XBT temperature fields by using the seasonally varying potential temperature/salinity relation from the Climatology of Australian Regional Seas (CARS; Ridgway et al. 2002), for which considerable effort was made to prevent averaging properties across island chains and ridges (Dunn and Ridgway 2002). Geostrophy was then applied to the gridded density fields to produce velocity relative to the deepest available depth (either 750 m or the bottom, whichever is shallowest).
As the net transports across the line are sensitive to how well boundary currents are resolved, the scales used to normalize distances before the parametric fit was performed were reduced near the shelf breaks. In particular, off Western Australia the normalizing zonal scale is 0.2° and off Java the normalizing meridional scale is 0.25°. Experiments showed, however, that results are less sensitive to mapping scales than to the choice of the number of profiles fitted, as a larger number of profiles required fitting data over a wider geographical area. We found that fitting 750 observed temperature profiles every 30 km along the line resulted in the best resolution of flow structures and yet produced solutions that are smooth. For this number of profiles in the 20-yr dataset, we are typically fitting data that span a 150- and 300-km along-track radius around the fitted grid point for IX1 and PX2/IX22, respectively. Since the parametric fit involves 2 degrees of freedom (linear and quadratic in space), the resolved scales might be estimated as being roughly 70 and 150 km for IX1 and PX2/IX22. Differences in total transports were somewhat sensitive to these choices but they remained within the error bars reported here.
The standard errors in the temperature field maps derived from the robust fits are also used here. To validate them in the context of our dataset, our parametric model was tested on synthetic datasets with the same temporal and spatial sampling as the XBT data, but perturbed with specified levels of Gaussian random noise to generate 100 datasets. On fitting the 100 datasets with our method, we found that the standard errors reported by the robust fit matched the standard deviations of the ensemble of solutions around the “true” solution. Hence, we can confidently interpret the reported standard errors as the expected standard deviation of the solution given the number of observations and their noise.
Generally the temperature errors are largest where the temperature variance is large—in the seasonal thermocline, tropical thermocline, and coastal waveguides. See Wijffels and Meyers (2004) for a more detailed description of the temperature variability along these lines at different time scales. The errors for mean temperature were translated to a transport error by distributing them across the section in such a way as to maximize the transport error: cold errors at one end of the section linearly progressing to warm errors at the other. Transports calculated with the error-perturbed temperature field were differenced from the original transports to produce the transport error. Thus, the transport errors reported here are the maximum given the errors in mapping the temperature field.
Liu et al. (2005) find that the choice of salinity/temperature relation used to calculate density also has an effect on the resulting transports. Changing from the more highly smoothed Levitus (1982) climatology to the finer CARS climatology increased total geostrophic section transport by 20%, which is within the error bars cited here.
Ekman transports were calculated using two wind products: the National Centers for Environmental Prediction’s NCEP1 (on a 2° by 2° grid; Kalnay et al. 1996) and the Quick Scatterometer (QuikSCAT) winds (on a 1° by 2° grid; CERSAT 2002). The Ekman velocity was distributed linearly from the surface to a seasonally varying mixed layer depth determined from the varying temperature data at each grid point. Mixed layer depth was defined as the shallowest depth where temperature is 0.5°C colder than that at the surface. Errors in the Ekman flux were estimated as 20% of the total amplitude based on differences among wind stress products (Wijffels et al. 1994). A comparison of the Ekman transports for the shorter QuikSCAT record (from late 1996 to the present) compared to the longer NCEP1 wind stress dataset shows agreement within this error bar.
Using the geostrophic plus Ekman velocity fields and the CARS salinity field, we have integrated the transports of the major currents and calculated their transport-weighted temperature (TWT) and salinities (Table 1). The zero velocity line was used to demarcate the boundaries between the currents, rather than using fixed depth or latitude definitions. The depth-averaged flow vectors are shown in Fig. 2, dyed by their associated transport-weighted salinities, and mean cross-transect velocity sections are presented in Fig. 3.
3. Regional currents and flow pathways
The locations of the major flows across the XBT lines (Fig. 2) are similar to those deduced by Meyers et al. (1995). The ITF crosses the PX2 line along the Java Sea shelf break south of Makassar Strait near 115°E (Fig. 2a). We shall refer to this strong southward flow as the Makassar Jet (MJ). The MJ flows over the complex and wide slope between Makassar and the Java Sea shelf break, transporting about 5.6 Sv (1Sv ≡ 106 m3 s−1) at 20.4°C (Table 1). It is subsurface intensified (Fig. 3b), with a velocity maximum core near 100 m, which Gordon et al. (1999) observed with direct mooring measurements upstream in the Makassar Strait near 3°S at the Labani Channel. They estimated transports of 9 Sv; the difference is most likely due to our shallow reference depth over the Flores Sea slope of about 300– 400 m. Hence the transport is likely to be higher and temperatures lower in the MJ if the barotropic component is included. Interannual variability may also account for some differences between the mooring data and our 20-yr average.
Upper 100-m transport-weighted salinities indicate that the fresh MJ directly feeds the eastward flow along the northern coast of the Nusa Tenggara island chain, which we shall call the Nusa Tenggara Current (NTC). The PX2 line crosses this flow at an oblique angle and so does not measure it well, though mass conservation and vorticity dynamics would suggest it should follow the topography eastward. The transport of the NTC is also 5.6 Sv at 18.5°C and, since part of the MJ must also deviate westward to feed the flow through Lombok Strait (∼1.7 Sv; Murray and Arief 1988); this again suggests that our shallow reference velocities in the MJ cause an underestimate of its transport by 2–3 Sv. It may also be true that where the PX2 line transects the NTC at such an oblique angle, ageostrophic effects might also reduce the accuracy of our results regarding the NTC.
The PX2 line cuts across the entrance regions of the Ombai and Timor Throughflows, and these feeder flows are clearly evident in Fig. 2a as southward vectors east of 125°E. The Ombai Throughflow transports about 3.1 Sv, and similar to the MJ features a distinct subsurface maximum but one that is located deeper, near 150-m depth. In addition the Ombai Throughflow features strong shear down to 700 m (Fig. 3b) and is cooler than the MJ (15.5°C, Table 1). In comparison, the Timor Throughflow is surface intensified, with most of its shear above 200 m, it transports 4.0 Sv, and at 21.6°C it is warmer than either the MJ or the Ombai Throughflow. There is a hint that the shear in Timor reverses at depth in the main channel. This striking difference in vertical structure between the Ombai and Timor Throughflows has been observed by direct mooring observations (Molcard et al. 1996, 2001).
Interestingly, the inflow to the Timor Passage is divided into two jets along PX2—a deep narrow flow close to Timor and a broad shallower one over the Australian shelf slope (Fig. 2a). Farther east along PX2 is a northward flow along the upper Australian slope—the Rochford current—first noted by Godfrey and Mansbridge (2000). In the upper 100 m, the salinities of the Ombai Throughflow are fairly fresh, while those in the Timor Throughflow are slightly saltier (Fig. 2a) and thus suggest that while the Ombai Throughflow is directly supplied by the NTC, the shallow Timor Throughflow is supplied mainly via a recirculation from the northern Banda Sea where the upper thermocline is saltier. Below 100 m, the salinities also support this pathway (Fig. 2b).
The IX22 line independently samples the Ombai and Timor Throughflows just downstream of PX2 (Fig. 2). On this line, the flow through the Ombai/Savu Sea has about the same transport, temperature (though slightly warmer), and salinity as found on PX2 (Fig. 2a; Table 1), and features the same deep subsurface velocity maximum near 150 m and deep-reaching shear (Fig. 3c). This general agreement between these two independent estimates of the flow gives greater confidence in our results.
IX22 samples the Timor Passage at its western end (Fig. 2a) where we find a single surface intensified jet (Fig. 3c). Transport increases between PX2 and IX22 from 4 Sv at inflow to 5.6 Sv at outflow, with temperatures and salinities both increasing (Table 1). How is this possible in only 600 km? South of Timor Passage along IX22, we find shallow flow of warm and salty waters toward Australia, here termed the NWS inflow [or the eastern extension of the Eastern Gyral Current (EGC); Fig. 3c]. The change in the Timor flow can then be accounted for if 1.6 Sv of the NWS inflow recirculates northward along the Australian slope to augment the Timor Throughflow, leaving 1 Sv to possibly supply the Leeuwin Current via flows on the shelf that are not captured by the XBT lines.
In the Banda Sea, the IX22 line crosses the NTC just west of the Ombai Strait (Fig. 2). Here we find a deep-reaching flow of 4.7 Sv but in contrast to the MJ, it is surface intensified (Fig. 3c), again suggestive of the importance of recirculations within the Banda Sea. North of the NTC on IX22, we encounter a weak shallow westward flow in the central Banda Sea (0.6 Sv) and then a very strong subsurface intensified eastward flow south of Buru Island transporting 3.6 Sv of water saltier than that found in the MJ. Since Lombok Strait carries about 2–3 Sv (Hautala et al. 2001), the total eastward transport through the Banda Sea of 7.7 Sv across IX22 is likely supplied by southward salty flow west of Buru from the Maluku Sea, in addition to the Makassar inflow.
The Lombok, Savu, and Timor Throughflows join to form the northern component of the South Equatorial Current (SEC; South Eq. C I in Fig. 3a), seen as strong westward surface-intensified flow carrying 9.5 Sv between 10° and 15°S on IX1. North of the SEC, very fresh, warm, and shallow eastward flow hugs the Java coast—the 1.1-Sv South Java Current (SJC) flowing from the high-rainfall region just west of Sumatra. While the SJC reverses seasonally (Quadfasel and Cresswell 1992), our data show that on the annual mean it is eastward. Its warm fresh contribution must ultimately recirculate eastward in the SEC, but clearly its properties are much modified before doing so. The freshening effect of the SJC must also be offset by the shallow salty eastward transport of 2.6 Sv of the EGC, though some part of this flow feeds the southward Leeuwin Current off Western Australia. Below the main thermocline, salinities increase by 0.1 psu between the exit straits and the SEC along IX1 (Fig. 2b), indicative of strong lateral mixing in the region. Below the shallow SJC, we find a second core of eastward flow, the South Java Undercurrent (Figs. 2b, Fig. 3a). A core of high-salinity water has implied such a mean flow (Fieux et al. 1994; Wijffels et al. 2002); we detect it here transporting 1.1 Sv eastward at 10.3°C.
South of 15°S is a westward subsurface current organized into two cores—one with a maximum velocity at 400 m near 20°S and the other at 150 m near 17°S. The two-core structure is robust in our analysis. We label these two cores together the South Equatorial Current II. Qu and Meyers (2005) also note this deep and second component of the SEC, which is the deep expression of the South Indian Ocean’s Sverdrup gyre that carries relatively newly ventilated high-salinity waters from the south (Fieux et al. 1994; Wijffels et al. 2002). The high salinity of the southern core of the SEC II means that the bulk of this flow must be supplied by the northward flow offshore and below the Leeuwin Current, which has similar salinities and encompasses the Leeuwin Undercurrent (Domingues et al. 2007). The fresher northern shallower core of the SEC II is likely partly supplied by modified EGC and ITF waters.
Based on the flows across the XBT lines, transport-weighted salinities, and rough mass conservation, a schematic of the main currents and their sizes is presented in Fig. 4. The primary pathway remains as that deduced by Meyers et al. (1995)—Makassar to Nusa Tengarra Current to Ombai and Timor Straits. However, the smaller regional flows augmenting and connecting to this pathway are now evident and can be quantified.
4. The total interbasin volume and heat exchange and regional heat budget
The 20-yr average volume transports (geostrophic relative to 750 m plus Ekman) and associated transport-weighted temperature and salinities for the IX1 and PX2 sections are shown in Table 2. While the two shelf-to-shelf XBT lines, IX1 and PX2, comprise two completely independent estimates of the ITF fluxes relative to a 750-m reference level, the PX2 line is less suited to estimate total transports as it traverses the broad and complex slope/reef region in the western Flores Sea where bottom velocities below the MJ are likely significant. Larger internal tides and a less narrowly defined shipping track also reduce its reliability for transport estimates. We believe the IX1 line represents the most accurate estimate of the 20-yr average ITF transport relative to 750 m, which is 8.9 ± 1.7 Sv. Across PX2, the transport is about 2 Sv less than that across IX1, and the difference can be accounted for by a plausible ∼3 cm s−1 flow at the shallow bottom reference depth under the MJ.
The section average vertical structure of transport clearly shows the subsurface maxima in the geostrophic flow (Fig. 5a) near 120 m on IX1 and 100 m on PX2. Near-surface average geostrophic velocity is toward the Pacific Ocean along IX1, but is largely cancelled by the strong Ekman flux in the section mean. In potential temperature classes the two sections give somewhat similar profiles, with a shallow transport peak near 27° and another near 17°C, though differences highlight the fact that the deeper transports are not well constrained by these data. Transports across PX2 can be matched to those at IX1 by increasing the MJ transport via the addition of a near-bottom flow as suggested above. Such an adjustment does not greatly enhance the match in the temperature/transport distribution between the two lines (Fig. 5b), but as we shall see below, a close match might not be expected.
Molcard et al. (2001) found that the transport measured by moorings through the major outflow straits (Lombok, Ombai, and Timor) during different years summed to 11.4 ± 3 Sv, which compares well with the 1996–97 full transport average reported by Hautala et al. (2001) of 8.4 ± 3.4 Sv. The Arlindo moorings deployed in Makassar Strait delivered an 18-month record of the velocity and temperatures in the thermocline and below, but the near-surface flow was not measured well. Hence the resulting transport and its associated temperatures have a large range due to uncertainty of the flow above 100 m. Total transport estimates from the Makassar moorings range from 6 to 12 Sv, which easily accomodates the IX1 result (Gordon et al. 1999; Vranes et al. 2002; Wajsowicz et al. 2003; Susanto and Gordon 2005).
Vranes et al. (2002) use several surface extrapolation schemes to estimate transports and temperature fluxes through the Makassar Strait that we will use below. We express the ITF heat and freshwater fluxes as transport-weighted temperature and salinity (Table 2). Vranes et al. (2002, hereafter VGF) found a TWT in Makassar Strait of between 11° and 16°C, depending on how velocity was extrapolated to the surface. The TWT from the XBT lines ranges from 20.7°C at PX2 to 21.6°C at IX1 farther downstream, substantially warmer than those suggested for Makassar. If we match the IX1 volume transport at PX2 by adding a barotropic flow under the MJ, the TWT at PX2 is 1° cooler but still far from agreeing with VGF (Fig. 6a). Can these estimates be reconciled?
First, we consider the impact of interannual variability on the Arlindo measurements that were collected during the end of a strong La Niña and development of the intense 1997 El Niño and Indian Ocean Dipole event. Analyzing the XBT data for the time period of the Arlindo measurements, we find that while transports are close to their average values, the transport-weighted temperature at PX2 during Arlindo is 1.4°C cooler than on average, while that at IX1 is close to average. Thus, the differences in volume and heat transports between the XBT means and Arlindo results cannot all be accounted for by interannual variability.
It is certain that the XBT transport estimates are biased low and warm because of neglecting the deep portion of the ITF. Property fields (Fieux et al. 1994; Wijffels et al. 2002; Gordon and Fine 1996; Talley and Sprintall 2005) suggest that the ITF comprises both a thermocline and intermediate water core. Inspired by the Vranes et al. approach, we attempt to bound the effect of a missing deep portion of the ITF using three simple velocity models to extend the XBT-based shear to the deepest ITF sill depth of 1250 m. The three models of velocity are 1) that velocity at 750 m is unknown (a reference-level velocity) but that it is constant below 750 m; 2) that velocity at 750 m is unknown (a reference-level velocity) but that it linearly decreases to zero at the sill; and 3) that velocity at 750 m is zero but that it linearly increases to some unknown value at the sill (reference velocity). Each scheme results in a different relationship between the unknown reference velocity, the total volume flux, and its associated TWT at IX1. By varying the unknown reference velocity a continuous curve is produced in the volume transport and TWT plane for each deep shear model (Fig. 6a). Surprisingly, our curves do not intersect with VGF’s values—VGF’s warmest scenario at Makassar Strait still does not intersect our coldest one at IX1. This suggests that the “missing deep” component in the XBT estimates cannot account for the different estimates produced by VGF.
Surface heat fluxes also will modify the TWT of the ITF as it passes through the Indonesian region from Makassar Strait to IX1. We examined several estimates of the long-term average heat flux over the Flores/Banda Seas—all show a substantial net ocean uptake of between 20 and 50 W m−2, particularly over the Nusa Tengarra; one estimate is shown in Fig. 7. The effect of this warming heat flux on the TWT of the ITF is estimated here. First we define an area (Fig. 7) for which we assume that all of the incoming heat flux is absorbed by the Makassar ITF and exported to the Indian Ocean across IX1; that is, we assume that ocean heat divergence by the mean ITF is the dominant term balancing surface heat uptake. The area chosen includes half the shallow Java Sea and all of the Arafura Sea and Gulf of Carpentaria. We assume that lateral ocean heat fluxes are relatively small (but not zero) in the Java Sea because of damping of the flow by bottom friction, while the Torres Strait is shallow and reef choked. From this we can roughly predict the temperature of the inflow at Makassar Strait that is needed to match the TWT of the outflow at IX1 for a given heat flux estimate.
We find that the temperature changes driven by the local heat fluxes are substantial and for some heat flux estimates it can account for the all of the difference between the VGF estimates and those at IX1 (Fig. 6b) without requiring any additional deep flow at IX1. However, the transport estimates from both the XBT lines and the Makassar moorings both do not measure the intermediate portion of the ITF and are both likely underestimates. A prediction of the size of the ITF can be derived from mean wind stress observations using the island rule, which assumes a Sverdrup balance in the interior ocean basins and a local balance between alongshore winds and pressure gradients at coasts (Godfrey 1989). This island rule suggests transports of between 12 and 14 Sv using scatterometer-based wind estimates (Fig. 6b). Accounting for both a missing deep portion of the ITF and the local heat fluxes gives the best agreement with the VGF estimate of 12 Sv at 16°C. We stress, however, that a deep portion of the ITF passes east of Makassar Strait and must flow out across IX1. Hence, the VGF data do not preclude the higher volume flows (suggested by scatterometer datasets) and colder total TWTs.
The low TWTs of the ITF determined by VGF lead them to conclude that the ITF contributes minimally to heat flux divergence in the Indian Ocean north of 30°S, while recognizing that the ITF waters could clearly be releasing heat back to the atmosphere in the Aghulas Outflow south of 30°S. Our results suggest that, as well as the heat absorbed in the Pacific Ocean by the ITF, it must also carry the heat absorbed in the Indonesian Seas throughout to the Indian Ocean. It is also worth pointing out that as well as the heat loss over the Aghulas Outflow, significant ocean cooling occurs over the Leeuwin Current system and this heat is highly likely to have derived from the ITF (Domingues et al. 2007).
The difference between the temperature distribution of transport at PX2 versus IX1 (Fig. 5b) also hints that mid- and lower-thermocline waters warm between the sections moving transport from the temperature classes below 12°C to those above this isotherm. Such a conversion is possible given the strong tidal mixing that likely occurs in this region (Ffield and Gordon 1996; Ffield and Robertson 2005; Katsumata and Wijffels 2006). However, to be more quantitative about this conversion rate, the barotropic flow in the straits requires more accurate measurements, such as those to be taken during the INSTANT program.
5. Summary
The regional velocity structure and transports derived from the 20-yr JAFOOS XBT dataset show a remarkably clear, quantitative, and consistent picture of the ITF as it passes through the Indonesian Seas, underscoring the value of such long sustained sampling. The thermocline portion of the ITF passes through the Makassar Strait and forms the Makassar Jet at the western shelf break of the Flores Sea. This jet likely feeds the Lombok Throughflow, but most of it turns eastward to flow along the northern edge of the Nusa Tengarra—as the Nusa Tengarra Current—which then feeds the Ombai and Timor Throughflows. Salinity changes suggest that flow to the Timor Passage must occur via recirculation in the Banda Sea, which shows a vigorous eastward flow along its northern edge. Distinct vertical shear profiles for the Ombai and Timor Throughflows, captured independently on two different XBT lines, also suggest strong modification of the flow in the Banda Sea before exiting via Timor Passage, as it features a surface-intensified and warm profile compared to Ombai, which hosts a much colder throughflow. This difference in structure and heat content of the Timor Throughflow suggests that part of the Nusa Tengarra Current recirculates in the Banda Sea before exiting through Timor Passage. A crude regional heat budget indicates that the ITF can be warmed by several degrees by local air–sea fluxes between the Labani Channel and the IX1 XBT line. The thermocline waters warmed in the Banda and Arafura Seas might well form a source for the shallow surface-intensified component of the Timor outflow.
The distinctly different shear structure measured in the Ombai Strait and Timor Passage is very interesting. Since the ITF is essentially a western boundary current that will seek the westernmost route into the Indian Ocean, any mean flow through Ombai Strait above the depth of the Lombok Sill (∼300 m) must arise because the flow through Lombok is dynamically saturated (either by hydraulic control or friction), forcing the excess east to Ombai Strait. Similarly the existence of a Timor Throughflow above the Savu Sea sill depth (∼800 m) implies that the Ombai transport is also saturated. In light of this reasoning, the deeper flow signature in Ombai compared to Lombok makes sense, but the surface-intensified nature of the Timor Throughflow remains unexplained.
The total interbasin exchange is likely larger than the estimates based on the JAFOOS XBT lines (as they miss the barotropic component at 750 m and all the flow below) or the Labani Channel estimates of VGF and others. Using various simple models to bound the deep flow, and comparing these results with predictions from wind products using the island rule (Godfrey 1989), the total mean ITF likely lies between 9 and 14 Sv with transport-weighted temperatures of 13°–18°C on inflow at Labani, and 16°–21°C on outflow across IX1, with local heat fluxes accounting for the 2°–3°C warming on route. Warmer ITFs are not consistent with the XBT or mooring data collected so far. However, the ITF is still a major contributor to the thermocline of the southeast Indian Ocean, comprising the bulk of the nascent South Equatorial Current near 110°E.
Acknowledgments
We thank those who initiated, ran, and contributed to the Australian Ship-of-Opportunity network: Peter Jackson, Ann Gronell, Rick Bailey, Graeme Ball, Lisa Cowen, Marty Rutherford, the captains and crew of the many ships that have contributed; Royal Australian Navy and Australian Bureau of Meteorology. This work was funded by the CSIRO Wealth from Oceans Flagship and the Australian Greenhouse Office through the Australian Climate Change Science Program.
REFERENCES
Bailey, R., A. Gronell, H. Phillips, E. Tanner, and G. Meyers, 1994: Quality control cookbook for XBT data [expendable bathythermograph data]. Version 1.1. Marine Laboratories CSIRO Rep. 221, 84 pp.
CERSAT, 2002: CERSAT QUIKSCAT Scatterometer mean wind field products user manual, version: 1.0. Department of Oceanography from Space, IFREMER, Ref. C2-MUT-W-03-IF, 47 pp.
Cresswell, G. R., and J. R. Luick, 2001: Current measurements in the Maluku Sea. J. Geophys. Res., 106 , 13953–13958.
Cresswell, G. R., A. Frische, J. Peterson, and D. Quadfasel, 1993: Circulation in the Timor Sea. J. Geophys. Res., 98 , C8. 14379–14389.
da Silva, A. M., C. Young, and S. Levitus, 1994: Algorithms and Procedures. Vol. 1, Atlas of Surface Marine Data 1994, NOAA Atlas NESDIS 6, 83 pp.
Domingues, C. M., M. E. Maltrud, S. E. Wijffels, J. A. Church, and M. Tomczak, 2007: Simulated Lagrangian pathways between the Leeuwin Current System and the upper ocean circulation of the southeast Indian Ocean. Deep-Sea Res. II, 54 , 797–817.
Dunn, J. R., and K. R. Ridgway, 2002: Mapping ocean properties in regions of complex topography. Deep-Sea Res. I, 49 , 591–604.
Ffield, A., and A. L. Gordon, 1996: Tidal mixing signatures in the Indonesian Seas. J. Phys. Oceanogr., 26 , 1924–1937.
Ffield, A., and R. Robertson, 2005: Indonesian Seas: Finestructure variability. Oceanography, 18 , 108–111.
Fieux, M., C. Andrie, P. Delecluse, A. G. Ilahude, A. Kartavtseff, F. Mantisi, R. Molcard, and J. C. Swallow, 1994: Measurements within the Pacific-Indian Oceans throughflow region. Deep-Sea Res. I, 41 , 1091–1130.
Godfrey, J. S., 1989: A Sverdrup model of the depth-integrated flow for the World Ocean allowing for island circulations. Geophys. Astrophys. Fluid Dyn., 45 , 1–2. 89–112.
Godfrey, J. S., 1996: The effect of the Indonesian throughflow on ocean circulation and heat exchange with the atmosphere: A review. J. Geophys. Res., 101 , 12217–12237.
Godfrey, J. S., and J. V. Mansbridge, 2000: Ekman transports, tidal mixing, and the control of temperature structure in Australia’s northwest waters. J. Geophys. Res., 105 , C10. 24021–24044.
Gordon, A. L., and R. A. Fine, 1996: Pathways of water between the Pacific and Indian Oceans in the Indonesian Seas. Nature, 379 , 146–149.
Gordon, A. L., R. D. Susanto, and A. Ffield, 1999: Throughflow within Makassar Strait. Geophys. Res. Lett., 26 , 3325–3328.
Grist, J. P., and S. A. Josey, 2003: Inverse analysis adjustment of the SOC air–sea flux climatology using ocean heat transport constraints. J. Climate, 16 , 3274–3295.
Hautala, S., J. Sprintall, J. T. Potemra, J. C. Chong, W. Pandoe, N. Bray, and A. Ilahude, 2001: Velocity structure and transport of the Indonesian Throughflow in the major straits restricting flow into the Indian Ocean. J. Geophys. Res., 106 , 19527–19546.
Holland, P. W., and R. E. Welsch, 1977: Robust regression using iteratively reweighted least-squares. Commun. Stat. Theory Methods, A6 , 813–827.
Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77 , 437–471.
Katsumata, K., and S. Wijffels, 2006: Semidiurnal M2 internal tides in the Indo-Australian Basin. Geophys. Res. Lett., 33 .L17601, doi:10.1029/2006GL026861.
Levitus, S., 1982: Climatological Atlas of the World Ocean. NOAA Prof. Paper 13, 173 pp. and 17 microfiche.
Liu, Y., M. Feng, J. Church, and D. Wang, 2005: Effect of salinity on estimating geostrophic transport of the Indonesian Throughflow along the IX1 XBT section. J. Oceanogr., 61 , 795–801.
Luick, J. L., and G. R. Cresswell, 2001: Current measurements in the Maluku Sea. J. Geophys. Res., 106 , C7. 13953–13958.
Meyers, G., 1996: Variation of Indonesian throughflow and the El Nino-Southern Oscillation. J. Geophys. Res., 101 , 12255–12263.
Meyers, G., R. J. Bailey, and A. P. Worby, 1995: Geostrophic transport of Indonesian Throughflow. Deep-Sea Res. I, 42 , 1163–1174.
Molcard, R., M. Fieux, and A. Ilahude, 1996: The Indo-Pacific throughflow in the Timor Passage. J. Geophys. Res., 101 , 12411–12420.
Molcard, R., M. Fieux, and F. Syamsudin, 2001: The throughflow within Ombai Strait. Deep-Sea Res. I, 48 , 1237–1253.
Murray, S. P., and D. Arief, 1988: Throughflow into the Indian Ocean through the Lombok Strait, January 1985–January 1986. Nature, 333 , 444–447.
Oberhuber, J. M., 1988: An atlas based on the COADS data set: The budgets of heat, buoyancy and turbulent kinetic energy at the surface of the global ocean. Max Planck Institut für Meteorologie Rep. 15, 19 pp.
Pegion, P. J., M. A. Bourassa, D. M. Legler, and J. J. O’Brien, 2000: Objectively derived daily “winds” from satellite scatterometer data. Mon. Wea. Rev., 128 , 3150–3168.
Qu, T., and G. Meyers, 2005: Seasonal characteristics of circulation in the southeastern tropical Indian Ocean. J. Phys. Oceanogr., 35 , 255–267.
Quadfasel, D., and G. R. Cresswell, 1992: A note on the seasonal variability of the South Java Current. J. Geophys. Res., 97 , C3. 3685–3688.
Ridgway, K. R., J. R. Dunn, and J. L. Wilkin, 2002: Ocean interpolation by four-dimensional weighted least squares—Application to the waters around Australasia. J. Atmos. Oceanic Technol., 19 , 1357–1375.
Schneider, N., 1998: The Indonesian Throughflow and the global climate system. J. Climate, 11 , 676–689.
Sprintall, J., J. T. Potemra, S. L. Hautala, N. A. Bray, and W. W. Pandoe, 2003: Temperature and salinity variability in the exit passages of the Indonesian Throughflow. Deep-Sea Res. II, 50 , 2183–2204.
Sprintall, J., and Coauthors, 2004: INSTANT: A New International Array to Measure the Indonesian Throughflow. Eos Trans. Amer. Geophys. Union, 85 .369, doi:10.1029/2004EO390002.
Susanto, R. D., and A. L. Gordon, 2005: Velocity and transport of the Makassar Strait throughflow. J. Geophys. Res., 110 .C01005, doi:10.1029/2004JC002425.
Talley, L. D., and J. Sprintall, 2005: Deep expression of the Indonesian Throughflow: Indonesian Intermediate Water in the South Equatorial Current. J. Geophys. Res., 110 .C10009, doi:10.1029/2004JC002826.
Vranes, K., A. L. Gordon, and A. Ffield, 2002: The heat transport of the Indonesian Throughflow and implications for the Indian Ocean heat budget. Deep-Sea Res. II, 49 , 1391–1410.
Wajsowicz, R. C., and E. K. Schneider, 2001: The Indonesian throughflow’s effect on global climate determined from the COLA Coupled Climate System. J. Climate, 14 , 3029–3042.
Wajsowicz, R. C., A. L. Gordon, A. Ffield, and R. D. Susanto, 2003: Estimating transport in Makassar Strait. Deep-Sea Res. II, 50 , 2163–2181.
Wijffels, S., and G. Meyers, 2004: An intersection of oceanic waveguides: Variability in the Indonesian Throughflow region. J. Phys. Oceanogr., 34 , 1232–1253.
Wijffels, S., E. Firing, and H. L. Bryden, 1994: Direct observations of the Ekman balance at 10°N in the Pacific Ocean. J. Phys. Oceanogr., 24 , 1666–1679.
Wijffels, S., J. Sprintall, M. Fieux, and N. Bray, 2002: The JADE and WOCE I10/IR6 Throughflow sections in the southeast Indian Ocean. Part 1: Water mass distribution and variability. Deep-Sea Res. II, 49 , 1341–1362.
Yu, L., R. A. Weller, and B. Sun, 2004: Improving latent and sensible heat flux estimates for the Atlantic Ocean (1988–99) by a synthesis approach. J. Climate, 17 , 373–393.
Geography of the study region, with names of the islands, straits, and seas discussed in the text. The gray shaded region is between 0- and 100-m depths.
Citation: Journal of Physical Oceanography 38, 9; 10.1175/2008JPO3987.1
The depth-averaged cross-track geostrophic plus Ekman velocity along the repeat XBT lines. (a) Average in the upper 100 m. (b) Average from 100 to 750 m. Velocities are Ekman plus geostrophic relative to a 750-m reference level. Scale is shown in central Australia. The color of the arrows indicates the transport-weighted salinity of the flow; note the change in this scale from (a) to (b). Flow across the IX22 line has been rotated to zonal for clarity of presentation.
Citation: Journal of Physical Oceanography 38, 9; 10.1175/2008JPO3987.1
The mean geostrophic plus Ekman velocities across the repeat XBT lines (colors) and the mean salinity field (black contours): (a) IX1 line, (b) PX2 line, and (c) IX22 line. The zero velocity contour is marked as a solid white contour line, and thus pale orange through white and blue colors mark the flow toward the Indian Ocean (negative values).
Citation: Journal of Physical Oceanography 38, 9; 10.1175/2008JPO3987.1
Schematic of flow pathways and associated total transports from the 20-yr repeat XBT dataset where flows cross the XBT lines. Transports (Sv) include the geostrophic flow relative to 750 m or the bottom if shallower, and the Ekman flow estimated from the NCEP1 product. Dashed arrows indicate subsurface flows. The light shading indicates depths between 0 and 100 m. The dark shading is land. The Makassar Jet transport (indicated by an asterisk) is likely greatly underestimated because of an assumed bottom reference velocity of zero over the shallow Flores Sea slope (300–400-m depth).
Citation: Journal of Physical Oceanography 38, 9; 10.1175/2008JPO3987.1
Vertical profiles of the geostrophic and Ekman (line above 50 m) transport between Australia and Asia from the JAFOOS XBT lines. (a) Transport per unit depth on depth levels. (b) Transport in 1° potential temperature bins. Solid line is for the IX1 section and dashed line is for the PX2 line. In (a), the Ekman component is shown at the top, while in (b) it is combined with the geostrophic component. In (b), the dotted line is for the adjusted PX2 transports where a barotropic velocity has been added under the MJ to match the IX1 transport (see text).
Citation: Journal of Physical Oceanography 38, 9; 10.1175/2008JPO3987.1
(Continued) Indonesian Throughflow transport-weighted temperature as a function of total volume flux under various assumptions about the shear below 750 m (see text). (a) Stars are 0–750-m estimates from the XBT lines: IX1 (black filled), PX2 (white), and PX2 adjusted (gray); gray circles are estimates from the Makassar Strait moorings as reported by Vranes et al. (2002); continuous lines show total transport using three models of the deep shear (see legend and text). (b) As in (a), with X showing a prediction of properties at Makassar based on the IX1 observations and the effect of local surface heat fluxes from five climatologies: NCEP1 from Kalnay et al. (1996); Comprehensive Ocean–Atmosphere Data Set (COADS) from da Silva et al. (1994) and Oberhuber (1988); Woods Hole Oceanographic Institution (WHOI) objectively analyzed air–sea fluxes from Yu et al. (2004); National Oceanography Centre, Southampton (NOC1.1a) from Grist and Josey (2003); vertical shaded lines show the annual average transport plus and minus one standard deviation based on Godfrey’s (1989) island rule and three wind stress climatologies: NCEP1 (as above); and COAPS from Pegion et al. (2000) and CERSAT (2002).
Citation: Journal of Physical Oceanography 38, 9; 10.1175/2008JPO3987.1
Surface heat flux (W m−2) into the ocean over the Indonesian Seas as estimated by Grist and Josey (2003). White line and circles define the surface region between the Labani Channel and the IX1 XBT line assumed to warm the Indonesian Throughflow.
Citation: Journal of Physical Oceanography 38, 9; 10.1175/2008JPO3987.1
Transports of major currents measured along the three JAFOOS XBT lines analyzed. Grid points along the XBT line were designated to a current system and summed to produce the transports reported. Positive values are eastward/northward (toward the Pacific). Total transport refers to the sum of the Ekman and geostrophic (relative to 750 m) transports. Errors on the total transport are a sum of those derived formally from the temperature mapping (see text) combined with a 20% assumed uncertainty in the Ekman component.
Total transport and its geostrophic and Ekman components, across the two coast-to-coast JAFOOS XBT lines, with their standard error bars (see text) and associated heat and freshwater fluxes expressed as transport-weighted properties. The values in square brackets indicate the Ekman transports derived from the QuikSCAT wind product (see text).
