Abstract

Ocean boundary currents are poorly represented in existing coupled climate models, partly because of their insufficient resolution to resolve narrow jets. Therefore, there is limited confidence in the simulated response of boundary currents to climate change by climate models. To address this issue, the eddy-resolving Ocean Forecasting Australia Model (OFAM) was used, forced with bias-corrected output in the 2060s under the Special Report on Emissions Scenarios (SRES) A1B from the CSIRO Mark version 3.5 (Mk3.5) climate model, to provide downscaled regional ocean projections. CSIRO Mk3.5 captures a number of robust changes that are common to most climate models that are consistent with observed changes, including the weakening of the equatorial Pacific zonal wind stress and the strengthening of the wind stress curl in the Southern Pacific, important for driving the boundary currents around Australia.

The 1990s climate is downscaled using air–sea fluxes from the 40-yr European Centre for Medium-Range Weather Forecasts Re-Analysis (ERA-40). The current speed, seasonality, and volume transports of the Australian boundary currents show much greater fidelity to the observations in the downscaled model. Between the 1990s and the 2060s, the downscaling with the OFAM simulates a 15% reduction in the Leeuwin Current (LC) transport, a 20% decrease in the Indonesian Throughflow (ITF) transport, a 12% increase in the East Australian Current (EAC) core transport, and a 35% increase in the EAC extension. The projected changes by the downscaling model are consistent with observed trends over the past several decades and with changes in wind-driven circulation derived from Sverdrup dynamics. Although the direction of change projected from downscaling is usually in agreement with CSIRO Mk3.5, there are important regional details and differences that will impact the response of ecosystems to climate change.

1. Introduction

Ocean boundary currents are poorly represented in the current climate models that contribute to the Coupled Model Intercomparison Project phase 3 (CMIP3), an initiative by the World Climate Research Programme (WCRP). This representation is partly due to an insufficient horizontal resolution of about 1°–2° (about 100–200km) in the ocean component of climate models, too large to realistically simulate these narrow jets. As a result there is limited confidence in the structural changes in these boundary currents projected by climate models. However, the response of these currents to climate change may directly affect marine ecosystems and regional climate (e.g., Stock et al. 2011). Around Australia, both the eastern and western boundary currents flow poleward, bringing warm tropical water to the colder regions. The East Australian Current (EAC) is a relatively strong western boundary current with an annually averaged volume transport of 20–30Sv (1Sv ≡ 106m3s−1) (Mata et al. 2000; Ridgway and Dunn 2003). The Leeuwin Current (LC) is a narrow, weak eastern boundary current, with an annually averaged volume transport of ~3.4Sv at 32°S (e.g., Feng et al. 2003). These currents are responsible for maintaining the marine ecosystems along the east and west coasts of Australia, both as a result of higher temperatures and alongshore dispersal (e.g., Booth et al. 2007; Lenanton et al. 2009).

As both the EAC and LC are primarily driven by large-scale wind forcing that is resolved by coarse-resolution climate models, more realistic simulations of the currents in a future climate could be obtained with an eddy-resolving ocean general circulation model using climate model output as forcing. The present study uses the Ocean Forecasting Australia Model (OFAM; Oke et al. 2005) to downscale a future climate scenario from the Commonwealth Scientific and Industrial Research Organisation Mark version 3.5 (CSIRO Mk3.5) climate model (Gordon et al. 2002). Dynamical downscaling methods have been routinely applied to regional atmospheric circulation studies and hurricane simulations and projections (e.g., Emanuel 2006; Emanuel et al. 2008; Caldwell et al. 2009). However, there are relatively few studies of dynamical downscaling of ocean circulation for climate projections (e.g., Meier 2006; Auad et al. 2006; Ådlandsvik and Bentsen 2007; Ådlandsvik 2008). This study applies ocean dynamical downscaling to investigate the impact of climate change on the boundary currents in the Australian region. In section 2, we present the models and forcing configurations for the downscaling experiments for two periods, the 1990s and the 2060s. In section 3, we compare the downscaling results with the climate model projections and discuss the changes between the two periods. In section 4, we summarize and discuss the strengths and weaknesses of our ocean downscaling approach for climate change projections.

2. Methods

a. The models

CSIRO Mk3.5 is one of the contributing CMIP3 climate models that informed the Intergovernmental Panel on Climate Change Fourth Assessment Report (IPCC AR4; Solomon et al. 2007). In CSIRO Mk3.5, the ocean component is the Modular Ocean Model version 2 (MOM2) with a horizontal resolution of 1.875° × 0.933° (Gordon et al. 2002). The CSIRO Mk3.5 climate model is chosen in this study for the following reasons: 1) it is a model configured by CSIRO, so we have ready access to both the developers and the forcing files; 2) it captures the average climate signals globally (Reichler and Kim 2008) and in the Southern Hemisphere (Sen Gupta et al. 2009); and 3) it simulates the teleconnection between the Indian and Pacific Oceans (Cai 2006). This teleconnection is the principal driving mechanism for the Leeuwin Current.

In this study, the bias-corrected atmospheric output from CSIRO Mk3.5 is used to force the OFAM (Oke et al. 2008) to simulate future ocean circulations. This approach is motivated by the understanding that most climate models have biases when simulating present climate. OFAM is based on MOM4 (Griffies et al. 2005). While the domain is global, the resolution is enhanced to a 10-km resolution in the greater Australian region (72°S–16°N, 90°–180°E). Outside this domain, the horizontal resolution decreases to 0.9° across the Pacific and Indian basins and to 2° in the Atlantic Ocean. OFAM has 47 vertical levels, 35 of which are in the top 1000m and 20 in the top 200m with 10-m resolution. OFAM is capable of simulating the current systems in the Australian region realistically, including their seasonal cycles and volume transports (e.g., Schiller et al. 2008). A major advantage of using OFAM for marine downscaling is its global configuration; there is no need to nest it inside a lower-resolution ocean model or to apply open boundary conditions. The disadvantage is that it has higher computational costs compared to a regional-domain model.

b. Forcing configurations for downscaling

1) The 1990s forcing

To evaluate the downscaling approach and provide a reference state for interpreting results for the 2060s, a control experiment is run, forced by surface fluxes (heat, freshwater, and wind stress) derived from the 40-yr European Centre for Medium-Range Weather Forecasts Re-Analysis (ERA-40) (Uppala et al. 2005) in the 1990s. A repeat-year forcing dataset typical of the average climatological conditions at the end of the twentieth century is used to remove the complexity associated with interannual variability from external forcing. This was achieved by constructing a monthly climatology of air–sea fluxes based on the years 1993–2001 from ERA-40. A correction to the heat and freshwater fluxes is added. This is calculated from an existing OFAM spinup run for 1993–2001, forced by the ERA-40 fluxes with the sea surface temperature (SST) and sea surface salinity (SSS) weakly restored to observed values with a 30-day time scale. This ensures that the forcing applied does not cause the upper-ocean state to drift. In addition, to account for daily variability, ERA-40 daily anomalies from year 1995 are added (defined as the difference between daily values and monthly mean). The year 1995, which is neither an El Niño nor an La Niña year, is chosen as “normal,” to obtain daily variability independent of ENSO signals. To summarize, the forcing terms are

 
formula

The experiment with this repeat-year forcing is denoted “CTRL.” The initial condition is the December climatology from the OFAM spinup run from 1993 to 2001 (Table 1).

Table 1.

Summary of surface forcing and initial conditions in different experiments.

Summary of surface forcing and initial conditions in different experiments.
Summary of surface forcing and initial conditions in different experiments.

2) The 2060s forcing

In CMIP3 present-climate simulations, large regional differences exist between the simulated surface fluxes of heat, freshwater, and momentum, and the observed fluxes (e.g., Sen Gupta et al. 2009). Figure 1 shows the bias in zonal wind stress in four CMIP3 climate models compared with ERA-40. The existence of biases in the CMIP3 climate models requires some thought when using these fluxes as forcing for OFAM. We hence employ a bias-correction technique, commonly used for atmospheric downscaling, to account for the surface flux biases of the CSIRO Mk3.5 simulation. We take the difference of the CSIRO Mk3.5 surface fluxes between the decades of 2060s and 1990s and add it to present-day surface fluxes (used in the CTRL experiment) to produce bias-corrected surface fluxes for the 2060s. The CSIRO Mk3.5 change in surface fluxes (heat, freshwater, and momentum fluxes) are from the Special Report on Emissions Scenarios (SRES) A1B simulation in the 2060s and the Twentieth-Century Climate in Coupled Model (20C3M) simulation in the 1990s. A monthly climatology is computed from the monthly fields of CSIRO Mk.35 output in the 1990s and 2060s to generate a repeat-year forcing dataset similar to that in the CTRL experiment. This approach of generating the surface fluxes for the 2060s assumes that both the bias in the CSIRO Mk3.5 surface fluxes and the daily variability in the surface fluxes do not change between the 2060s and the 1990s.

Fig. 1.

Zonal wind stress biases in four CMIP3 climate models: (a) Geophysical Fluid Dynamics Laboratory Climate Model version 2.1 (GFDL CM2.1), (b) Hadley Centre Global Environmental Model version 1(HadGEM1), (c) CSIRO Mk3.5, and (d) Max Planck Institute (MPI) ECHAM5. Bias is calculated over 1981–2000 in Nm−2, using ERA-40 as a benchmark.

Fig. 1.

Zonal wind stress biases in four CMIP3 climate models: (a) Geophysical Fluid Dynamics Laboratory Climate Model version 2.1 (GFDL CM2.1), (b) Hadley Centre Global Environmental Model version 1(HadGEM1), (c) CSIRO Mk3.5, and (d) Max Planck Institute (MPI) ECHAM5. Bias is calculated over 1981–2000 in Nm−2, using ERA-40 as a benchmark.

The terms used in the 2060s forcing are

 
formula

The initial condition for the OFAM downscaling run for the 2060s is created by adding projected changes in December between the two decades in the CSIRO Mk3.5 ocean fields (temperature, salinity, and velocity), interpolated to the OFAM grid, to the initial condition used in the CTRL run. This experiment is denoted as “FUTR.” No further information from the CSIRO Mk3.5 ocean state is used in this experiment.

To deal with the mismatch near the ocean/land boundaries in the low-resolution CSIRO Mk3.5 fields and the high-resolution OFAM fields, the land values in CSIRO Mk3.5 are masked out first, then interpolated (and extrapolated near the land boundary) to the high-resolution OFAM grid for both the initial condition and forcing fields. This approach is based on the understanding that local winds near the coast are unimportant compared to the large-scale wind field in the Pacific in driving the EAC and LC (Ridgway and Dunn 2003; Feng et al. 2003).

A summary of the downscaling experiments is provided in Table 1. Note that both experiments are run with repeat-year forcing. While the CTRL run is 26yr long, the FUTR run is 16yr long. It is known that tropical and subtropical thermocline equilibrates to external forcing on a 10–20-yr time scale (Sarmiento 1983; Cox and Bryan 1984). However, computational constraint precludes long runs. We tested for the stability of our simulations and found little drift in the upper ocean after 5yr. In the following analyses, the last 10yr of simulation for each experiment are used to compute the annual and seasonal fields.

3. Results

a. Overview

In presenting our results, we focus on the similarities and differences between climate projections from the CSIRO Mk3.5 coupled climate model and ocean downscaling simulations from OFAM. Overall, the CSIRO Mk3.5 simulates relatively broad currents with little fine detail because of its coarse resolution (Fig. 2a). The CSIRO Mk3.5 time-mean circulation does not show a well-defined Leeuwin Current—no coherent flow down the Western Australia coast and around Cape Leeuwin—only weak broad southward flow in the south Indian Ocean (Fig. 2a). In comparison, the time-mean circulation in the 1990s simulated by OFAM (Fig. 2b) shows much stronger flows with finer structure. In particular, the Leeuwin Current can be seen clearly in OFAM as starting from the Northwest Cape (around 22°S), going down the Western Australia coast, turning around the Cape Leeuwin at about 35°S, and extending all the way to the Tasmania coast to form the South Australian Current (between the eastern Great Australian Bight and western Bass Strait) and the Zeehan Current (off western Tasmania). These features of the LC are consistent with observations (Ridgway and Condie 2004).

Fig. 2.

(top) Time-mean circulation in the 1990s from (a) CSIRO Mk3.5 and (b) OFAM and (bottom) examples of snapshots from (c) CSIRO Mk3.5 and (d) OFAM. Colors show the depth-averaged current speed over 0–200m in ms−1; color scale is the same in all panels and is shown in (d). Vectors in (a)–(c) show the direction and magnitude of the depth-averaged current; vector scales are shown. Vectors are omitted in (d) to help visualize the eddies and jets in the flow field. Boxes denote the domain shown in Figs. 68. Magenta lines in (a),(b) denote the sections used to calculate the ITF transport. Locations of Ombai and Lombok Straits that are also used in OFAM for ITF transport calculations are shown in Fig. 3a. Shaded gray areas show the land masks in the model.

Fig. 2.

(top) Time-mean circulation in the 1990s from (a) CSIRO Mk3.5 and (b) OFAM and (bottom) examples of snapshots from (c) CSIRO Mk3.5 and (d) OFAM. Colors show the depth-averaged current speed over 0–200m in ms−1; color scale is the same in all panels and is shown in (d). Vectors in (a)–(c) show the direction and magnitude of the depth-averaged current; vector scales are shown. Vectors are omitted in (d) to help visualize the eddies and jets in the flow field. Boxes denote the domain shown in Figs. 68. Magenta lines in (a),(b) denote the sections used to calculate the ITF transport. Locations of Ombai and Lombok Straits that are also used in OFAM for ITF transport calculations are shown in Fig. 3a. Shaded gray areas show the land masks in the model.

The difference in the upper-ocean circulation of the two models is even more pronounced in monthly snapshots (Figs. 2c,d). In CSIRO Mk3.5, the circulation is still smooth (Fig. 2c) and shows little difference to the time-mean circulation (Fig. 2a). By contrast, a snapshot of OFAM’s circulation is dominated by eddies and jets (Fig. 2d), features that are averaged out in the time-mean circulation (Fig. 2b).

The EAC is a western boundary current that flows poleward from the southern Coral Sea to the coast of northern New South Wales, and then separates from the coast between 32° and 34°S to form the eastward-flowing current along the Tasman Front and the EAC extension, a southward-flowing eddy field (Ridgway and Dunn 2003). CSIRO Mk3.5 simulates the EAC as a western boundary current; however, the current magnitude of the EAC is too weak (Figs. 2a,c). Note that the current speed shown in Fig. 2 is depth averaged over the top 200m. This depth-averaged current speed is proportional to the volume transport in the top 200m, so it can be used as a proxy as volume transport. Hereafter, we will use the words circulation and transport interchangeably when referring to depth-averaged current speed. Note that the banded structure in the EAC extension from OFAM (Fig. 2c) is a consequence of long-term averaging of warm-core eddies moving down the coast, which produce a pattern of southward flow near the coast (first band of high flow) and northward flow offshore (second band of high flow) separated from the EAC (Fig. 2d).

To aid in the comparison of the CSIRO Mk3.5 and OFAM simulations, we focus on three upper-ocean regions identified in Figs. 2a,b: the magenta line shows the section to compute the Indonesian Throughflow (ITF) transport, the blue box for the Leeuwin Current, and the red box for the EAC.

b. The ITF

In OFAM there are three exit straits of the ITF: the straits of Lombok, Ombai, and Timor, while CSIRO Mk3.5 only resolves outflow through Timor Passage, which is too wide (Fig. 3a). Thus, the strength of the ITF in OFAM is estimated as the combined transport through the three exit straits, following Schiller et al. (2008). The ITF strength in CSIRO Mk3.5 is approximated by the net zonal transport across a section at 115°E (Fig. 3b), similar to the approach by England and Huang (2005), where they calculated the ITF transport in an ocean reanalysis product. The CSIRO Mk3.5 time-mean zonal velocity along 115°E is mostly westward down to 1200-m depth, similar to but weaker than the OFAM zonal velocity along 115°E (Fig. 3d). OFAM-simulated zonal velocity at Timor Strait at 124.5°E is shown in Fig. 3c, which has predominantly westward flow with the greatest flow in the upper 200m.

Fig. 3.

(a) Land masks in CSIRO Mk3.5 (gray area) and OFAM (black area). Red lines denote the three locations where ITF outflow straits in OFAM were used to compute the ITF transport. Blue line is the section at 115°E used to calculate ITF transport for CSIRO Mk3.5. (b),(d) 1990s time-mean zonal velocity at 115°E from CSIRO Mk3.5 and OFAM. (c) 1990s time-mean zonal velocity at Timor Strait from OFAM (124.5°E). Colors in (b)–(d) are zonal velocity in ms−1: eastward is positive (red) and westward is negative (blue).

Fig. 3.

(a) Land masks in CSIRO Mk3.5 (gray area) and OFAM (black area). Red lines denote the three locations where ITF outflow straits in OFAM were used to compute the ITF transport. Blue line is the section at 115°E used to calculate ITF transport for CSIRO Mk3.5. (b),(d) 1990s time-mean zonal velocity at 115°E from CSIRO Mk3.5 and OFAM. (c) 1990s time-mean zonal velocity at Timor Strait from OFAM (124.5°E). Colors in (b)–(d) are zonal velocity in ms−1: eastward is positive (red) and westward is negative (blue).

Both CSIRO Mk3.5 and OFAM simulate a reduction in the ITF transport between the 2060s and 1990s (Table 2). In CSIRO Mk3.5, the ITF transport decreases from 14.5Sv in 1990s to 13.0Sv in the 2060s (Table 2). In comparison, the downscaled ITF transport declines by about 20%, from 9.6Sv in the 1990s (CTRL) to 7.7Sv in the 2060s (FUTR). The CTRL estimate of 9.6 ± 2.1Sv (standard deviation of the annual transport) is consistent with the estimate of 9.7 ± 4.4Sv (standard deviation of the daily transport) from an ocean reanalysis, the Bluelink ReAnalysis (BRAN), from October 1992 to June 2006 (Schiller et al. 2008). However, these values are lower than a recent calculation of 15Sv from the 3-yr International Nusantara Stratification and Transport (INSTANT) field observation from January 2004 to December 2006, but they agree with other pre-INSTANT observations reported in the literature (e.g., Sprintall et al. 2009; Gordon et al. 2010). The differences among the pre-INSTANT observations, INSTANT observations, CSIRO Mk3.5, and OFAM values could be due to many factors, such as interannual and decadal variability, different large-scale wind forcings in CSIRO Mk3.5 and OFAM, model resolution, model parameterizations, etc. However, it is beyond the scope of this paper to address this issue in more detail here.

Table 2.

Time-mean volume transports of the ITF, LC, and the EAC from CSIRO Mk3.5 and downscaling, and estimates from observations (OBS). Note the numbers after ± are standard deviations of annual volume transports. LC transport is averaged over a latitude band between 32° and 34°S, the EAC core transport over 28°–32°S, and the EAC extension transport over 38°–42°S. BRAN is from 1992 to 2006 (Oke et al. 2008; Schiller et al. 2008). INSTANT estimate is from January 2004 to December 2006 (Sprintall et al. 2009; Gordon et al. 2010).

Time-mean volume transports of the ITF, LC, and the EAC from CSIRO Mk3.5 and downscaling, and estimates from observations (OBS). Note the numbers after ± are standard deviations of annual volume transports. LC transport is averaged over a latitude band between 32° and 34°S, the EAC core transport over 28°–32°S, and the EAC extension transport over 38°–42°S. BRAN is from 1992 to 2006 (Oke et al. 2008; Schiller et al. 2008). INSTANT estimate is from January 2004 to December 2006 (Sprintall et al. 2009; Gordon et al. 2010).
Time-mean volume transports of the ITF, LC, and the EAC from CSIRO Mk3.5 and downscaling, and estimates from observations (OBS). Note the numbers after ± are standard deviations of annual volume transports. LC transport is averaged over a latitude band between 32° and 34°S, the EAC core transport over 28°–32°S, and the EAC extension transport over 38°–42°S. BRAN is from 1992 to 2006 (Oke et al. 2008; Schiller et al. 2008). INSTANT estimate is from January 2004 to December 2006 (Sprintall et al. 2009; Gordon et al. 2010).

The seasonal cycle of ITF transport from CSIRO Mk3.5 and OFAM, as well as individual transport from each outflow strait in OFAM, are shown in Fig. 4. OFAM exhibits a more pronounced seasonal cycle than CSIRO Mk3.5. There is also a suggestion from OFAM that there is some seasonality in the projected change, with the greatest decline in the months of January–April.

Fig. 4.

Seasonal cycle of ITF transport from (a) CSIRO Mk3.5 and (b)–(e) OFAM: 1990s (black) and 2060s (red). CSIRO Mk3.5 transport is from a section at 115°E, from 22° to 8.2°S, with westward transport positive. Note that scales of the transport on the plots are different.

Fig. 4.

Seasonal cycle of ITF transport from (a) CSIRO Mk3.5 and (b)–(e) OFAM: 1990s (black) and 2060s (red). CSIRO Mk3.5 transport is from a section at 115°E, from 22° to 8.2°S, with westward transport positive. Note that scales of the transport on the plots are different.

c. The LC

On the west coast, the Leeuwin Current is the prominent ocean current. However, probably because of its coarse resolution, CSIRO Mk3.5 lacks a clearly defined boundary current north of 30°S and has only a weak broad southward flow in the southern Indian Ocean next to the Western Australia coast (Fig. 5). In January, there is no southward-flowing boundary current in CSIRO Mk3.5; the flow is instead slightly northward. The differences in the CSIRO Mk3.5 upper-ocean circulation off the coast of Western Australia between the 2060s and the 1990s show little changes in summer (Fig. 5c) but a clear weakening in winter (Fig. 5f).

Fig. 5.

CSIRO Mk3.5 seasonal circulation off Western Australia from (a),(b) 1990s, (c),(d) 2060s, and (e),(f) differences between the two periods for (left) January and (right) July. Colors show the magnitude of currents (ms−1) depth averaged over the top 200m. Vectors show the direction of the depth-averaged current; vector scale shown in (b). Gray shading indicates model land areas; black lines indicate the actual coastline.

Fig. 5.

CSIRO Mk3.5 seasonal circulation off Western Australia from (a),(b) 1990s, (c),(d) 2060s, and (e),(f) differences between the two periods for (left) January and (right) July. Colors show the magnitude of currents (ms−1) depth averaged over the top 200m. Vectors show the direction of the depth-averaged current; vector scale shown in (b). Gray shading indicates model land areas; black lines indicate the actual coastline.

By contrast, OFAM shows a well-defined LC both in austral summer and winter, with a much stronger current in winter (Fig. 6), consistent with observed seasonality (Feng et al. 2003). A zoomed view of the LC extending from the coast to 110°E is shown in Fig. 7. The LC is much weaker in austral winter in the 2060s than in the 1990s (Fig. 7f), while the change in austral summer, when the current is much weaker, is small (Fig. 7c).

Fig. 6.

OFAM time-mean circulation off Western Australia from (a),(d) CTRL and (b),(e) FUTR. Colors show the magnitude of currents (depth averaged over 0–200m) in ms−1. Vector lengths correspond to velocity magnitude, scale shown in (a) Velocity difference in (c) January and (f) July between the two time slices of 2060s and 1990s. Magenta lines indicate 100- and 2000-m isobaths.

Fig. 6.

OFAM time-mean circulation off Western Australia from (a),(d) CTRL and (b),(e) FUTR. Colors show the magnitude of currents (depth averaged over 0–200m) in ms−1. Vector lengths correspond to velocity magnitude, scale shown in (a) Velocity difference in (c) January and (f) July between the two time slices of 2060s and 1990s. Magenta lines indicate 100- and 2000-m isobaths.

Fig. 7.

As in Fig. 6, but for a zoomed view.

Fig. 7.

As in Fig. 6, but for a zoomed view.

To provide a more quantitative analysis, we examine the LC strength at a latitude band between 32° and 34°S. Within this latitude band, a well-defined boundary current exists throughout the year in both models, and there exists an observational estimate of LC transport at 32°S (Feng et al. 2003). In CSIRO Mk3.5, we choose the longitude of 108°E as the western extent to estimate the CSIRO Mk3.5 LC transport from the southward flow in the top 200m. The CSIRO Mk3.5 LC transport (averaged over 32° and 34°S) is 1.7Sv in the 1990s and 1.3Sv in the 2060s, a reduction of about 20%. In OFAM, the LC transport is calculated from the southward flow in the top 200m from the coast to 114°E between 32° and 34°S. The OFAM annual mean LC transport is 2.2Sv in the 2060s from FUTR, about 15% weaker than 2.7Sv in the 1990s from CTRL (Table 2).

Both models show that the LC is stronger in winter and weaker in summer (Fig. 8), with the highest transport in June, consistent with observed seasonality (e.g., Feng et al. 2003). The biggest projected reduction in the LC transport in OFAM is during the season when the mean flow is at its maximum from April to July, while CSIRO Mk3.5 has the greatest reduction from June to October.

Fig. 8.

Seasonal cycle of the LC transport from (a) CSIRO Mk3.5 and (b) OFAM: 1990s (black) and 2060s (red). Southward transport is positive. CSIRO Mk3.5 transport is estimated from southward flow averaged over 32°–34°S (0–200m, coast–108°). LC transport from OFAM is calculated from southward flow (0–200m, coast–114°E) averaged over 32°–34°S.

Fig. 8.

Seasonal cycle of the LC transport from (a) CSIRO Mk3.5 and (b) OFAM: 1990s (black) and 2060s (red). Southward transport is positive. CSIRO Mk3.5 transport is estimated from southward flow averaged over 32°–34°S (0–200m, coast–108°). LC transport from OFAM is calculated from southward flow (0–200m, coast–114°E) averaged over 32°–34°S.

d. The EAC

With climate change, CSIRO Mk3.5 simulates a strengthening in the EAC extension region, which is in general agreement with most other CMIP3 climate models (Sen Gupta et al. 2009). However, CSIRO Mk3.5 simulates a slight weakening in the core of the EAC between 26° and 32°S (Fig. 9, left). By contrast, OFAM simulates a vigorously strengthening EAC system; both the core of the EAC and the EAC extension strengthen with climate change. The strengthening can be seen along the EAC main path between 24° and 33°S, where it separates from the coast, and also farther south in the EAC extension region (Fig. 9, right).

Fig. 9.

Time-mean circulation off the east coast of Australia from (left) CSIRO Mk3.5 and (right) OFAM in the (top) 1990s and (middle) 2060s, and (bottom) differences between the two decades. Colors show current speed depth averaged over 0–600m in cms−1. Vector length represents current speed averaged over 0–600m in cms−1; and the corresponding vector scales are shown in (a),(b),(d),(e). Note the scale in (f) is 10 times larger than in (c).

Fig. 9.

Time-mean circulation off the east coast of Australia from (left) CSIRO Mk3.5 and (right) OFAM in the (top) 1990s and (middle) 2060s, and (bottom) differences between the two decades. Colors show current speed depth averaged over 0–600m in cms−1. Vector length represents current speed averaged over 0–600m in cms−1; and the corresponding vector scales are shown in (a),(b),(d),(e). Note the scale in (f) is 10 times larger than in (c).

To quantify the difference in the EAC transport between CSIRO Mk3.5 and OFAM, we compute the EAC transport as depth-integrated southward flow from 148° to 157°E in the top 600m as a function of latitude (Fig. 10). The CSIRO Mk3.5 EAC transport varies with latitude smoothly and gradually, and peaks at about 29°S in both periods of the 1990s and the 2060s. The EAC transport from OFAM in both periods varies more drastically with latitude and peaks around 32°S, but it remains nearly constant from about 35° to 42°S. Both the EAC core and extension strengthen in OFAM, but there is a slight decrease in the recirculation region around 33°–36°S. Note that the absolute value of the EAC transport will depend on the choice of longitudinal extent and the vertical extent chosen. To obtain representative values in CSIRO Mk3.5 and OFAM for the core and extension of the EAC transport, we computed the transport over two latitude bands: 28°–32°S for the EAC core and 38°–42°S for the EAC extension. Between the 1990s and the 2060s, the OFAM EAC transport increases about 12% (from 34 to 38Sv) in its core region and 35% (from 18 to 25Sv) in its extension, while the CSIRO Mk3.5 EAC transport decreases about 6% (from 19 to 18Sv) in the core and increases about 40% (from 6 to 9Sv) in its extension (Table 2).

Fig. 10.

Annual-mean EAC transport along its path: CSIRO Mk3.5 (dashed) and OFAM (solid). Red line is 2060s; black line is 1990s. Transport is calculated as depth-integrated southward flow from 148° to 157°E, 0–600m.

Fig. 10.

Annual-mean EAC transport along its path: CSIRO Mk3.5 (dashed) and OFAM (solid). Red line is 2060s; black line is 1990s. Transport is calculated as depth-integrated southward flow from 148° to 157°E, 0–600m.

With climate change, CSIRO Mk3.5 simulates a slight weakening in the EAC transport along its main path (north of 32°S) between the two periods; however, OFAM simulates a strengthening in the EAC transport both north and south of the peak value at 32°S, with the development of a second maximum at 39°S as the EAC extension entrains more offshore water into its flow.

4. Summary and discussion

The present study seeks to quantify the response of the Australian boundary currents, the EAC and LC in particular, to climate change using ocean dynamical downscaling. The approach uses bias-corrected surface fluxes from climate change projections under the SRES A1B scenario by the CSIRO Mk3.5 climate model to force an eddy-resolving ocean model in the Australian region, the OFAM model. The EAC and LC are generally poorly represented by coarse-resolution climate models. However, they are primarily driven by large-scale wind fields, which are resolved by climate models. It is therefore feasible to simulate the changes in these boundary currents in a future climate by forcing a high-resolution ocean model with output from climate model projections. To this end we investigate the impact of climate change on the LC, EAC, and ITF by examining the difference in these currents between the 2060s and the 1990s.

The 1990s climate is downscaled by driving OFAM with the air–sea fluxes from ERA-40. OFAM is able to reproduce key features of the EAC and LC in the 1990s, such as their spatial structure, seasonality, and volume transports, which are poorly represented in CSIRO Mk3.5. To produce the downscaled climate in the 2060s, we apply a bias-correction technique, whereby the difference between the CSIRO Mk3.5 2060s and 1990s air–sea fluxes (momentum, heat, and freshwater fluxes) are added to the ERA-40 forcing used in the control experiment.

While it is possible to assess the response of ocean boundary currents to climate change from coupled climate models, important features, such as spatial structure, are missing in climate models because of the resolution. The downscaling captures finer-scale features and realistic volume transports of the boundary currents, thus providing additional information on the impact of climate change.

The downscaling projects a 15% decrease in the LC transport (between 32° and 34°S) between the 1990s and the 2060s. The weakening of the LC in the 2060s can be attributed to changes in the large-scale wind forcing in the equatorial Pacific simulated by CSIRO Mk3.5. A weakening of the tropical atmospheric circulation in response to global warming is a robust feature across an ensemble of 22 IPCC AR4/CMIP3 climate models, including CSIRO Mk3.5 (Vecchi and Soden 2007). CSIRO Mk3.5 zonal wind stress in the equatorial Pacific displays a weakening trend from the 1850s to 2100 under the SRES A1B scenario (Fig. 11), consistent with the sign of change observed in the twentieth century (Vecchi et al. 2006, their Fig. 4). The weakened zonal wind stress in the equatorial Pacific leads to weakened ITF and LC in both CSIRO Mk3.5 and the downscaling, as the LC is primarily forced by the meridional pressure gradient associated with the ITF (Godfrey and Ridgway 1985; McCreary et al. 1986). The projected weakening of the ITF and LC in the 2060s from the CSIRO Mk3.5 and ocean downscaling are similar, but CSIRO Mk3.5 does not resolve the coastal waveguide and has an unrealistically wide LC that is nonexistent in the summer. In the downscaling projection, the largest reduction in the LC occurs in austral winter, when the current is also the strongest. The weakening trend of the ITF and LC into the future is consistent with observations from the past several decades (Feng et al. 2004; Wainwright et al. 2008), although a recent study suggests the trend over the past 15yr has reversed sign with increasing transport, likely a result of decadal variability (Feng et al. 2010, 2011).

Fig. 11.

Equatorial Pacific zonal-mean zonal wind stress anomaly (Nm−2) from the CSIRO Mk3.5 20C3M experiment (1860–2000) and SRES A1B (2001–2100) simulations. Zonal wind stress anomaly is averaged over 5°S–5°N, 120°E–70°W. Cyan curve is the monthly mean, black line is the annual mean, and red line is the trend. Note that zonal wind stress in the equatorial Pacific is westward (negative in this figure), so a positive anomaly indicates a weakening of wind stress there. Monthly difference in the wind stress over the global domain from the two time slices (2060s and 1990s) is used to compute a 10-yr averaged monthly climatology in the future forcing used in the downscaling experiments.

Fig. 11.

Equatorial Pacific zonal-mean zonal wind stress anomaly (Nm−2) from the CSIRO Mk3.5 20C3M experiment (1860–2000) and SRES A1B (2001–2100) simulations. Zonal wind stress anomaly is averaged over 5°S–5°N, 120°E–70°W. Cyan curve is the monthly mean, black line is the annual mean, and red line is the trend. Note that zonal wind stress in the equatorial Pacific is westward (negative in this figure), so a positive anomaly indicates a weakening of wind stress there. Monthly difference in the wind stress over the global domain from the two time slices (2060s and 1990s) is used to compute a 10-yr averaged monthly climatology in the future forcing used in the downscaling experiments.

The EAC strengthens in the 2060s because of the strengthening and southward shift of the basinwide wind stress curl in the South Pacific simulated by CSIRO Mk3.5, which is also a robust feature in the majority of CMIP3 climate models (Cai et al. 2005; Sen Gupta et al. 2009). This strengthening is consistent with observations and modeling results over past decades (e.g., Cai 2006; Ridgway 2007; Ridgway et al. 2008; Hill et al. 2008) as a result of the increased wind stress curl in the South Pacific, as explained by Sverdrup dynamics (e.g., Roemmich et al. 2007; Hill et al. 2008). However, in CSIRO Mk3.5, the EAC core does not strengthen and only the EAC extension strengthens. The downscaling projects a consistent strengthening of both the EAC core and EAC extension, about a 10% increase in the EAC core and a 35% increase in the EAC extension.

To assess if the difference between the two decades of the 2060s and the 1990s is representative of long-term trends, we compute the CSIRO Mk3.5 EAC core transport (averaged over 28–32°S) and the EAC extension transport (averaged over 38–42°S) from 1980 to 2100 (Fig. 12). The change between the two decades used in this study is consistent with the CSIRO Mk3.5 long-term trend, which shows that the EAC core slightly weakens but the EAC extension strengthens. The changes in ITF and LC between the 1990s and the 2060s in CSIRO Mk3.5 are also consistent with the long-term trend in the CSIRO Mk3.5 simulations (not shown).

Fig. 12.

CSIRO Mk3.5 EAC transport from 1980 to 2100. (a) EAC core transport averaged over 28°–32°S; (b) EAC extension transport (averaged over 38°–42°S). CSIRO Mk3.5 transport is integrated from southward flows over 148° to 160°E, 0–600m. Cyan curve is the monthly mean, blue line is the annual mean, and red line is the trend.

Fig. 12.

CSIRO Mk3.5 EAC transport from 1980 to 2100. (a) EAC core transport averaged over 28°–32°S; (b) EAC extension transport (averaged over 38°–42°S). CSIRO Mk3.5 transport is integrated from southward flows over 148° to 160°E, 0–600m. Cyan curve is the monthly mean, blue line is the annual mean, and red line is the trend.

The present study focuses on hydrodynamic changes in the boundary currents, which provides an important first step for investigating the ecological impacts of climate change. The differences between the ocean downscaling and CSIRO Mk3.5 in the projected changes in the 2060s for both the LC and EAC will have important implications for marine biology, connectivity, and water mass formation (e.g., Poloczanska et al. 2007, 2008; Dietze et al. 2009; Stock et al. 2011).

One caveat of marine downscaling is that there is no feedback of the ocean state to the atmosphere. The use of an ocean-only model for downscaling neglects the potential feedback a change in the ocean state may have on the exchange of heat, water, and momentum between the atmosphere and ocean. In this study, the ocean climate change projection is based on the difference of climate projections between two decades from one climate model and one emission scenario. To assess the robustness of climate projection, different projections from different climate models under various scenarios could be used to force the ocean downscaling model. Further, the use of the climate anomaly approach [Eq. (1)] to reduce climate model biases assumes we could treat climate change independently of the ocean state of the climate model. The consequence of this assumption on the downscaled projections needs to be investigated. It is conceivable that downscaled atmospheric forcing could also alter ocean downscaling (e.g., Langlais et al. 2009) and should be investigated.

Finally, as in all climate model projections, the assessment of projections is limited by the lack of future data. An important next step is to assess the robustness of projections from ocean downscaling against past trends and to investigate the limitations discussed above on the climate projections.

Acknowledgments

The OFAM simulations have been carried out at the Australian National Computing Infrastructure (NCI) supercomputing facility. Initial test runs were performed at the iVEC supercomputing facility in Western Australia. The FERRET program was used for analysis and graphics (http://ferret.pmel.noaa.gov/Ferret/). The authors thank Graham Symonds, Stuart Godfrey, Jay McCreary, Tony Hirst, Wenju Cai, and Evan Weller for their helpful discussions; Russ Fiedler for his OFAM tips; Chris Hines, Stephen Leak, Stephen Phipps, Margaret Khan, and David Singleton for their supercomputing help; Gareth Williams for his help archiving model output data at the iVEC; Alf Uhlherr, Robert Bell, Ben Evans, Paul Tildesley, and Robert Mollard for their help with the supercomputing resources; and the FERRET user community for its useful tips. Constructive and insightful comments from two anonymous reviewers helped improve the paper significantly. This work is supported by the Western Australian Marine Science Institution (WAMSI).

REFERENCES

REFERENCES
Ådlandsvik
,
B.
,
2008
:
Marine downscaling of a future climate scenario for the North Sea
.
Tellus
,
60
,
451
458
.
Ådlandsvik
,
B.
, and
M.
Bentsen
,
2007
:
Downscaling a twentieth century global climate simulation to the North Sea
.
Ocean Dyn.
,
57
,
453
466
,
doi:10.1007/s10236-007-0125-2
.
Auad
,
G.
,
A. J.
Miller
, and
E.
Di Lorenzo
,
2006
:
Long-term forecast of oceanic conditions off California and their biological implications
.
J. Geophys. Res.
,
111
,
C09008
,
doi:10.1029/2005JC003219
.
Booth
,
D. J.
,
W. F.
Figueira
,
M. A.
Gregson
,
L.
Brown
, and
G.
Beretta
,
2007
:
Occurrence of tropical fishes in temperate southeastern Australia: Role of the East Australian Current
.
Estuarine Coastal Shelf Sci.
,
72
,
102
114
,
doi:10.1016/j.ecss.2006.10.003
.
Cai
,
W.
,
2006
:
Antarctic ozone depletion causes an intensification of the Southern Ocean super-gyre circulation
.
Geophys. Res. Lett.
,
33
,
L03712
,
doi:10.1029/2005GL024911
.
Cai
,
W.
,
G.
Shi
,
T.
Cowan
,
D.
Bi
, and
J.
Ribbe
,
2005
:
The response of the southern annular mode, the East Australian Current, and the southern mid-latitude ocean circulation to global warming
.
Geophys. Res. Lett.
,
32
,
L23706
,
doi:10.1029/2005GL024701
.
Caldwell
,
P.
,
H.-N.
Chin
,
D. C.
Bader
, and
G.
Bala
,
2009
:
Evaluation of a WRF dynamical downscaling simulation over California
.
Climatic Change
,
95
,
499
521
.
Chiswell
,
S. M.
,
J.
Toole
, and
J.
Church
,
1997
:
Transports across the Tasman Sea from WOCE repeat sections: The East Australian Current 1990–94
.
N. Z. J. Mar. Freshwater Res.
,
31
,
469
475
.
Cox
,
M. D.
, and
K.
Bryan
,
1984
:
A numerical model of the ventilated thermocline
.
J. Phys. Oceanogr.
,
14
,
674
687
.
Dietze
,
H.
,
R.
Matear
, and
T.
Moore
,
2009
:
Nutrient supply to anticyclonic meso-scale eddies off Western Australia estimated with artificial tracers released in a circulation model
.
Deep-Sea Res.
,
56
,
1440
1448
.
Emanuel
,
K.
,
2006
:
Climate and tropical cyclone activity: A new model downscaling approach
.
J. Climate
,
19
,
4797
4802
.
Emanuel
,
K.
,
R.
Sundararajan
, and
J.
Williams
,
2008
:
Hurricanes and global warming: Results from downscaling IPCC AR4 simulations
.
Bull. Amer. Meteor. Soc.
,
89
,
347
367
.
England
,
M. H.
, and
F.
Huang
,
2005
:
On the interannual variability of the Indonesian Throughflow and its linkage with ENSO
.
J. Climate
,
18
,
1435
1444
.
Feng
,
M.
,
G.
Meyers
,
A.
Pearce
, and
S.
Wijffels
,
2003
:
Annual and interannual variations of the Leeuwin Current at 32°S
.
J. Geophys. Res.
,
108
,
3355
,
doi:10.1029/2002JC001763
.
Feng
,
M.
,
Y.
Li
, and
G.
Meyers
,
2004
:
Multidecadal variations of Fremantle sea level: Footprint of climate variability in the tropical Pacific
.
Geophys. Res. Lett.
,
31
,
L16302
,
doi:10.1029/2004GL019947
.
Feng
,
M.
,
M. J.
McPhaden
, and
T.
Lee
,
2010
:
Decadal variability of the Pacific subtropical cells and their influence on the southeast Indian Ocean
.
Geophys. Res. Lett.
,
37
,
L09606
,
doi:10.1029/2010GL042796
.
Feng
,
M.
,
C.
Böning
,
A.
Biastoch
,
E.
Behrens
,
E.
Weller
, and
Y.
Masumoto
,
2011
:
The reversal of the multi-decadal trends of the equatorial Pacific easterly winds, and the Indonesian Throughflow and Leeuwin Current transports
.
Geophys. Res. Lett.
,
38
,
L11604
,
doi:10.1029/2011GL047291
.
Godfrey
,
J. S.
, and
K. R.
Ridgway
,
1985
:
The large-scale environment of the poleward-flowing Leeuwin Current, Western Australia: Longshore steric height gradients, wind stresses, and geostrophic flow
.
J. Phys. Oceanogr.
,
15
,
481
495
.
Gordon
,
A. L.
, and
Coauthors
,
2010
:
The Indonesian Throughflow during 2004-2006 as observed by the INSTANT program
.
Dyn. Atmos. Oceans
,
50
,
115
128
.
Gordon
,
H. B.
, and
Coauthors
,
2002
:
The CSIRO Mk3 Climate System Model
.
CSIRO Tech. Rep. 50, 130 pp
.
Griffies
,
S. M.
,
M. J.
Harrison
,
R. C.
Pacanowski
, and
A.
Rosati
,
2003
:
A technical guide to MOM4
.
GFDL Ocean Group Tech. Rep. 5, 337 pp
.
Griffies
,
S. M.
, and
Coauthors
,
2005
:
Formulation of an ocean model for global climate simulations
.
Ocean Sci.
,
1
,
45
79
.
Hill
,
K. L.
,
S. R.
Rintoul
,
R.
Coleman
, and
K. R.
Ridgway
,
2008
:
Wind forced low frequency variability of the East Australia Current
.
Geophys. Res. Lett.
,
35
,
L08602
,
doi:10.1029/2007GL032912
.
Langlais
,
C.
,
B.
Barnier
,
J. M.
Molines
,
P.
Fraunié
,
D.
Jacob
, and
S.
Kotlarski
,
2009
:
Evaluation of a dynamically downscaled atmospheric reanalyse in the prospect of forcing long term simulations of the ocean circulation in the Gulf of Lions
.
Ocean Modell.
,
30
,
270
286
.
Lenanton
,
R. C.
,
N.
Caputi
,
M.
Kangas
, and
M.
Craine
,
2009
:
The ongoing influence of the Leeuwin Current on economically important fish and invertebrates off temperate Western Australia—Has it changed?
J. Roy. Soc. West. Aust.
,
92
,
111
127
.
Mata
,
M.
,
M.
Tomczak
,
S.
Wijffels
, and
J.
Church
,
2000
:
East Australian Current volume transports at 30°S: Estimates from the World Ocean Circulation Experiment hydrographic sections PR11/P6 and the PCM3 current meter array
.
J. Geophys. Res.
,
105
(
C12
),
28509
28526
.
McCreary
,
J. P.
,
S. R.
Shetye
, and
P. K.
Kundu
,
1986
:
Thermohaline forcing of eastern boundary currents: With application to the circulation off the west coast of Australia
.
J. Mar. Res.
,
44
,
71
92
.
Meier
,
H. E. M.
,
2006
:
Baltic Sea climate in the late twenty-first century: A dynamical downscaling approach using two global models and two emission scenarios
.
Climate Dyn.
,
27
,
39
68
.
Oke
,
P. R.
,
A.
Schiller
,
G. A.
Griffin
, and
G. B.
Brassington
,
2005
:
Ensemble data assimilation for an eddy-resolving ocean model
.
Quart. J. Roy. Meteor. Soc.
,
131
,
3301
3311
.
Oke
,
P. R.
,
G. B.
Brassington
,
D. A.
Griffin
, and
A.
Schiller
,
2008
:
The Bluelink Ocean Data Assimilation System (BODAS)
.
Ocean Modell.
,
21
,
46
70
.
Poloczanska
,
E. S.
, and
Coauthors
,
2007
:
Climate change and Australian marine life
.
Oceanogr. Mar. Biol.: Annu. Rev.
,
45
,
407
478
.
Poloczanska
,
E. S.
,
S. J.
Hawkins
,
A. J.
Southward
, and
M. T.
Burrows
,
2008
:
Modelling the response of populations of competing species to climate change
.
Ecology
,
89
,
3138
3149
.
Reichler
,
T.
, and
J.
Kim
,
2008
:
How well do coupled models simulate today’s climate?
Bull. Amer. Meteor. Soc.
,
89
,
303
311
.
Ridgway
,
K. R.
,
2007
:
Long-term trend and decadal variability of the southward penetration of the East Australian Current
.
Geophys. Res. Lett.
,
34
,
L13613
,
doi:10.1029/2007GL030393
.
Ridgway
,
K. R.
, and
J. S.
Godfrey
,
1997
:
Seasonal cycle of the East Australian Current
.
J. Geophys. Res.
,
102
,
22921
22936
.
Ridgway
,
K. R.
, and
J. R.
Dunn
,
2003
:
Mesoscale structure of the mean East Australian Current system and its relationship with topography
.
Prog. Oceanogr.
,
56
,
189
222
.
Ridgway
,
K. R.
, and
S. A.
Condie
,
2004
:
The 5500-km-long boundary flow off Western and southern Australia
.
J. Geophys. Res.
,
109
,
C04017
,
doi:10.1029/2003JC001921
.
Ridgway
,
K. R.
,
R. C.
Coleman
,
R. J.
Bailey
, and
P.
Sutton
,
2008
:
Decadal variability of East Australian Current transport inferred from repeated high-density XBT transects, a CTD survey and satellite altimetry
.
J. Geophys. Res.
,
113
,
C08039
,
doi:10.1029/2007JC004664
.
Roemmich
,
D.
,
J.
Gilson
,
R.
Davis
,
P.
Sutton
,
S.
Wijffels
, and
S.
Riser
,
2007
:
Decadal spinup of the South Pacific subtropical gyre
.
J. Phys. Oceanogr.
,
37
,
162
173
.
Sarmiento
,
J. L.
,
1983
:
A tritium box model of the North Atlantic thermocline
.
J. Phys. Oceanogr.
,
13
,
1269
1274
.
Schiller
,
A.
,
P. R.
Oke
,
G.
Brassington
,
M.
Entel
,
R.
Fiedler
,
D. A.
Griffin
, and
J. V.
Mansbridge
,
2008
:
Eddy-resolving ocean circulation in the Asian–Australian region inferred from an ocean reanalysis effort
.
Prog. Oceanogr.
,
76
,
334
365
.
Sen Gupta
,
A.
,
A.
Santoso
,
A. S.
Taschetto
,
C. C.
Ummenhofer
,
J.
Trevena
, and
M. H.
England
,
2009
:
Projected changes to the Southern Hemisphere ocean and sea ice in the IPCC AR4 climate models
.
J. Climate
,
22
,
3047
3078
.
Solomon
,
S.
,
D.
Qin
,
M.
Manning
,
M.
Marquis
,
K.
Averyt
,
M. M. B.
Tignor
,
H. L.
Miller
Jr.
, and
Z.
Chen
, Eds.,
2007
:
Climate Change 2007: The Physical Science Basis
.
Cambridge University Press, 996 pp
.
Sprintall
,
J.
,
S. E.
Wijffels
,
R.
Molcard
, and
I.
Jaya
,
2009
:
Direct estimates of the Indonesian Throughflow entering the Indian Ocean: 2004–2006
.
J. Geophys. Res.
,
114
,
C07001
,
doi:10.1029/2008JC005257
.
Stock
,
C. A.
, and
Coauthors
,
2011
:
On the use of IPCC-class models to assess the impact of climate on living marine resources
.
Prog. Oceanogr.
,
88
,
1
27
,
doi:10.1016/j.pocean.2010.09.001
.
Uppala
,
S. M.
, and
Coauthors
,
2005
:
The ERA-40 Re-Analysis
.
Quart. J. Roy. Meteor. Soc.
,
131
,
2961
3012
,
doi:10.1256/qj.04.176
.
Vecchi
,
G. A.
, and
B. J.
Soden
,
2007
:
Global warming and the weakening of the tropical circulation
.
J. Climate
,
20
,
4316
4340
.
Vecchi
,
G. A.
,
B. J.
Soden
,
A. T.
Wittenberg
,
I. M.
Held
,
A.
Leetmaa
, and
M. J.
Harrison
,
2006
:
Weakening of tropical Pacific atmospheric circulation due to anthropogenic forcing
.
Nature
,
441
,
73
76
.
Wainwright
,
L.
,
G.
Meyers
,
S.
Wijffels
, and
L.
Pigot
,
2008
:
Change in the Indonesian Throughflow with the climatic shift of 1976/77
.
Geophys. Res. Lett.
,
35
,
L03604
,
doi:10.1029/2007GL031911
.