JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 100, NO. C9, PAGES 18,409-18,420, SEPTEMBER 15, 1995

Effects of Indo-Pacific throughflow on the uppertropical Pacific and Indian OceansM. A. VerschellCenter for Ocean-Atmospheric Prediction Studies, Florida State University, Tallahassee

J. C. KindleNaval Research Laboratory, Stennis Space Center, Mississippi

James J. O'BrienCenter for Ocean-Atmospheric Prediction Studies, Florida State University, Tallahassee

Abstract. The dynamics and influence of the connection between the Pacific andIndian Oceans in the Indonesian Archipelago (PACIO) on the upper tropical Pacificand Indian Oceans are investigated using a reduced gravity global model. Twodifferent model geometries are used; the first has an open PACIO region (opencase) that permits normal transport, the second has a closed PACIO region (closedcase) that allows no transport. The model is forced by either monthly 1980-1989European Centre for Medium-Range Weather Forecasting wind stresses or by animposed Rossby wave pulse. The average upper layer thickness of the IndianOcean is significantly greater and more variable because of the influence of PACIOtransport. The effect in the Pacific Ocean is much smaller, and there appearsto be no appreciable effect of PACIO transport on interannual variability in theeastern tropical Pacific Ocean. RMS differences between the open and closed casesshow areas of low-frequency (> ^ year) variability over most of the southern IndianOcean, the western boundary of the Indian Ocean, the region of the Mindanao Eddy,and the eastern equatorial Pacific waveguide. Areas of significant high-frequencyvariability are the East African Coastal Current, the Mindanao Eddy region, theCoral Sea Basin, and in the eastern equatorial Pacific waveguide. The large effecton the southern Indian Ocean is linked to off-equatorial Rossby waves generated bytransport through the PACIO region.

Introduction steric height in the Indian Ocean determined that nettransport from the Pacific Ocean to the Indian Ocean

The connection between the Pacific and Indian Oceans through the PACIO region could have a large effect onin the Indonesian Archipelago (PACIO region) is com- the Sverdrup circulation pattern in the southern In-posed of three major passages (Figure 1). These pas- dian Ocean, which would otherwise be more like thatsages are Lombok Strait to the south of Java, Savu of the other tropical oceans. They hypothesized thatStrait to the north of Timor Island, and Timor Strait throughflow of Pacific water is also an important factorbetween Timor Island and Australia. The minor pas- in the Leeuwin current, the East Australian flow, andsages are both narrow and shallow, and no significant the South Equatorial current and that the throughflowtransport is assumed to pass through them. There have may be responsible for the observed lack of upwellingbeen few long-term observational studies in the region, (or lack of cold, nutrient-rich water) along the West-and most determinations of net transport and regional ern Australian coast. Godfrey [1989] calculated annualdynamics have been made from indirect observations, mean depth-integrated steric height and stream func-theoretical studies, and numerical modeling (Table 1). tion and determined closing off the PACIO region could

Godfrey and Golding [1981] performed an early study cause a 6°C drop in the upper 500 m over the entirethat investigated the dynamic importance of the PA- South Indian Ocean, with smaller changes elsewhere.CIO region. Their calculations of depth-integrated Other studies have looked at the effect of the trans-

port through the PACIO region on Indian Ocean cir-culation. Kundu and McCreary [1986] hypothesizedthat southward bending of the throughflow may con-

Copyright 1995 by the America* Geophysical Union. tribute to ̂ Leeuwin ̂ ^ although it would not

Paper number 95JC02075. be the dominant mechanism. However, Godfrey andoi48-0227/95/95JC-02075$05.0o Weaver [1991] determined that warm west Pacific wa-

18,409

18,410 VERSCHELL ET AL.: EFFECTS OF INDO-PACIFIC THROUGHFLOW

Figure 1. Pacific and Indian Oceans in the IndonesianArchipelago (PACIO region) and the three major pas-sages (Lombok, Savu, and Timor Straits). Light grayshading indicates ocean depth <200 m. Overlaid is the10-year average current pattern as determined by thenumerical model.

ter can flow through the Indonesian Archipelago, creat-ing basin-scale, buoyancy-driven circulation in the In-dian Ocean, and that Pacific heating and winds maybe the main driving force behind the Leeuwin current.Hirst and Godfrey [1993] examined the effects of thethroughflow on surface heat flux and sea surface tem-perature using a coarse resolution general circulationmodel (GCM). They found, in general, that throughflowwarms the Indian Ocean and cools the Pacific Ocean,but any large changes in surface temperatures or heatflux were confined to certain "well-defined" regions inthese basins. Wajsowicz [1994a] described numerical

model results that suggest PACIO transport is an essen-tial element for a good simulation of the eastern IndianOcean.

The question of how to balance mass, heat, and saltfluxes between the Indian and Pacific Oceans was ad-dressed by several studies. Wunsch et al [1983] con-cluded as a result of inverse calculations on sectionsacross the South Pacific Ocean that there was negligiblenet transport through the PACIO region, implying thatPACIO throughflow was unimportant in mass, heat,and salt flux balances. Other researchers have con-cluded that the PACIO region is important in balancingthese fluxes. Piola and Gordon [1984], Took [1987], andKindle et al. [1989] all maintain that an accurate deter-mination of the mass, heat, and salt fluxes in the Pacificand Indian Oceans requires adequate knowledge of themagnitude and variability of the PACIO throughflow.

The PACIO region has been suggested as the ma-jor route for return flow of thermocline water from thePacific Ocean to the Atlantic Ocean [Gordon and Piola,1983; Gordon, 1985, 1986]. Broecker [1991] believes thatthe PACIO region accounts for only about one quarterof this return flow but suggests that this flow is still animportant component of what he calls "the great oceanconveyor," the major pathway of global circulation.

Attention has also turned to this region because ofits possible connection to the El Nine-Southern Os-cillation (ENSO). Bye and Gordon [1982] claim thatbasin-wide exchange between the Pacific and IndianOceans through the Indonesian Archipelago is an im-portant factor in the ocean's role in the Southern Os-cillation. Nicholls [1984] determined that anomalouslywarm sea surface temperatures in the Indonesian Seaprecede ENSO warm events by a few months, and Whiteet al. [1985] found a large increase in the heat contentof the warm pool in the tropical western Pacific Oceanbefore the 1982-1983 ENSO that might be importantto the onset of this event. Kindle et al. [1987, 1989]found anomalously high Pacific to Indian Ocean trans-ports following a modeled ENSO warm event. They be-

Table 1. Previous Estimates of PACIO Throughflow

Study

Godfrey and Golding [1981]Piola and Gordon [1984]Fine [1985]Kindle et al. [1987]Murray and Arief[198S]Godfrey [1989]Broecker [1991]Clarke [1991]Inoue and Welsh [1993]Wajsowicz [1993]Cresswell et al. [1993]

Fieux et al. [1994]Molcard et al [1994]Meyers et al. [1994]

Estimate,Sv

10.014.05.04.51.7

16.05.06.19.8

14.07.0

18.64.55.2

Notes

from depth-integrated steric heightfreshwater box modeltritium box modelglobal reduced-gravity numerical modelcurrent meters in Lombok Strait onlyglobal Sverdrup modelresidence time of radiocarbon datageostrophic estimate from sea level datatwo-basin reduced-gravity numerical modelGodfrey [1989], with added frictionacoustic Doppler current profiler and drop-

sonde, Oct. 1987, March 1988geostrophic estimate from Aug. 1989 cruisecurrent meters in Savu and Timor Straitsgeostrophic estimate from expendable bathy-

thermograph data

VERSCHELL ET AL.: EFFECTS OF INDO-PACIFIC THROUGHFLOW 18,411

lieve that an important factor in the maintenance andvariability of the warm pool of the tropical western Pa-cific Ocean may be the extent to which the through-flow modifies the regional circulation. Therefore thethroughflow may have a significant effect on the inter-annual variability of the coupled ocean-atmosphere sys-tem in that region. Wajsowicz [1994b], using Godfrey'sisland rule, showed that depth-integrated PACIO trans-port decreased by 4 Sv (1 Sv = 106 m3 s"1) near theonset of an ENSO warm event and increased a similaramount near the onset of a cold event. Clarke and Liu[1994] also made a similar connection to PACIO trans-port and ENSO.

Several researchers have examined the reflection ofRossby wave energy by the irregular western bound-ary of the Pacific Ocean and the possible connection toPACIO throughflow. Clarke [1991] and du Penhoat andCane [1991] concluded that the reflection of lower-modeRossby waves from a realistic western Pacific boundarywould not be significantly different from the reflectionfrom a meridional wall. McCalpin [1987] predicted thatunlike reflection from a meridional wall, as the period ofthe wave decreased past the minimum reflection pointnear 65 days, reflection would increase. Du Penhoat andCane [1991] also determined that there is no anomaloustransport through the PACIO region associated witheither Rossby wave reflection or ENSO.

Recent indications that the PACIO region may beimportant in general ocean circulation, particularly inlarge-scale phenomena such as ENSO, make this animportant area to study. We examine the transportthrough the PACIO region and its influence on the dy-namics of the upper tropical Pacific and Indian Oceans.This is done using a 1—layer reduced gravity oceanmodel with two different model geometries as follows:global domain with an open PACIO region and globaldomain with a closed PACIO region, i.e., no Pacific-Indian throughflow permitted. We force the model withrealistic wind stresses in some experiments, while inother experiments the model is initialized with analyt-ically determined fields.

In the first section we describe the model and modelparameters used. Then we show the two different modelgeometries and discuss the various experiments that willbe run. The final sections detail the results of theseexperiments, along with a discussion of their physicalmeaning.

We will show that the model PACIO throughflowis mostly westward and has a mean transport of 7.5Sv, a value consistent with previous studies. This flowhas a significant effect on the seasonal and interannualvariability of the southern Indian Ocean, linked to off-equatorial Rossby waves generated by PACIO trans-port, while its influence on any large-scale Pacific dy-namics is much smaller. The mean upper layer thicknessof the Indian Ocean would be reduced by nearly 20 m ifthe PACIO region were to be closed off. Additionally,there are also several locations where the open PACIOregion effects the high-frequency (< | year) variabil-ity. These include the area off Mindanao Island and

the Coral Sea in the Pacific Ocean and on the westernboundary of the Indian Ocean, near the East AfricanCoastal Current (EACC).

The Model

The numerical model used in this study is the 1^-layer version of the Naval Research Laboratory (NRL)global multilayer model [ Wallcraft, 1991; Hurlburt et al,1995] which is based on the original semi-implicit, free-surface model of Hurlburt and Thompson [1980]. Themodel is formulated in a spherical coordinate system onan Arakawa C grid [Mesinger and Arakawa, 1976] andincludes realistic coastline geometry determined by the200 m depth contour. The model domain extends lat-itudinally from 72°S to 73° N and has a horizontal res-olution of 0.5° in latitude by 0.7° in longitude betweenlike variables. Northern and southern boundaries areclosed. Mass entrainment into the upper layer preventslower layer surfacing.

The version of the model used in the 10 year runs in-corporates thermodynamics by allowing horizontal den-sity gradients. The density at each point in the upperlayer can be affected by entrainment, detrainment, hor-izontal diffusion, advection, and a relaxation to the Lev-itus [1982] mean density climatology. Additional detailsof the thermodynamic model are given by Kindle andPhoebus [1995] and Shinoda and Lucas [1995].

The vertically integrated equations of motion for theupper layer in the thermodynamic version are

^- + (V -V + V- V)v + kxfV =ot-h \g'V(h -H) + hVg' - £- V

L 2Po— + AH V2V - max(0, -w)vPo

(1)

(2)

(3)

where V is upper layer transport, t is time, v is upperlayer velocity, k is the positive upward unit vector, /is the Coriolis parameter, h is upper layer thickness, g'is the reduced gravity constant, H is the upper layerthickness at rest, g is the acceleration due to gravity,po is the constant reference density, pi is upper layerdensity, rw is vector wind stress, AH is the coefficientof horizontal eddy viscosity, u is vertical mixing veloc-ity, <TP is the reference coefficient of density climatologyrelaxation, pi is upper layer density climatology, andKH is the coefficient of horizontal density diffusivity.

Newtonian friction has been applied to Lombok Straitto parameterize the effects of a passage narrower than

18,412 VERSCHELL ET AL.: EFFECTS OF INDO-PACIFIC THROUGHFLOW

allowed by the 0.7° longitudinal resolution of the modelgeometry. Without this additional friction the trans-port that would normally pass through Savu or TimorStraits would instead go though Lombok Strait. Thisadditional friction has been designed to produce a meantransport consistent with the measurements of Murrayet al [1990]. The resultant flow through Lombok Straitdisplays the observed seasonal and interannual varia-tions as measured by Murray et al [1990].

Experiments

Two different sets of experiments were performed inthis study. The first experiment involved the use of re-alistic winds to force the model after it was spun upwith climatological winds. The second experiment usedan analytically derived solution of a first-mode Rossbywave pulse imposed on an ocean at rest. Each experi-ment set was composed of two independent model runswhich were identical, with the exception that differentcoastline geometries were used to understand the effectsof PACIO throughflow. The model geometries differonly in whether PACIO throughflow is permitted (Fig-ure 2). All model runs used the same set of parameters(Table 2).

Model Forced by Realistic Winds

Each model run in this experiment set was spun upfrom rest for 26 years using the hydrodynamic versionof the model. The wind used for spin-up was a monthlyclimatology created from the 1980-1989 European Cen-tre for Medium-Range Weather Forecasting (ECMWF)global wind stress data set.

Table 2. Parameter Set Used for All Model Runs

Value

110°E 114'E 122°E 126°E 130°E 134°E

Figure 2. Model geometry for PACIO region showingthe three major passages (Lombok, Savu, and TimorStraits). Black shading indicates land and ocean depth< 200 m. Model grid is overlaid with vertices showingmodel-produced upper layer thickness (ULT) locations.Gray shading shows where the three passages are closedfor the closed experiment.

Eddy viscosity, m2 s 1

Density difFusivity, m2 s"1

Gravity, m s~2

Vertical mixing velocity, cm s"1

Thickness to start entrainment, mThickness to start detrainment, mLand-sea boundary depth, mMean active layer thickness, m

2000.05000.0

9.80.3

75.0650.0200.0250.0

Density values were calculated for each grid point us-ing the final upper layer thickness field and the Levitus[1982] mean density climatology. The resulting densityfield was used, along with the final output fields of thehydrodynamic model, to start up the thermodynamicmodel for an additional 26 years of spin-up. Final den-sity fields were recalculated as described above.

The final model runs were done with the thermody-namic version, forced by 1980-1989 ECMWF monthlymean wind stresses. The global fields of model-producedupper layer thickness (ULT), transports, and densitywere retained every 6 days for analysis. The resultingtwo data sets will'be referred to as the open exper-iment, the experiment with the realistic coastline ge-ometry, and the closed experiment, the one with themodified coastline geometry that closes off the threenear-equatorial passages between the Pacific and IndianOceans.

Model With Imposed Analytical Rossby Wave

The zonal and meridional velocity fields for a first-mode symmetric Rossby wave were calculated fromthe linear shallow-water equations following Matsuno[1966]. The ULT in the model was allowed to adjustto the imposed velocity fields, and the amplitude of theimposed Rossby wave pulse was empirically chosen toproduce an initial 20-m response in the ULT field.

All model runs were started at rest and run for 2years. The analytical velocity fields were imposed inthe model's Pacific basin at 99°W. The imposed Rossbywave pulse was exponentially ramped up for a \ day tothe first input time step and run for 20 days. The wavewas then exponentially ramped down for another \ day.An open boundary condition of the form developed byCamerkngo and O'Brien [1980] was applied just to theeast of the wave generation region to prevent eastwardpropagation which would otherwise lead to reflectionsfrom the eastern Pacific boundary.

Results

Model Forced by Realistic Winds

Ten-year time series of transports through Lombok,Savu, and Timor Straits, along with the combinedtransport through all three passages, were calculatedfrom the model run with open PACIO geometry (Fig-ures 3a-3d and Table 3). A strong annual signal is

VERSCHELL ET AL.: EFFECTS OF INDO-PACIFIC THROUGHFLOW 18,413

£ -6.0 -S -8.0- (o) Lombok St.

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

Figure 3. Transport through PACIO region for opencase (in Sverdrups), including (a) Lombok Strait, (b)Savu Strait, (c) Timor Strait, and (d) total PACIOtransport. Positive values show Indian to Pacific trans-port. Note that the transport scale for figure 3d isdifferent.

apparent, and an estimate of the autospectral densityfunction shows that about 50% of the variance is at the1-year period. Transport is mostly westward through-out the model run. During the 10-year duration of theexperiment, eastward flow occurred 4.9% (180 days) ofthe time. There are two significant episodes of east-ward transport, November 1983 and November 1987.The maximum eastward transport was 3.8 Sv duringthe November 1983 episode. The average transport overthe 10-year experiment was 7.5 Sv from the Pacific tothe Indian Ocean, with most of the transport valuesbeing between 2.8 and 12.3 Sv. Transport through the

three passages is highly correlated as expected. Thelargest mean transport is through Savu Strait, closelyfollowed by Timor Strait. Both Savu Strait and TimorStrait have significantly higher average transports thanLombok as dictated by the additional friction imposedon the flow through Lombok Strait.

An annual climatology was constructed from the mo-del output (Figures 4a-4d). Transport values are small-est during northern winter (October to January), withan average of 4.0 Sv, and largest during northern sum-mer (June to August), with an average of 11.4 Sv. Max-imum transports occur from mid-June to mid-August,while the transport minimum is in October to Novem-ber. Thus it takes roughly 5 months for the transitionfrom maximum to minimum transport and 7 months forthe return transition.

Using the above climatology, a 10-year time series ofPACIO transport anomalies was generated (Figures 5a-5d). Anomalously large westward transports occur inthe northern summer of 1984 and during 1988-1989,whereas anomalously eastward transports occur in thenorthern winters of 1983 and 1987. The entire periodfrom mid-1988 through 1989 had transport values largerthan normal. This differs from Wajsowicz [1993], whofound the magnitude of the transport predicted fromSouth Pacific wind stress curl to decrease in the late1980s. She attributes part of this to a switch from usingECMWF to Japan Meteorological Agency (JMA) windstresses in 1987 for her study.

There is some debate over the accuracy of estimatingvolume transport from midchannel velocity alone, es-pecially when the model resolution is such that there isonly one point in the channel. To check our transportestimates for the individual straits, we calculated thetransport time series for a cross section at 7.5°S andcompared these values with the sum of the transporttime series of the separate straits. The mean differ-ence was 0.12 Sv, and the RMS difference was 0.66 Sv.These results support the use of single-point, midchan-nel transport calculations in this model for qualitativestudies.

The focus of this study is to investigate what effectthe PACIO region has on the tropical Pacific and In-dian Oceans by closing off the passages between thetwo oceans. Time series of average ULT in the Indian

Table 3. Summary of Transport Statistics for the Open Experiment

PercentTotal

^Ta 0"Tb Transport

LombokSavuTimorPACIO

-2.0-2.8-2.6-7.5

1.02.02.04.8

273835

Maximum MaximumWest East

Transport Transport

4.47.59.1

19.0

1.82.01.23.8

PercentEast,

(Time)

3.87.17.54.9

PercentEast,

(Volume)

1.61.81.30.9

All transports are in sverdrups (1 Sv = 106 m3 s"1). PACIO refers to all threepassages together.

aValues indicate mean transport, with negative values indicating westward transports.bValues are the standard deviations of the transport.

18,414 VERSCHELL ET AL.: EFFECTS OF INDO-PACIFIC THROUGHFLOW

-2.0 -8. -3.0 :

(a) Lombok St.

0.0

-1.0 -

-2.0 -

-3.0 --4.0 --5.0 - (b) Savu St.

0.0

-1.0 -

-2.0 -

-3.0 =

(c) Timor St.

~4-° i-6.0 =

-8.0 --10.0 =-12.0

(d) Total

Figure 4. Annual climatology of transport throughPACIO region for open case (in Sverdrups), including(a) Lombok Strait, (b) Savu Strait, (c) Timor Strait,and (d) total PACIO transport. Positive values showIndian to Pacific transport. Note that the transportscale for figure 4d is different.

and Pacific Ocean basins between 20°S and 15°N werecalculated for both the open and closed experiments(Figure 6). Annual climatologies were also constructedfrom the 10-year average ULT series. While the PacificOcean average ULT differs only slightly between theclosed and open experiments, the Indian Ocean averageULT is significantly smaller in the closed experimentthan in the open experiment. The Indian Ocean av-erage ULT in the closed experiment is similar to thePacific Ocean average ULT in both experiments (Ta-ble 4).

ULT anomaly time series were constructed from theabove 10-year ULT time series by extracting the weeklyvarying climatology. There is no significant difference inthe Pacific Ocean for the open and closed cases; the twocurves nearly parallel each other, separated by a meanabsolute difference of 0.5 m (Figure 7). In the IndianOcean the difference between the ULT anomalies in theopen and closed experiments is much more pronounced(Figure 8). The mean absolute difference between thetwo cases is 1.0 m, and the curves only roughly followthe same path (Table 4).

The areas that are most affected by the artificialboundaries between the Indian and Pacific Oceans wereidentified by calculating the RMS 10-year ULT and up-per layer velocity magnitude (ULV) differences betweenthe closed and the open experiments. The boundaries

of the RMS fields are the same as used in calculatingthe average ULT.

The area of greatest ULT variability is the south-ern Indian Ocean, with the maximum extending west-ward along 15°S and originating in the PACIO region(Plate la). Other areas of prominent ULT variabilityare near the low-latitude western boundary currents ofthe Pacific and Indian Oceans; the Pacific Ocean ex-hibits a local maximum in the region extending east-ward from the Mindanao Current (130°E, 5 - 10°N),while the Indian Ocean exhibits a secondary maximumin the region of the EACC at 45°E, 8°S. A region ofsmaller ULT variability is in the eastern equatorial Pa-cific Ocean between 100 and 120°W. Finally, a regionof small ULT variability extends across the entire tropi-cal Pacific Ocean (10°N-10°S) and exhibits an eastwardintensification that extends poleward along the easternboundary.

The total ULT RMS difference field can be dividedinto fields of low-frequency variability and fields of high-frequency variability. The closed and open data setswere smoothed via two passes of a 180-day running av-erage, filtering out energy at frequencies less than halfa year. The low-frequency field is the RMS analysis ofthese filtered data sets (Plate Ib). The high-frequencyfield is found by subtracting the low-frequency fieldfrom the original difference field (Plate Ic).

M | I I I I I M I I I I | I I

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

"I"1""""!"1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

"I"1""""!"""1""!1

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

Figure 5. Transport anomalies through PACIO re-gion for open case (in Sverdrups), including (a) Lom-bok Strait, (b) Savu Strait, (c) Timor Strait, and (d)total PACIO transport. Positive values show anoma-lous eastward transport. Note that the transport scalefor figure 5d is different.

VERSCHELL ET AL.: EFFECTS OF INDO-PACIFIC THROUGHFLOW 18,415

Pocific (Open)

Pacific (Closed)

Indian (Open)

Indian (Closed)

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

Figure 6. Mean upper layer thickness (in meters) inthe tropical Pacific and Indian Oceans for open andclosed experiments.

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

Figure 7. Anomaly mean upper layer thickness (in me-ters) in the tropical Pacific Ocean for open and closedexperiments.

The low-frequency ULT RMS variability dominatesthe height field variations of the open-closed through-flow field, primarily because of energetic long Rossbywaves emanating from the PACIO region that cross thesouthern Indian Ocean (Plates la and Ib). The low-frequency variability also accounts for the broad bandof weak activity across the tropical Pacific Ocean be-tween 10°N and 10°S and contributes to the ULT RMSvariability in the Mindanao Eddy region. The high-frequency component of the open-closed ULT variabil-ity is concentrated primarily in three areas that, as dis-cussed below, are regions characterized by pronounceddynamic instabilities.

The variability in the ULV RMS difference field isdominated by the high-frequency variations in the west-ern tropical Indian Ocean, the western Pacific Oceannear Mindanao, and in the eastern equatorial PacificOcean (Plates 2a-2c). Observations of intraseasonalvariability in the EACC region has been discussed byMysak and Mertz [1984] and Schott et al. [1988]. Kindleand Thompson [1989], using a one-layer reduced grav-ity model, hypothesized that the offshore extension ofthe EACC is barotropically unstable and responsiblefor the 50-day oscillations observed in this region. Thehigh-frequency variability in the eastern tropical Pacific

Table 4. Summary of Average Upper Layer ThicknessStatistics

Open Closed DifFerencea

p>Pacific 131.2Indian 157.2

Pacific anomalyb

Indian anomalyb

<j

4.22.5

4.12.2

^

136.0138.4

(T fl

4.0 -4.82.0 18.8

3.91.7

(7

0.71.3

0.71.2

All values are in meters; /i is mean ULT and <r is stan-dard deviation of the ULT.a Difference values are calculated from the time seriesopen-closed.

bAnomaly ULT n is 0 by definition.

Ocean may be related to the type of instabilities dis-cussed by McCreary and Yu [1992]. Interestingly, whilethe southern Indian Ocean is an area of pronouncedULT variability, the ULV response in that same area isweak in comparison to that near the western boundary.This is due to the weak velocity response associatedwith the off-equatorial long Rossby waves relative tothe energetic mesoscale variability of the EACC.

It should be noted that some of the high-frequencyvariability in the open-closed fields cannot be attributedto the effects of the Indonesian boundary, but instead,to the lack of coherence between the open and closed ex-periments in the regions characterized by unstable flow.However, analysis does show that ULV variability is in-creased slightly in the Pacific Ocean when the through-flow is closed, whereas a noticeable increase in ULVvariability occurs in the western Indian Ocean when theIndonesian boundary is open. Given the simplicity andresolution of the simulation, it is beyond the scope ofthis report to explore the dynamics of the flow instabil-

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989

Figure 8. Anomaly mean upper layer thickness (inmeters) in the tropical Indian Ocean for open and closedexperiments.

18,416 VERSCHELL ET AL.: EFFECTS OF INDO-PACIFIC THROUGHFLOW

J————r——————————J——————~i~_______________I \

40°E 80°E 120°E 160°E 160°W 120°W 80°W

40°EI I I

80°E 120°E 160°E 160°W 120°W 80°W

40°E 80°E 60°WI T T

120°W 80°W

Plate 1. (a) Total RMS differences (open-closed), (b) low-frequency (periods > | year) RMSdifferences, and (c) high-frequency (periods < ^ year) RMS differences in upper layer thickness(in meters) over a 10-year period for the tropical Pacific and Indian Oceans. Colored contoursshow variability greater than 1 standard deviation above the mean.

ities, except to identify regions potentially affected bythe throughflow. Modeling efforts with greater verticaland horizontal resolution and bottom topography willbe needed to examine these features in detail as well asto explain the response of the Indonesian Archipelagocirculation to high-frequency forcing.

Model With Imposed Analytical Rossby Wave

Plates 3a-3c show the results of the imposed analyti-cal Rossby wave pulse on an otherwise at-rest ocean inthe open case. The wave pulse can be seen travelingacross the equatorial Pacific Ocean (Plate 3c), produc-ing the PACIO transport seen in Plate 3b. The wavesignal travels through the PACIO region, becoming apoleward traveling coastal Kelvin wave that separatesoff the west Australian coast to become an off-equatorialRossby wave in the Indian Ocean (Plate 3a).

RMS difference plots of open-closed ULT fields in theimposed analytical Rossby wave pulse experiment showlittle difference in the Pacific Ocean. Snapshots of ULTdifference (open-closed) suggest a small reflected Kelvinwave that is apparently present in the closed case only.

Discussion

The PACIO region is a dynamically important areafor understanding the Indian Ocean and, to a lesser ex-tent, the eddy-generating regions near Mindanao, theCoral Sea, and the southwestern Indian Ocean. Thesouthern Indian Ocean is significantly affected by trans-port through this region, whereas the much larger areaof the Pacific Ocean and its large ambient interannualsignal minimize the effect of the throughflow in thatbasin.

Upper layer transport through the PACIO region ismainly westward. Throughout most of the 10-year ex-periment the mean ULT is larger on the Pacific Oceanside of the PACIO region than on the Indian Oceanside. The associated pressure gradient between the twobasins (positive to the east) maintains this predomi-nately westward transport. Calculation of the autospec-tral density function (not shown) from the time seriesof PACIO transport shows that transport through thePACIO region is largely dominated by the annual sig-nal.

VERSCHELL ET AL.: EFFECTS OF INDO-PACIFIC THROUGHFLOW 18,417

I i i i I (a)

40°E 80°E 120°E 160°E 160°W 80°W

40°E 80°E 120°E 160°E 160°W 120°W

40°E 80°E 120°E 160°E 160°W 120°W 80°W

Plate 2. (a) Total RMS differences (open-closed), (b) low-frequency (periods > | year) RMSdifferences, and (c) high-frequency (periods < | year) RMS differences in upper layer velocitymagnitude (in meters second"1) over a 10-year period for the tropical Pacific and Indian Oceans.Colored contours show variability greater than 1 standard deviation above the mean.

There are two significant episodes of eastward trans-port, late 1983 and 1987. Examination of the mean ULTfor the Pacific Ocean and Indian Ocean sides of the PA-CIO region shows a reversal (Indian Ocean ULT largerthan Pacific Ocean ULT) for these episodes. These twoperiods occur in the normally low-flow northern winter,just after an ENSO warm event but before the sea levelin the western Pacific Ocean returns to its normallyhigh values. Equally important to the scenario is theoccurrence of high sea level on the Indian Ocean side ofthe PACIO region owing to westerly winds during thenorthwest monsoon.

The nearly continuous westward transport throughthe PACIO region deepens the mean ULT in the IndianOcean by nearly 14% as compared with the mean ULTfor the Indian Ocean with no PACIO throughflow. Thisthickening of the ULT may affect many of the dynamicproperties of the Indian Ocean, e.g., wave speeds, seasurface temperature, etc. The PACIO throughflow isalso linked to greater variability in Indian Ocean ULT,the standard deviation of which in the Indian Ocean isnearly double that of the Pacific Ocean (Table 4). Theeffects on the Pacific Ocean are much smaller, becauseof the large area relative to that of the Indian basin.

The PACIO transport that leads to the nearly 14% in-crease in mean ULT in the Indian Ocean causes only a3.5% decrease in mean ULT in the Pacific Ocean.

The calculations of ULT RMS differences show thatthe greatest effects of PACIO throughflow are mainlyin the southern Indian Ocean due to the "poleward"propagation of the PACIO signal along the northwest-ern Australian coast as an internal Kelvin wave and thesubsequent westward radiation of Rossby waves. TheRossby waves tend to leave the coast at approximately14°S, a result consistent with McCalpin [1995].

The large RMS difference in the southern IndianOcean confirms the importance of the transport of wa-ter from the Pacific Ocean through this region on IndianOcean dynamics. If we assume that upper ocean tem-perature is directly related to the ULT, then our modelresults showing a much shallower ULT in the IndianOcean support the hypothesis that closing off the PA-CIO region could cause a significant drop in the temper-ature of the upper ocean layer over the entire southernIndian Ocean [Godfrey, 1989].

The transport through the PACIO region also has adynamical effect on more localized regions in both thePacific and Indian Oceans. The areas of large high-

18,418 VERSCHELL ET AL.: EFFECTS OF INDO-PACIFIC THROUGHFLOW

Indian Ocean along 10S Equatorial Pacific Ocean,J_L..-l__....J._...̂ m. I........!.......1 J J ; I I I l I I I i I I I

(b)

T800E! 90°E 100°E1100E1200E130°E -1.0-0.8-0.6-0.4-0.2 0.0 HO°E 150°E 160°E 170°E 180° 170°W 160°W

209

208

207

206

205

204

203

202

201

200

199

198

l' 197

Layer Thickness (m) PACIO transport (Sv) Layer Thickness (m)

Plate 3. Results of analytical Rossby wave pulse experiment including (a) longitude versus timeplot of Indian Ocean ULT averaged over 11°S-9°S (in meters), (b) time series of total PACIOtransport (in Sverdrups), and (c) longitude versus time plot of Pacific Ocean ULT averaged over3°S-3°N (in meters).

frequency RMS differences for both ULT and ULV arelocations of considerable eddy generation. In the Pa-cific Ocean the areas of large high-frequency variabilityare found off Mindanao Island, the location of the Min-danao eddy, and the eastern equatorial Pacific waveg-uide. The differences here are slightly decreased bythe open boundary with the Indian Ocean, whereasthe throughflow increases the intraseasonal variabilityin the region of the EACC.

The results of the analytical Rossby wave experimenthelp to explain the small effect of the throughflow onthe Pacific Ocean variability. The analytical Rossbywave generated was chosen to represent a first-modesymmetrical Rossby wave, with its height field centerednear ±5° about the equator. When this Rossby wavearrives at the western Pacific boundary, some of theenergy is reflected as an equatorial Kelvin wave, someis transmitted into the Indian Ocean through the PA-CIO region, and the rest is lost to dissipation near thewestern Pacific boundary. In the open experiment any

reflected Kelvin wave is indistinguishable among back-ground noise. The amount of reflected energy is signifi-cantly smaller than that computed by McCalpin [1987],Clarke [1991], and du Penhoat and Cane [1991]. Thismay be the result of the more realistic western bound-ary and the presence of gaps in the PACIO region inour study.

Acknowledgments. The authors wish to thank JamiePringle and Joe Metzger for their invaluable assistance inrunning the NRL ocean model. Special thanks go to EdHarrison and Steve Hankin for the graphical data analy-sis program "ferret," from which all the figures were pro-duced. This work was supported in part by a grant of highperformance computer (HPC) time from the Department ofDefense HPC Shared Resource Center at the Naval Oceano-graphic Office. The work by M. A. Verschell was supportedby the NRL fellowship program at Florida State Universityand the DOD AASERT fellowship at Florida State Uni-versity. The work by J. C. Kindle was supported by the

VERSCHELL ET AL.: EFFECTS OF INDO-PACIFIC THROUGHFLOW 18,419

6.1 Dynamics of Low Latitude Western Boundary Currentproject and the 6.1 Dynamics of Coupled Air-Ocean Modelsstudy sponsored by the Office of Naval Research (programelement 61153N). The Center for Ocean-Atmospheric Pre-diction Studies receives its base support from the Secretaryof the Navy grant to James J. O'Brien.

References

Broecker, W. S., The great ocean conveyor, Oceanography,4, 79-89, 1991.

Bye, J. A. T., and A. H. Gordon, Speculated cause of inter-hemispheric ocean oscillation, Nature, 296, 52-54, 1982.

Camerlengo, A. L., and J. J. O'Brien, Open boundary con-ditions in rotating fluids, J. Comput. Phys., 35, 12-35,1980.

Clarke, A. J., On the reflection and transmission of low-frequency energy at the irregular western Pacific Oceanboundary, J. Geophys. Res., 96, 3289-3305, 1991.

Clarke, A. J., and X. Liu, Interannual sea level in the north-ern and eastern Indian Ocean, J. Phys. Oceanogr., 24,1224-1235, 1994.

Cresswell, G., A. Frische, J. Peterson, and D. Quadfasel, Cir-culation in the Timor Sea, J. Geophys. Res., 98, 14,379-14,389, 1993.

du Penhoat, Y., and M. A. Cane, Effect of low-latitude west-ern boundary gaps on the reflection of equatorial motions,J. Geophys. Res., 96, 3307-3322, 1991.

Fieux, M., C. Andrie, P. Delecluse, A. G. Ilahude, A. Kar-tavtseff, F. Mantisi, R. Molcard, and J. C. Swallow, Mea-surements within the Pacific-Indian Oceans throughflowregion, Deep Sea Res., 41, 1091-1130, 1994.

Fine, R. A., Direct evidence using tritium data for through-flow from the Pacific into the Indian Ocean, Nature, 315,478-480, 1985.

Godfrey, J. S., A Sverdrup model of the depth-integratedflow for the world ocean allowing for island circulations,Geophys. Astrophys. Fluid Dyn., 45, 89-112, 1989.

Godfrey, J. S., and T. J. Golding, The Sverdrup relationin the Indian Ocean, and the effect of Pacific-IndianOcean throughflow on Indian Ocean circulation and onthe East Australian Current, J. Phys. Oceanogr., 11, 771-779, 1981.

Godfrey, J. S., and A. J. Weaver, Is the Leeuwin Currentdriven by Pacific heating and winds?, Prog. Oceanogr.,27, 225-272, 1991.

Gordon, A. L., Indian-Atlantic transfer of thermocline wa-ter at the Agulhas Ret reflect ion, Science, 227, 1030-1033,1985.

Gordon, A. L., Interocean exchange of thermocline water,J. Geophys. Res., 91, 5037-5046, 1986.

Gordon, A. L., and A. R. Piola, Atlantic Ocean upper layersalinity budget, J. Phys. Oceanogr., 13, 1293-1300, 1983.

Hirst, A. C., and J. S. Godfrey, The role of Indonesianthroughflow in a global ocean GCM, J. Phys. Oceanogr.,23, 1057-1086, 1993.

Hurlburt, H. E., and J. D. Thompson, A numerical studyof Loop Current intrusions and eddy shedding, J. Phys.Oceanogr., 10, 1611-1651, 1980.

Hurlburt, H. E., A. J. Wallcraft, W. J. Schmitz, E. J. Met-zger, and P. J. Hogan, Dynamics of the Kuroshio/Oyashiocurrent system using eddy-resolving models of the NorthPacific, J. Geophys. Res., in press, 1995.

Inoue, M., and S. E. Welsh, Modeling seasonal variability inthe wind-driven upper-layer circulation in the Indo-Pacificregion, J. Phys. Oceanogr., 23, 1411-1436, 1993.

Kindle, J. C., and P. A. Phoebus, The ocean response tooperational westerly wind bursts during the 1991-1992 ElNino, J. Geophys. Res., 100, 4893-4920, 1995.

Kindle, J. C., and J. D. Thompson, The 26-day and 50-dayoscillations in the western Indian Ocean: Model results,J. Geophys. Res., 94, 4721-4736, 1989.

Kindle, J. C., G. W. Heburn, and R. C. Rhodes, An esti-mate of the Pacific to Indian Ocean throughflow from aglobal numerical model, in Further Progress in EquatorialOceanography, edited by E. J. Katz and J. M. Witte, pp.317-321, Nova Univ. Press, Ft. Lauderdale, Florida, 1987.

Kindle, J. C., H. E. Hurlburt, and E. J. Metzger, On the sea-sonal and interannual variability of the Pacific to IndianOcean throughflow, paper presented at Western PacificInternational Meeting and Workshop on TOGA COARE,Centre ORSTOM de Noumea, Noumea, New Caledonia,1989.

Kundu, P. K., and J. P. McCreary, On the dynamics of thethroughflow from the Pacific into the Indian Ocean, J.Phys. Oceanogr., 16, 2191-2198, 1986.

Levitus, S., Climatological atlas of the world ocean, NO A AProf. Pap. 13, U.S. Govt. Print. Off., Washington, D.C.,173 pp., 1982.

Matsuno, T., Quasi-geostrophic motions in the equatorialarea, J. Meteorol. Soc. Jpn., 44, 25-42, 1966.

McCalpin, J. D., A note on the reflection of low-frequencyRossby waves from realistic western boundaries, J. Phys.Oceanogr., 17, 1944-1949, 1987.

McCalpin, J. D., Rossby Wave generation by poleward prop-agating Kelvin waves: The mid-latitude, quasigeostrophicapproximation, J. Phys. Oceanogr., 25, 1415-1425, 1995.

McCreary, J. P., and Z. Yu, Equatorial dynamics in a 21/2-layer model, Prog. Oceanogr., 29, 61-132, 1992.

Mesinger, F., and A. Arakawa, Numerical methods used inatmospheric models, GARP Publ. Ser. 17, 64 pp., WorldMeteorol. Organ., Geneva, 1976.

Meyers, G., T. Qu, and J. R. Donguy, Ocean mechanisms ofSST variability in the south tropical Indian Ocean, paperpresented at International Conference on Monsoon Vari-ability and Prediction, International Centre for Theoreti-cal Physics, Trieste, Italy, 1994.

Molcard, R., A. G. Ilahude, M. Fieux, J. C. Swallow, andJ. Banjarnahor, Low frequency variability of the currentsin Indonesian channels, Deep Sea Res., 41, 1643-1661,1994.

Murray, S. P., and D. Arief, Throughflow into the IndianOcean through the Lombok Strait, January 1985 - Jan-uary 1986, Nature, 333, 444-447, 1988.

Murray, S. P., D. Arief, J. C. Kindle, and H. Hurlburt, Char-acteristics of the circulation in an Indonesian archipelagostrait from hydrology, current measurements, and model-ing results, in The Physical Oceanography of Sea Straits,pp. 3-23, Kluwer Academic, Norwell, Mass., 1990.

Mysak, L. A., and G. J. Mertz, A 40- to 60-day oscillationin the source region of the Somali Current during 1976,J. Geophys. Res., 89, 711-715, 1984.

Nicholls, N., The Southern Oscillation and Indonesian seasurface temperature, Mon. Weather Rev., 112, 424-432,1984.

Piola, A. R., and A. L. Gordon, Pacific and Indian upper-layer salinity budget, J. Phys. Oceanogr., 14, 747-753,1984.

Schott, F., M. Fieux, J. Kindle, J. Swallow, and R.Zantropp, The boundary currents east and north of Mada-gascar, 2, Direct measurements and model comparisons,J. Geophys. Res., 93, 4963-4974, 1988.

Shinoda, T., and R. Lukas, Lagrangian mixed layer mod-eling of the western equatorial Pacific, J. Geophys. Res.,100, 2523-2541, 1995.

Toole, J. M., Problems of interbasin exchanges and marginal-sea overflows, Bull. Am. Meteorol. Soc., 68, 136-140,1987.

18,420 VERSCHELL ET AL.: EFFECTS OF INDO-PACIFIC THROUGHFLOW

Wajsowicz, R. C., Interannual variations in the south Pa-cific wind stress curl and the predicted magnitude of theIndonesian throughflow, Toga Notes 11, pp. 8-12, NovaSoutheastern University Press, Dania, Florida, 1993.

Wajsowicz, R. C., Interannual variations in the Indo-Pacificthroughflow forced by ECMWF wind-stress anomalies for1985-1989, Toga Notes 15, pp. 6-11, Nova SoutheasternUniversity Press, Dariia, Florida, 1994a.

Wajsowicz, R. C., A relationship between the interahnualvariations in the south Pacific wind stress curl, the In-donesian throughflow and the west Pacific warm waterpool, J. Phys. Oceanogr., 24, 2180-2187, 1994b.

Wallcraft, A. J., The Navy Layered Ocean Model Users'Guide, NOARL Rep. 25, 21 pp., Nav. Oceanogr. and At-mos. Res. Lab., Stennis Space Center, Miss., 1991.

White, W. B., G. Meyers, J. R. Donguy, and S. E. Pazan,

Short-term climatic variability of the Pacific Ocean during1979-1982, J. Phys. Oceanogr., 15, 917-935, 1985.

Wunsch, C., D. Hu, and B. Grant, Mass, heat, salt and nutri-ent fluxes in the south Pacific ocean, J. Phys. Oceanogr.,13, 725-753, 1983.

J. C. Kindle, Naval Research Laboratory, Code 323, Sten-nis Space Center, MS 39529.

J. J. O'Brien and M. A. Verschell, Center for Ocean-Atmospheric Prediction Studies, Florida State University,020 Love Building, Tallahassee, FL 32306-3041. (e-mail:markv@coaps.fsu.edu)

(Received July 8, 1994; revised March 24, 1995;accepted July 3, 1995.)