Sep 15, 2000 - Indonesian Throughflow (ITF) transport (measured in. Sverdrups) is not sufficient for an adequate simulation of equatorial watermasses.
GEOPHYSICAL RESEARCH LETTERS, VOL. 27, NO. 18, PAGES 2941-2944, SEPTEMBER 15, 2000
Sensitivity of Equatorial Watermasses
to the
Pacific and Indian Ocean
Position
of the
Indonesian
Throughflow Keith B. Rodgers 1, Mojib Latif, and StephanieLegutke Max-Plam:k-Institut fiir Meteorologie, Hamburg, Germany
Abstract. The sensitivity of the thermal struct•re of the equatorial Pacific and Indian Ocean pycnoclines to a model's representation of the Indonesian Straits connecting the two basins is investigated. Two integrations are performed using the global HOPE ocean model. The initial conditions and surface forcing for both cases are identical; the only difference between the runs is that one has an openingfor the Indonesian Straits which spansthe equator on the Pacific side, and the other has an opening which lies fully north of the equator. The resulting sensitivity throughout much of
the upper oceanis greaterthan 0.5øC for both the equatorial
Indian
and Pacific.
A realistic
sinrelation
of net
Model Description The model used for this study is the global HOPE
(HamburgOceanPrimitiveEquation)model[Wolff et al., 1997]. The modelwasoriginallyappliedto tropical circulationstudiesby Latif [1987]. Model calculationsareperformedonan ArakawaE-grid[Amkawaand Lamb, 1977],and horizontalderivativesare calculated using second-ordercentered differences. The
resolution
used here
is the
sanhe as that
used
for the ocean componentof the coupled GCM ECHO-
2 (ECHAM/HOPE)integrations as discussed by Latif and Barnett [1994,1996]and Frey et al. [1997]. The
IndonesianThroughflow(ITF) transport (measuredin ocean has 20 unevenly spaced vertical layers, with 10 Sverdrups)is not sufficientfor an adequatesimulation of these in the upper 225 meters. The meridional resof equatorial watermasses.The ITF must also contain a realistic source
admixture
of northern
and southern
Pacific
water.
olution is 2.8ø in the extratropics, and is gradually increaseduntil it reaches0.5ø within 10ø of the equator, enabling the model to have a realistic representationof the narrow equatorial current system and waves. The zonal resolution
Introduction
The purpose of this study is to better understand the sensitivity of equatorial Pacific and Indian Ocean watermassesto the position of the Indonesian Straits connectingthe Pacific and Indian Oceans. It is known
that the IndonesianThroughflow(ITF), with a mean transport of between 5 and 10 Sverdrups,plays an important role in determining the watermass properties
is 2.8 ø for the entire domain.
The daily forcing fields for momentum and heat used for the experimentsdescribedhere •ire taken from an integration of the ECHAM4 AGCM at T42 resolu-
tion [Roeckner et al., 1996]. The lowerboundarycondition for that integration consistedof AMIP climatological SSTs for the period 1979-1988. In addition to using the heat fluxes from the ECHAM4 integration, a relaxation term to AMIP climatologicalSSTs is accom-
plishedwith a restoringterm of 40 W m-2 K -•. Sea of the equatorialIndian and PacificOceans[Gordon, surface salinities were relaxed to observed values with 1986; Godfrey,1996]. Previousoceanmodelingstuda timescale of 30 days. For both cases,the model was ies [Hirst and Godfrey,1993;Murtuguddeet al., 1998; initializedwith annualmeantemperature[Levitusand Rodgerset al., 1999;Rodgerset al., 2000],comparing Boyer•1994]and salinity [Levituset al., 1994]climaruns with an "open" versus "closed"ITF, have shown a strong sensitivity in the resulting sea surfacetemper-
tologies, and integrated for 40 years. The two configurationsusedfor the sensitivitystudy ature (SST) in the EasternEquatorial Pacificof order described here differ only in their land-sea mask de1øC. The sensitivitystudy which we presenthere differs scribingthe northern coast of New Guinea. The landfrom the earlier onesin that we comparetwo runs which sea masks for casesA and B are shown in Figures l a both include an ITF, but with different representations and lb, respectively. In the figures,grey shadingindiof the coastline of New Guinea. cates land points for the model donhain,with the realworld coastlinessuperposedas black contours. For case Now at LSCE, Gif sur Yvette, France A, the northern coast of New Guinea doesn't reach the equator; its northernmost extent is 3øS. For caseB, it Copyright2000 by theAmericanGeophysical Union. reaches2øN. Case A is the control run; it is the landsea mask which has been used for the coupledECHO-2 Papernumber1999GL002372. 0094-8276/00/1999GL002372505.00 runs. This study was motivated by a desire to under2941
2942
RODGERS
ET AL.: ITF: OPEN VS CLOSED
'
-. - ..L_•..I..•.,
1 x
Flowfield, Ca•
Flowfield, 6ase A. Z=125
•.•
B, Z=125
Figure 1. The modelflowfields in the westernequatorialPacificat 125m depthfor (a) caseA and (b) caseB.
stand why the ECHO simulationsare characterizedby a cold bias in eastern equatorial Pacific SSTs.
It is knownfrom observations [Godfrey,1996]that
Current (SEC) water flowingwestwardin the Pacific between 2øS and 8øS impinging on the coast of New Guinea near 150øW joins the ITF, and thus flows to the Indian Ocean. The ITF, however, is fed primarily by Northern Hemisphericthermoclinewater in caseB
the vast majority of the Throughflow passesthrough the Makassar Strait in the real ocean. In Figure la the Makassar Strait correspondsto the passagecross- (Fig. lb). Away from this region, the differencesin ing the equator at approximately 116øE. The Makassar circulation fields for the two runs are quite small. As Strait is bounded to the east by Sulawesi,whosenorth- we shall see, this relatively local change in circulation ern boundary at 3øN forms the southern boundary of has basin-wide consequencesfor the upper ocean heat the throughflow opening on the Pacific side. For the content and watermasspropertiesof the tropical Pacific model domain shown in Figure la, the position of the and Indim• Oceans. The annual mean vertically integrated Indonesian westernboundary point for the Throughflow is shifted 2 gridpoints,or approximately6ø longitude, to the east of throughflowtransport for caseA is 14.7 Sverdrups,and the real-world Makassar Strait. Given that the strength for case B it is 15.2 Sverdrups. Thus the differencein of the Throughflowcalculatedby the model is quite sen- mean transport between the two runs is lessthan 4 per sitive to the width of the passage,and our interest in cent. Howeverthere is a significantdifferencein the amunderstandingthe sensitivityof the oceancomponentof plitude of the seasonalcyclefor the two runs: for caseA the coupledECHO model to the position of the mouth the amplitudeof the seasonalcycleis 11.0 Sverdrups, of the Throughflow, the extendednorthern boundary of whereas for case B it is 4.8 Sv. The difference (caseB minuscaseA) in annualmean New Guinea shown in Figure lb is meant to represent dynamically the effective barrier posed by the coast of temperaturesat 125 m depth is shownin Figure 2.
Clearly the Pacificthermoclineis warmerfor caseB
Sulawesi in the real ocean.
After 20 years of model integration, the differences between the two runs change very little for the Pacific and Indian
Ocean thermoclines.
The results shown are
for the annual mean of the 30th year of model integration. First we describe differences in circulation fields, and then we describe differences in the temperature fields for cases A and B.
Results The differences
in annual
mean flowfields
for the west-
ern equatorial Pacific at 125 meters depth for casesA and B are shownin Figures la and lb, respectively.For each case,the even componentof the land-seamask cor30'E 60'E 90'E 120'E 150'E 18½ 153'W 12ffW ,90'W 60'W x respondingto the E-grid is shown. This depth horizon TemperatureDifference,Case B minusCase A, Z=125 is chosenbecauseit correspondsto the depth of maximum ITF transport for the model, where the differ- Figure 2. The annual mean temperaturedifference encesin circulation are expected to be largest. It can (caseB minuscaseA) at 125m depthfor the 30th year be seen(Fig. la) that much of the SouthEquatorial of model intergration.
RODGERS
ET AL.- ITF-
OPEN
VS CLOSED
2943
there is a 2øC meridionalgradient in temperature across the equator along isopycnalsurfacesin the equatorial
Pacific[Figure11aof Rodgers et al., 1999],whichis due to differences
in surface
fleshwater
fitLxes between
the
North and South Pacific. For isopycnalsurfaceswithin the pycnocline, South Pacific waters are warmer and saltier
than
North
Pacific
waters.
Secondly,the only significantchange in circulation between
case A and case B is local to the western bound-
ary region of the equatorial Pacific thermocline. When /•]•
....• 150'E
•
,
m 180'
,
I
,
,
150'W x
I 120*W
•
the ITF openingis to the southof the equator(caseA),
•
there is a tendencyfor warm southernhemisphericwater to feed the ITF, and for coldernorthern hemispheric waters to supply the EUC. Conversely,when the open-
•'W
Temperature Difference, CaseB minusCaseA,
Figure 3. The annum mean temperature difference ing is to the north (caseB), the ITF is primarily sup(caseB minus caseA) along Y=0øN for the 30th year plied by water from the Northern Hemisphere,and there of model integration. is a strongersoutherncomponentin the upwellingin the eastern equatorial Pacific. The structure of the open than for case A, and the Indian Ocean thermocline is ocean equatorial currents is largely the same for both cooler. Temperature differencesin excessof 0.5ø extend cases in both the Pacific and Indian Oceans. Thus, over much of the equatorial Pacific thermocline, with the basin-wide sensitivity in thermocline temperatures maximum differencesin excessof 1.5øCnear the equator which we have seenin this study results from localized changesin circulation. The flow scenariofor case B is at 160øE. These differences trace the downstream advective signature of the Equatorial Undercurrent. Like-
supportedby observations.Ffield and Gordon[1992]
wise for the Indian Ocean, where differences in excess have shown that the ITF is dominated by northern
of løC exist throughout much of the equatorial thermocline, maximum differences in excess of 2ø between
8øS and 12øS trace the position of the South Equatorial Current (SEC). Upon reachingthe coastof Africa at the western boundary, a portion of the SEC turns north, and then ventilates the equatorial thermocline. A vertical sectionof temperature differencesalong the equator (Figure 3) revealsa significantchangein the heat content of the upper tropical oceanswhen going from case A to caseB. Case B is warmer over much of the Pacific thermocline by more than 0.5øC, with differencesof greater than 1.0øC correspondingto the •pper part of the EUC. The maximum differenceoccurs
sources,and Tsuchiyaet al. [1989]and Giriou and Toole[1993]have shownusingsalinitymeasurements that the EUC is comprisedof more than 50% South Pacific
water.
It will be of interest to seewhat the sensitivity of the coupled system would be to the changesin the landsea mask
as described
here.
The
model
used here is
a component of the coupled atmosphere-oceanmodel
ECHO-G [Legutke and Voss,1998].An earlierversionof the coupled GCM, ECHO-2, exhibits SSTs which have
a cold bias in the easternequatorialPacific [Frey et al., 1997].Thus the land-seamaskshownin Figurelb
should improve matters, sincefor the stand-alone ocean model it results in more warm southernhemisphericwaport within the Indonesian Straits occurson a shallower ter being upwelled in the eastern equatorial Pacific. It isopycnalsurface than does the core of the EUC in the western equatorial Pacific. At 50 rn depth, the effects are clearly larger in the central and easternpart of the above the core of the EUC
since the maximum
trans-
basin.
The maximum temperat,re differencesseenin Figure 3 are not merely an artifact of changesin the mean depth of the pycnoline. These differencesare equally large when the model's output is analysedisopycnally. This can be seen in Figure 4, which showsthe annual mean temperature difference for the cr0=25.0 isopycnal surface. Here temperature differencesin excessof 0.5øC extendwithin the undercurrentto approximately 120øW.
Discussion
Figure 4. The annual mean temperature difference The differencesin thermoclinetemperaturesbetween (caseB minuscaseA) on the cr0=25.0isopycnalsurface casesA and B are the result of two factors. Firstly, for the 30th year of model integration.
2944
RODGERS
ET AL.:
ITF:
OPEN
VS CLOSED
remains to be seenwhat the effect would be on coupled Frcy, H., M. Latif, and T. Stockdale, The Coupled GCM ECHO-2. Part I: The Tropical Pacific. Mon. Wea. Rev., atmosphere-oceanmodes of variability, such as ENSO.
For the 1/6ø ParallelOceanClimatemodel(POCP), GordonandMcClean[1999]haveshownthat the simulated ITF is dominated by Southern Pacific water, with the consequencethat their Pacific thermocline is too
cold. Wajsowicz[1999]has arguedthat this is largely due to the choiceof wind stressclimatologyusedto force the model, and points out that there exist important discrepenciesbetween commonly used wind stress climatologiesin the position of the mean zero wind stress curl line at the western boundary or the basin. This effect could certainly have the samesign as the sensitivity to model bathymetry we have discussedhere. Important questions regarding the dynamical controls on the Throughflow transport pathways and composition will require higher resolution than is currently available for ocean componentsof coupled models such
as ours. For example,Nor [1996]hasshownthat the retrofiectionof the SEC could play an important role in determining the compositionof the Throughflow.
Vol. 125, No. 5, 703-720, 1997. Giriou, Y., and J. M. Toole, Mean circulation of the upper layers of the western equatorial Pacific Ocean. J. Geophys. Res., 98, 22,495-22,520, 1993.
Godfrey, J.S., The effect of the Indonesianthroughflowon oceancirculationand heat exchangewith the atmosphere: A review. J. Geophys. Res., 101, 12,217-12,237, 1996. Gordon, A.L., Interoceanexchangeof thermoclinewater, J. Gcophys. Res., 91, 5037-5046, 1986. Gordon, A.L., and J.L.McClean, Thermohaline Stratification of the Indonesian Seas: Model and Observations, J. Phys. Oceanogr., 29, 198-216, 1999. Hirst, A. C., and J. S. Godfrey, The role of Indonesian throughflowin a global Ocean GCM, J. Phys. Oceanogr., 23, 1057-1086, 1993.
Latif, M., Tropical Ocean Circulation Experiments, J. Phys. Occanogr., 17, 246-263, 1987. Latif, M., and T.P. Barnett, Causes of dccadal variability over the North Pacific and North America, Science, 266, 634-637,
1994.
Latif, M., and T.P. Barnett, Decadal climate variability over the North Pacific and North America: Dynamics and preidietability, J. Clim., 9, 2407-2423, 1995.
Lcvitns, S., R. Burgeft, and T.P. Boyer, World OceanAtlas 199•, vol. 3, Salinity, 99 pp., U.S. Dep. of Comm., Washington,D.C., 1994.
Conclusion
Levitus, S., and T. P. Boyer, World OceanAtlas 199•, vol. 4,
Two runs were conducted using the global HOPE Temperature,117 pp., U.S. Dep. of Comm., Washington, ocean model, one where the northern boundary of the D.C., 1994. model's representation of New Guinea extends to 3øS Murtuguddc, R., A.J. Busalacchi,and J. Beauchamp,Seasonalto-intcrannual effects of the Indonesian throughflow on (caseA), and another where the northern boundary of the tropical Indo-Pacific Basin. J. Gcophys. Res., 103, New Guinea extendsto 2øN (caseB). The initial con21,425-21,441. ditions and forcing for the two runs were identical. The Nof, D., What controlsthe origin of the Indonesianthroughresulting differencesin temperaturesthroughoutmuch flow?, J. Gcophys. Res., 101, 12,301-12314, 1996. of the equatorial Pacific and Indian Ocean thermoclines Rodgers,K.B., M.A. Cane, N.H. Naik, and D.P. Schrag,The are in excess of 0.5øC.
cific thermocline
Case B exhibits
and a cooler Indian
a warmer
Pa-
Ocean thermocline
role of the Indonesian Throughflow in equatorial Pacific thcrmoclinc ventilation, J. Geophys. Res., 10•, 20,55120,570, 1999.
than caseA, due to changesin the sourceof Indonesian Rodgers, K.B., D. P. Schrag, M.A. Cane, and N.H. Naik, Throughflow water. For case A, the Throughflow is The Boxnb-•4CTransientin the PacificOcean, J. Geophys. Res., 105, 8489-8512, 2000. supplied primarily by warm and salty water from the Southern Hemisphere, resulting in a pronouncedcold Roeckner, E., K. Arpe, L. Bengtsson, M. Christoph, M. Claussen,L. Diimenil, M. Esch, M. Giorgetta, U. Schlese, and fresh signature of North Pacific water in the EUC. and U. Schulzweida, The atmospheric general circulaFor caseB, the Throughflowis insteadsuppliedlargely tion model ECHAM-4: Model description and simulaby cool and fresh North Pacific water, with the result tion of present-day climate. Reports of the Max-Planckthat there is a stronger presenceof waxm south Pacific Institute, Hamburg, No. 218, 90pp., 1996. water within the EUC, in agreementwith observations. Tsuchiya, M., R. Lukas, R. Fine, E. Firing, and E. Lind-
strom, Sourcewaters of the Pacific Equatorial Undercurrent, Prog. Occanogr., 23, 101-147, 1989. Acknowledgments. We would like to thank Ernst MaierRcimer for for his support and suggestionsregarding the Wolff, J.E., E. Maier-Reimcr, and S. Legutke, The Hamburg Ocean Primitive Equation Model, Technical Report, No. HOPE ocean model. This work was supported by the Eu13. German Climate Computing Center (DKRZ), Hamropean Union through the SINTEX project. Keith Rodgers burg, 98 pp., 1997. wassupportedby the Visiting ScientistProgram of the MaxPlanck-Society. The ocean model integrationshave been Wajsowicz,RoxanaC., Modelsof the SoutheastAsian Seas, J. Phys. Oceanogr.,29, 986-1018, 1999. performedat the DeutchesKlimarechenzentrum(DKRZ). References Arak_awa,A., and V.R. Lamb, Computational designof the basic dynamical processesof the UCLA General Circulation Model. Methods Cornput. Phys., 17, 173-265, 1977. Fficht, A., and Gordon, A.L., Vertical mixing in the Indonesian thcrmoclinc, J. Phys. Occanogr.,22, 184-195, 1992.
K. B. Rodgers, Mojib Latif, and S. Lcgutke, MPI, Hamburg. (e-maih rodgers(•dkrz.de,latif(•dkrz.de, and lcgutke(•dkrz.dc) (Rcceiw•'dJuly 16, 1999; revised December 23, 1999; accepted January 21, 2000.)
