A Numerical Study of Water Circulation in A Thermally Stratified Embayment

Journal of Ocean University of Qingdao ( Oceanic and Coastal Sea Research )ISSN 1671-2463 April 302003 ,Vol.2 ,No.1 pp.24-34httpE ] /1ww. ouC. edu. ci JlbyvL ]E-mail :xbynvb@ mail.0ulC. edu. cnA Numerical Study of Water Circulation in A ThermallyStratified EmbaymentWANG x. H.!), WANG x. L.2)1 ) School of Geography and Oceanography , University Colle , ADFA , University of New South Wales ,Canberra ACT 2600 , Australia2 ) Department of Mechanics , Zhejiang University , Hangzhou 310029 , P. R. China( Received July 22 2002 ; acepted February 28 2003 )Abstract Princeton Ocean Model is used to study the response of Jervis Bay , NSW , Australia , to the local wind and remoteshelf coastal trapped wave( CTW )forcings in summer seasons when the water column is stratified by the water temperature.The study has revealed that the response of bay to the wind forcing is the generation of the wind driven currents and the in-ternal Kelvin waves( IKW ). However , both temperature and flow sub- inertial oscillations in the bay are weaker than thosefrom the observations and the correlation between the modeled and observed low frequency currents is low. In response tothe forcing of CTWs on the adjacent shelf , IKWs are also established in the bay and amplitudes of sub inertial oscillations oftemperature and currents agree better with the observations. It can be concluded that sub inertial baroclinic flows in the bayis dominantly forced by remote CTW on the shelf adjacent to Jervis Bay during thermally stratified summer seasons.Key words Princeton Ocean Model ; water circulation ; stratified embayment ;Jervis BayNumber ISSN 1671-2463 2003 )01-24-11waves( CTW ) have been observed propagating towardsthe north along the Australian East Coast. The peri-1 Introductionods of these waves are of 6- -10d( Church et al. , 1986a,The sub- inertial circulation in coastal embayment b ).and its exchange with the open shelf waters can haveThe scattering of the coastal trapped waves nearimportant environment consequences. An example of Jervis Bay can be decoupled by a barotropic CTW andsuch a system is Jervis Bay ,a small semi closed embay-a baroclinic internal Kelvin wave( Middleton , 1994 ). .ment located on the East Coast of Australia. The bay Craig and Holloway ( 1991 ) have shown that flowis approximately 15 km long and 8 km wide with argenerated inside the bay by a barotropic forcing on thearea of 124 km2. The averaged depth of the bay is 15m shelf is insignificant , but an internal Kelvin wave withand is connected to the continental shelf through an a period of 8d can force a stronger baroclinic flow inopening which is 3.75 km wide and 40m deep( Fig.1 ). the bay. The amplitude of this decoupled internalThe adjacent continental shelf gradually increases itsKelvin wave on the shelf is around 0.1 ms' near thedepth to 120m within a distance of 5 km offshore , and口人AYthe rest of the shelf remains flat until 20 km offshorewhere the shelf break starts.Based on observations ,Holloway et al.( 1992 )have1 LOCATIONwshown that Jervis Bay is a low energy environment withNEan energy peak at a sub- inertial frequency of 8 days insummer. Tidal currents in the bay are weak with M2being the dominant tidal constituent with current中国煤化工4speeds of 0.07 ms”near the bay entrance and less than0.01 ms”in the northern reaches of the bay. One ofHCNMHGthe mechanisms in forcing the sub-inertial circulation20in the bay is the scattering of the coast trapped wavesue gug morgon the adjacent continental shelf. The coastal trappeda)* Correspondinghubor. Tel 0061-02 -62688473E-mail :hua. wing@ adfa. edu. auWANG X.H. et al. : A Study of Water Circulation in a Thermally Stratified Embayment2521.0fers discussion and the conclusions of the paper formSection 5.2 Numerical Model/10'The three-dimensional ocean model of Jervis Bay isj3based on Princeton Ocean Model ( POM , Blumbergand Mellor , 1983 ; 1987 ). POM solves the primitiveequations for surface elevation，momentum，tempera-ture and salinity on the Arakawa C-grid in the hori-zontal and a σ-coordinate system in vertical. Mellor-0.021Yamada( 1982 ) level 2.5 turbulence closure scheme isx (km)(b)used to simulate vertical sub-grid mixing. The hori-zontal diffusivities are calculated by the Smagorinskydiffusion scheme and horizontal eddy diffusivity andFig.1(a) Map of Jervis Bay showing the locations of theheat diffusivity are assumed to be equal. Both hydro-static and Boussinesq approximations are applied. Themoorings and the meteorological station;( b) Jervis Baymodel domain , cross sections of lines AB and CD being in-full set of the equations is referred to Mellor( 1992 ).dicatedbay entrance( Middleton , 1994 )and 0.02 ms 1 inside 2.1 Model Setupof the bay( Craig and Holloway , 1991 ).In order to resolve the surface mixing layer , the mod-Furthermore strength of the circulation induced by el uses a logarithmic distribution for vertical σ coordi-the scattering of the CTWs is dependent on the strati-nate grids near surface boundary. There are 8 sigmafication that determines the buoyancy frequency N. In levels with the distribution σ=0，-0.031，-0.063 ,summer ,Jervis Bay is stratified by the water tempera--0.125 , -0.25 , -0.5 , -0.75，-1.0. The horizontalture, and the summer observations in the bay con- grid sizes are△x= Oy= 500m. The time steps forfirmed that there are strong correlation between sub-external and internal mode are 3s and 30 s respective-inertial currents and temperatures at different locations ly. As the freshwater discharge is negligible ( salinityinside the bay or between those at different depthsaccounts for less than 10% of the density variation( Craig and Holloway , 1991 ).across the bay ), the salinity S is set to a constant ofThe observations of Holloway et al. ( 1992 ) also 35 in the model so that temperature determines theshowed the temperature amplitude of the wave(of 8d water density. The Coriolis parameter f of the modelperiod ) reaching 0.9C at 15m at the south channelisset tobe -0.8x10-*s+.( SC)of the bay( Fig.1a). This amplitude of tempera-A sketch of the model domain is shown in Fig.1 b.ture corresponds to an isotherm oscillation of 4.5mat In cross shelf( x ) direction , 48 grid points are taken20m water depth with the observed buoyancy fre- with the last five grid size being linearly telescoped.quency of 0.03s. The bottom temperature amplitude The actual bathymetry of Jervis Bay is taken for thewas0.6C at the northern end of the bay where the model but the depth is constant for water deeper thanwater depth reaches 12 m.120 m. This agrees with the topography of the studyLocal wind forcing can also generate sub-inertial area where the continental shelf is relatively flat. Incurrents in Jervis Bay. Close correlation has been alongshelf( y ) direction , 73 grid points are used. Atfound between the 8d period longshelf winds and deepeach end of the cross- shore boundary , 10 grid pointstemperature and currents in the bay( Holloway et al.，are also linearly telescoped except for the wind forced1992 , Craig and Holloway ，1991 ), despite a weak model. This larger domain is used to reduce the effectlongshelf wind component observed at that frequency of relection of internal waves from these boundaries.(~1ms' ).The domain( xt-y coordinate system ) is turned by 37°. This paper uses Princeton Ocean Model to study to the east from the standard map orientation , so theJervis Bay sub- inertial circulation in summer seasons.alongshore direction does not align with north. ForTwo numerical experiments are conducted to exame convenience ，the positive y direction is still referred asthe response of the bay to the forcing mechanisms of中国煤化工the paper.the wind and the shelf CTWs respectively in order tone rest. As we are pri-elucidate the difference in its dynamic response. ThemariMYHCNMHGioninthebay,asum-organisation of the paper is as follows. Section 1 gives mer temperature profile measured in the center of thean introduction of the study. Description of the model bay with a water depth of 25 m is used to initialize thesetup，boundary conditions and model forcings is of- model.fered in Section 2. Sections 3 presents model resultsAlthough some temperature difference has been ob-and their boarons with observations. Section 4 of- served between the bay and its adjacent shelf ,no hori-26Journal of Ocean University of Qingdao .2003 ,Vol.2 ,No. 1zontal variation of temperature is considered in the cross- shelf. Some storm events gave a maximum windmodel for simplicity. For water deeper than 25m ,the speed of 10ms 1 ,although not generally sustained formodel uses the temperature measured on theshelf Fig.2 6d. Table 2 shows the wind amplitudes and phases atshows the initial temperature profile ，which givessub-inertial periods. Among the low frequency windsmaximum buoyancy of 0.03s' at 15m. The model pa- with periods from 6 to 10d , the 8d period wind hasrameters described above are summarized in Table 1.maximum amplitudes in both cross-shore and long-shore componentsof 2ms' and 1.1 ms 1 respectively.0|The total wind amplitude is 2.3ms-' . The ( kinemat-40tic ) unfiltered wind stress is calculated from the ob-served wind velocity using Oey and Chen( 1992 ):昌60|30τ=Cj|w|w00Ca=f1.275x 10。0≤|w|≤5( ms' ) .,(1)(0.95+0.065|w|)x106 5≤|w1≤20(ms' )20一1420Table 2 Amplitude( ms 1 ) for observed summer cross shoreTemperature (C)wind component( W ) ,and longshore wind component. ( Wa )at sub-inertial periodTFig.2 The observed summer temperature on 5 December 1988. (d) Amp Phase Amp Phase Amp Phase Amp Phase Amp PhaseW。0.6 217 1.7 -59 2.0 172 0.4 840.5134Table 1 Jervis Bay model parametersW。0.91870.71431.1620.5621.0-17NameSymbolValueCoriolis frequency-0.8X104g1Grid spacingOx= Oy500 mwhere Ca is the drag coefficient( multiplied by the ra-External time stepOte3stioof air and sea water densities ) and w is the obInternal time stepAt;30sserved wind velccity vector shown in Fig.3. The unfil-Cross- shelf grid pointsIM48Longshelf grid pointsJM73tered wind stress of( 1 ) is uniformly applied to the en-Maximum buoyancy frequencyNm0.03s1tire model domain for the summer season..10ms'AN2.2 Model Boundary ConditionsThe conditions used for closed boundaries such asthe land and the seabed are prescribed by POM. No0 100200 300 400 S0o 600 700 800 900Time (h)heat flux is specified at the sea surface. A sea wall is .placed at the offshore boundary and its effects on thecirculation in the bay are minimal and ignored. CyclicFig.3 The low pass filtered ( one-day cut off period ) windboundary conditions are used for the northern and .vector observed at the Governor Head between 6 December1988 and 13 January 1989( 900h)southern cross -shelf boundaries in the wind forcing ex-periment. For the CTW forcing experiment , internalIn the CTW forcing experiment , the model is driv-velocities of the forcing CTWs are specified at the two en by a simple internal Kelvin wave( IKW )on thecross shelf open boundaries. As the longshelf domain shelf representing the CTWs as no data is availableis small to compare with the longshelf CTW wave- there. The period of this IKW is 8d and coincideslength , zero phase difference is used for the specifiedwith the peak energy observed on the shelf. The ve-velocities at two boundaries. A separate model run locity of the IKW iswith a phase difference of 62 degrees ( correspondingto longshelf distance in the model domain ) betweenu(x心;t )=vco( πz/Ho )xin( 2πt/T )e R。(2)the specified velocity fields at the northern and south-ern open boundaries shows that sub-inertial flows gen-where Ho= 80m, T=8d , R; is internal Rossby ra-erated by the CTW are not sensitive to the phase change.dius of deformation and calculated to be 5000m by adepth averaged buoyancy frequency N=0.02s' ,V0 is2.3 Model Forcings0.11中国煤化工y Middleton( 1994 )byIn the wind forcing experiment , the model is driven assunW into an IKW nearby the observed wind. The hourly wind data wasJervisYH. CNM HGre Variation in its shefrecorded by a meteorological station at the Governor bathymetry. The model is run for 4 periods. After theHead ( Fig.1 ) between 6 December 1988 and 13 Janu-first cycle，the model has reached a quasi- steady state.ary 1989. The low passed( one day cut-ff period ) Thus the first period of the simulation is discarded andtime series of the wind velocity are shown in Fig.3.only last three cycles of the model simulations are ana-The alongshE&nponent is smaller than that of the lyzed and compared with observations.WANG X.H. et al. : A Study of Water Circulation in a Thermally Stratified Embayment27prediction show negative phase lags of approximate 1 dbetween stations PP and NE , and a positive phase lag3 Resultsof about 1 d between stations NE and NW .The results of the wind and CTW forcings experim-ents are compared with the observations which consistTable 3 Comparison of observed and modelled amplitude oftemperature oscillation with a period of 8dof temperature and current velocity measured at 6 sta-Temperature amplitude(C )tions shown in Fig.1. The names of the stations areBowen Island ( BI ), Creswell ( CW ), South ChannelInstrumentObservationWind driven CTW drivendepth( m )model .model( SC ), Northwest( NW ) , Northeast ( NE ) and PointBIPerpendicular( PP ). These measurements were taken0.730.140.68during the same period as the wind data. More detailsPPof the field programs are referred to Holloway et al.200.740.50( 1992 ).0.520.160.403.1 Data vs Wind Forcing Experiment7.40.250.190.3211.4 .0.420.22Fig.4 shows low passed( one- day cut off period )ob-15. 50.97served and model predicted time series of the current19.50.070.35speed at various levels for stations of SC , NW ancNE. The model predicted current speed is generally5.40.33 .0.300.28smaller than the observation , and the correlation be-9.50.590.490.70NEtween the observed and modelled currents is low.From top to bottom in Fig.4 , the correlation coeffi-.5.40.1811.50.630.55cients are calculated to be 0.15 ,0.14 ,0.05 ,0.23 ,0.08and 0.39 ,respectively. The filtered hourly time seriesTemperature phase lag(d)in Fig.4 were used in the above correlation coefficientcomputation.NW-BI0.1-0.11.4SC(7m)NE-NW0.71.90.3100 200 300 400500 600 700二 900PP-NE-0.8-1.61.3SC (11m)BI-PP-0.2-0.5moye9100200300400500600700 800 _ 900Table 4 shows the semi-major axis and surface toSc (20 m)bottom phase lags of the current ellipses with 8-day器0.2100 200 300 400 500 600 700 800 900period. Comparisons show that the model simulatedNW (5.4 m)o.xppcvsub- inertial currents are weaker than those observed in100 200 300 400 S500 600 700 8OO 5900all the stations. Both the model prediction and obser -NE (5.4 m)vation show significant surface to bottom phase lags.010200300400500600700800900NE(11.5m)Table 4 Comparison of observed and modelled semi-majoraxis and sub- surface/ bottom phase lags of low frequencyTime (h)currents with a period of 8dSemi-major axis( cms1 )Wind drivenCTW drivendepth( m)Fig4 The low. pass filtered time series of the observed andsCpredicted current speed from the wind driven model at vari-2.01.52.5ous depths for locationsof SC.NW and NE11.41.1The observed currents are in thick lines.15.51.80.81..19.51.60.6Table 3 shows the model predicted and observedNWtemperature amplitude for the 8 d frequency oscillation.2.1The comparison demonstrates that the local wind can.51.2only produce a weak sub-inertial temperature oscillation中国煤化工1.62.4at the southern stations. However ,a better compari-MHCNMH G_son is achieved at the northern stations of NW andSurface/Bottom phase lag(。)NE.Table 3 also shows the phase lags of the tempera-ture oscillation between stations BI，NW , NE , PPSC189161277( the phase lag, of NW-BI is the phase at NW subtract-717ed by theph8搋BI ). Both observation and modelE69129_528Journal of Ocean University of Qingdao .2003 ,Vol.2 ,No. 1velocity of 0.26ms'. A return flow of 0.1ms' is ob-3.2 Data vs the CTW Forcing Experimentserved at the bottom.In comparison，Fig.5b shows the surface velocityThe model predicted temperature and current am-remotely generated by the CT W forcing experiment onplitudes with 8d period are again shown in Tables 3day 8 and day 12 , when the outflow and inflow at theand 4. Both amplitudes of temperature and currentsbay entrance reach their maxima On day 8 , the shelfagree better with those of observations. Larger tem-currents flow southward , and an anti-cyclonic gyre isperature oscillations at southern stations of BI , PP andestablished in the bay. Four days later , the shelf cur-CW are simulated. The model has also predicted therents reverse their phases and the currents flow north-maximum temperature amplitude at 15m for station SC.ward on the shelf Two counter-rotating gyres arThe model predicted temperature phase lags be-formed in the bay , with a cyclone in the north and antween stations are similar to those from observations.A positive phase lags between NW and BI is predictedanti-cyclone in the south.To examine the model response to the wind forcingwhich is consistent with the observation. The predic-tion again shows large surface to bottom phase lags forat sub-inertial frequency，the amplitudes of the cross-shelf current and temperature oscillations with 8d pe-the currents( Table 4 ).iod are shown in Fig.6a for the cross-section at thebay entrance. The figure shows that the currents and3.3 Locally and Remotely Forced Currentstemperature reach a maximum value of 0.05 ms 1 andCurrents in Jervis Bay can be generated locally by 0.3 C respectively at the surface near Point Perpendic-the wind forcing ,or remotely by the CTW forcing onthe shelf Fig.5a shows the model predicted surface andFig.6b shows the amplitudes of the cross shelf sub-bottom currents on day 20 , driven locally by a south-inertial currents and the temperature at the bay en-westerly wind with a speed of 7ms'. Down wind cur-trance predicted by the CTW forcing experiment. Arents can be seen at the surface , reaching a maximum maximum valueof 0.1 ms' of the cross- shelf velocitySurface velocity (ms") Time-20.0 dSurface velocilty (msy) Time-8.0d。13130.26 ms'0.13 ms!x (km)Surface velocity (ms)Bottom velocity (ms") Time=20.0dTime=120d2318_13食”9418 22x(km)中国煤化工MHCNMHGFig.5( a) The wind driven surface and bottom velocities in the bay on day 20( A larger arrow indicates thedirection of the wind with its speed equaling to 7ms-' );( b)The CTW driven surface velocities in the bay万亦数捆and day 12 , when the outflow and inflow reached the maximum at the bay entranceWANG X.H. et al. : A Study of Water Circulation in a Thermally Stratified Embayment29PEPBU (cms")U (cms')t -200f0.2.00一一3.0ofContour interval-1.0Contour interval=1.00.015.050.0Tcmperature (C)Temperature (C)10t10一02030- 0.40mntour interval=0.1Contour interval-0.15.050150Distance (km)(间)Fig.6 The amplitudes of cross-shore velocity component and temperature of 8d period at the bay entrancefrom( a ) the wind driven model ;( b) the CTW forced modelcomponent is shown at the surface. Temperature am- the northern coast. A phase lag of about 100° existsplitude reaches 0.6C at 10 m water depth off Bowenbetween the north and south at a depth where theIsland. In comparison with the wind forcing experi-maximum temperature occurs. Unlike wind driven cur-ment , the CT W forcing has generated much strongerrents，this phase structure shows a wave propagatingsub-inertial flow at the bay entrance.' I he maximum from the south to the north.currents and temperature are twice as large as those inIn contrast，the cross- shelf currents predicted by theresponse to the wind forcing. Moreover ，they agree CTW forcing experiment are coastally trapped with awell with summer predictions by Middleton ( 1994 ) 60° phase lag between the northern and southern endswho estimated the maximum velocity and temperature of the bay( Fig.8 ). At the inflow phase the cross- shelftobe 0.1ms ! and 0.6C at the bay entrance due tocurrent has a negative component in the southern bay.the CTW scattering into the bay.Thus the phase of the currents there is computed to beTo further elucidate the difference in the responseof around 180 . The model predicts a maximum currentthe bay to the local and remote forcing ,the model preof 0.05 ms' near the centre of the southern bay，wheredicted amplitude of the cross shelf sub-inertial currentsthe current jet is located during the inflow phase of theand temperature along the section AB( Fig.1b ) are shelf flow( Fig.5b ).shown on Figs.7 and 8 from the wind and the CTWFig.8 shows two temperature maxima of 0.6C atforcing experiments. In response to the wind forcing，the bottom in the northern and southern parts of thecurrents are generated at the surface reaching a maxi->ay. The maximum phase delay of 100° between themum speed of 0.05ms' in the north. As the wind north and south is also predicted at the mid-depth bydriven current field is spatially uniform , there is littlethe中国煤化工amplitude and phase ofnorth to south phase lag for the cross-shelf current.Furthermore , the phase of the surface currents agrees the:HcNMHGillationat10mwaterwith the phase of cross-shelf wind component of 8d depth. As in Fig.7 , the wind forcing model predicts afrequency( 172° ). The phase of the cross-shelf cur-temperature maximum in the north of the bay( Fig9a ).rents at bottom has a lead of 180° , indicating a return-The temperature phase shows an increase from 10° ining flow there.. For the temperature , the wind driven the southern end of the bay to 160° at the northern endmodel preaiCsA糙maximum amplitude of 0.4C near of bay entrance.30Journal of Ocean University of Qingdao2003 ,Vol.2 ,No. 1。BU(cms)A。 B Phase(")^50.0 OO020|官20.0|0tCI= 0.5CI= 60.05007.515.050.0Distance (km)Distance (hm)Temperature (C)Phase(°).2一0.1@ 20-官20|CI=0.1CI- 60.00.01s.0Fig.7 The amplitudes and phases of cross shelf velocity component and temperature of 8d period at thecross-section AB( Fig.1b ) from the wind driven model ( CI abbreviates contour interval )B U (ems')3 Phase (")1010| 1500-2020 |340C1-0.5CI-60.0Temnperature (C)Phase (")03-10.0官20830|40 tCl-0.1中国煤化工50CNMHGDistannce (Km)Fig.8 Same as Fig.7 , but from the CTW forced modelWANG X.H. et al. : A Study of Water Circulation in a Thermally Stratified Embayment31Temperature (C) Depth=10.0 mPhase(") Depah=10.0 m2:18-18。135Cl=0.IC1- 30.0上(km)(田)23 rTemperature(C) Depth=10.0 m23Phase(") Depth=10.0 m原|。19F0.3-C1=0.1C1=30.049118 22x (km)[b)Fig.9 The simulated amplitude of the temperature osillation with 8d period from the wind forcedmodel ( a ) and the CTW forced model( b )( CI abbreviates contour interval )Simulations from the CTW forcing experimentthe shelf are weak , the IKWs in the bay are locally( Fig.9b ) shows that the amplitudes reach a minimum driven by the wind , and there is minimal interactionof0.3C at the centre of the bay and increase their between IKWs in the bay and on the shelf near thevalues towards the coast to about 0.5 C. A maximum bay entrance. A model run with a closed embaymenttemperature of 0.7 C is predicted at the southern endpredicts an identical temperature pattern to that shownof the bay. The phase increases from the entrance of in Fig.9a( figure not shown ).40° to about 160° in the northern bay. Furthermore ,In the CTW forcing experiment , the IKWs in thealong the northeast coast of the bay there is a de- bay are also generated. The waves again propagatecrease in phase from a maximumof 160° in the bay to around the bay in a clockwise direction ,and complete70° at the entrance ,so a small phase difference across their propagation in 3d. Due to the bottom friction ,the bay entrance is resulted.wave amplitudes are slightly dampened as they travelfrom the south to the north. This agrees with the ob-servation that temperature amplitude with 8d period4 Discussionreduces its value from a maximum of 0.9 C at station4.1 Internal Kelvin Waves in the BaySC to 0.6C at northern stations. Since the longshelfObservation of Figs.7 ,8 and 9 shows that the tem-wavelength of the shelf CTW ( which is basically anperature amplitude in the bay in response to both localIKW )is of an order of 500 km , phase change acrossand remote forcings is stronger near the coast than inbay e中国煤化工ation of the shelf IKWsthe centre of the bay. This shows that IKWs are gen-is nebay entrance is smallerated in the bay in both experiments.in ccMHCN MH Gof shelf IKM in bothIn the wind forcing experiment , the IKWs origi-longshelf and cross shelf directions , the shelf IKW cannate from the southern end of the bay , and propagate only be partially scattered into the bay( Fig.5b ) andaround the bay cyclonicly. Maximum amplitude is the energy of the unscattered IKW is still significant.reached in the, northern end and the IKWs exit the Hence , interaction of two groups of IKWs inside andbay 3d( 1305教振er their generation. As the IKWson outside of the bays is expected at the bay entrance. As32Journal of Ocean University of Qingdao2003 ,Vol.2 ,No. 1a result of this interaction，the temperature amplitudeis decreased from 0.7C to 0.4 C and phase is slightly4.2 Effect of Stratification on Inertial Flowincreased from 40° to 70° across the bay entrance. .In order to examine the effect of stratification onIn order to understand the mechanisms that gener-ate IKWs in the bay , the model experiments are con-the bay response to the wind forcing，the wind forcingducted using the forcing of steady northward andexperiment is repeated with a constant water tempera-southward shelf flows respectively. The velocity of thetureof 15C. The model predicts amplitude and phaseshelf flows is 0.2ms '. Fig.10 shows temperature dis-of the cross shelf currents along the cross section AB ,tributions along the cross-section CD after the flowand they are similar to those shown in Fig.7. Thisreaches a steady state for both experiments. A south-suggests that the wind driven currents at sub-inertialward shelf flow establishes an upwelling near the coast ,frequency is largely independent of the water stratifi-advecting cold shelf bottom water into the bay( Fig.10cation in the bay.a ). An anti-cyclonic coastal current similar to that4.3 Coastal Flow Separationshown in Fig.5b is established in the bay in response tothe upwelling event. In contrast ,a northward flow isThe southern anti-cyclonic gyre formed during theassociated with a downwelling event near the coastinflow phase of the shelf currents is a result of coastal( Fig.10b ). Correspondingly ,a cyclonic coastal flow is separation at the bay entrance. This process can bethen established in the bay. These experiments dem-found in some other coastal seas such as the Baie desonstrate that oscillating flow on the shelf is able to cre- Chaleurs in Canada( Gan et al. ,1997 ). Fig11 showsate alternating cross-shelf pressure gradient across thesurface vorticity fields calculated from the CTW forcedbay entrance. It is this oscillating cross- shelf pressuremodel surface velocity from day 10 to 13. Positivegradient that generates the IKWs in the bay.vorticity is generated on day 10 near BI when the in-Temperature (C) Time=10.0 dflow starts. As the flow increased its strength , the lo-cal vorticity is also intensified , and so is the non-lin28-earity of the flow there. Strong advection of the localvorticity downstream to the interior of the bay estab-56-lishes a vortex in the southern bay on day 13. The an-195-ti-cyclonic eddy observed there is the manifestation ofthe developing vortex during the inflow phase of theIKW intrusion.12-. In order to substantiate our arguments above，theCTW forcing experiment is repeated ，but the model140019.329.0omits the advection terms in the momentum equations.x (kxm)Fig.12 shows the surface vorticity from day 10 to 13.Temperaure(C) Time=10.0dDue to the absence of the non-linear term in the mo-mentum equations , no advection of local vorticity nearBI to the interior of the bay occurs. Therefore , no28 |vortex is established in the southern end of the bay ,and no flow separation is simulated there.84，5 Conclusions12The sub- inertial circulation of a temperature strati-C1=05fied Jervis Bay in response to the wind and CTW forc-140.0x (km)ings have been studied by the application of PrincetonOcean Model to the region.(b)In response to the wind forcing , the sub- inertial flowis weak and the correlation between the modelled cur-rents and the measurements is low. In addition to thewinc_are formed in the bay ,Fig.10 Model predicted temperature distributionpropa中国煤化工”south and exit the bayalong the croSS section CD3d l;HCNMH GThe model is fored by( a ) southward and( b)Under the CTw forcing , the model predicts a stron-northward steady shelf flow respectively( CI abbre-ger response. The amplitudes of currents and temper-viates contour interval ).ature at the sub-inertial frequency agree better withthose from the observation. As the CTWs can onlypartially scatter into the bay due to its narrow entrance ,WANG X.H. et al. : A Study of Water Circulation in a Thermally Stratified Embayment33Surface vrity (I/S*E-5) Time=10.0dSurface vorieity (1/S*E-5) Time=110d23厂。18CI=1.0.C1-108 2218 22x (km)，Surface voricity (S*F-5)_ Time=12.0dSurface vortieity(1/S*F-5) Time- 13.0dCI- 10C=1.0Fig.11 The surface vorticity field onday 10 ,11 ,12 and 13 predicted by the CTW forced model，during the inflow phase of CTW intrusionA vortex that was associated with and anti-cyclonic eddy was developed on day 13 in southern bay( CI abbreviates contour interval ).，Surace vricity(1/S*E-5) Time-10.0dSurface vorticity (I/S*E-S) 'Time - 10d1868县CI=1.0 .CI-1.018 2Surface vorticity (1/S*E-5) Time=12.0d .Surface vorticity (1/S*E-5) Time=13.0 d2:CI- 1.00418 23中国煤化工YHCNMHGFig.12 The surface vorticity field on day 10 ,11 ,12 and 13 predicted by a linear CTW forced mod-el , during the inflow phase of CTW intrusionrtex that was asosited with and ani-cvclonie edy was dereloped on day 13 in southem bayabbreviates contour interval ).34Journal of Ocean University of Qingdao .2003 ,Vol.2 ,No. 1interaction between two groups of IKWs inside andDimensional Models of Marine and Estuarine Dynamics. J.outside of the bay take place there. As a result of thisC. J. Nihoul ,and B. M. Jamart , eds. , Elsevier Oceanog-raphy Series ,45 :55 -88.interaction，there is a decrease in wave phase alongChurch ,J A. ,H. J. , Freeland , and R. L. , Smith , 1986 a.the northeast coast of the bay in a clockwise direction ,Coastal-trapped waves on the east Australian continentaland the phase lag across the bay entrance is small.shelf. Part I : propagation of modes. Journal of PhysicalThis agrees with a zero phase lag from station BI tcOceanography , 16 : 1929-1943.PP in the observation.Church ,J. A. ,H. J. , Freeland ,and R. L. , Smith , 1986 b.This study has shown that the sub -inertial baroclin-ic flow observed in Jervis Bay is dominantly driven byshelf. Part I : model verification. Journal of Physicalthe remote CTW on the adjacent shelf. The responseOceanography , 16 : 1945 - - 1957.of the bay to this CTW forcing is the formationof IK-Craig,P. D. ,and P. E. Holloway , 1991. The influence ofcoastally trapped waves on the circulation in Jervis Bay ,Ws in the bay. The flow associated with the IKWs inNew South Wales. In :Dynamics and Exchanges in Estuar-the bay is strongly non-linear due to a process of coast-ies and the Coastal Zone. D. Prandle ,ed. , Springer Verlag ,al flow separation that occurs at the bay entrance dur-747 pp.ing the inflow phase of the coastal flow on the shelf.Gan ,J.,R. G. Ingram ,and R. J. Greatbatch , 1997. Onthe unsteady separation/intrusion of the Gaspe Currentand variability in Baie des Chaleurs : modelling studies.AckowledgementsJournal of Geophysical Research , 102 : 15 567- 15 581.Holloway ,P. E., G. Symonds , and R. Nunes Vaz , 1992.Observations of circulation and exchange processes in JervisAuthors wish to thank Associate Professor PeterBay ,New South Wales. Australian Journal of Marine andHolloway and Dr Peter Craig for introducing XHW tcFreshwater Research ,43 : 1487 -1515.the topic of Jervis Bay research. Their commentson Mellor ,G. L. ,and T Yamada , 1982. Development of athe manuscript are constructive and greatly appreciat-turbulence closure model for geophysical fluid problems.ed.Rev. Geophys. Space Physics , 20( 4 ):851 -875.Mellor ,G. L. , 1992. User' s guide for a three-dimensional ,primitive equation numerical ocean model. Report :Pro-Referencesgram in Amos. and Ocean. Sci. Princeton University , Prin-ceton ,NJ 08544 ,35 pp.Blumberg ,A. F. ,and G. L. Mellor , 1983. Diagnostic andMiddleton ,J，F. , 1994. The baroclinic response of straitsand bays to coastal- trapped wave scattering. Journal ofprognostic numerical circulation studies of the South At-Physical Oceanography ,24 :521 -539.lantic Bight. Journal of Geophysics Research , 88 : 4579 -Oey ,L.-Y. and Chen, P., 1992. A model simulation o4592.Blumberg ,A. F. ,and G. L. Mellor , 1987. A description ofcirculation in the northeast Atlantic shelves and seas.Journal of Geoplhysical Research , 97 : 20087 -20115.a three-dimensional coastal ocean circulation model. Three中国煤化工MHCNMHG.