Direct Observations of Velocity and Transport in the Passages Between the Intra-Americas Sea and the Atlantic Ocean, 1984 -1996

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 104, NO. Cll, PAGES 25,805-25,820, NOVEMBER 15, 1999

Direct observations of velocity and transport in the passages between the Intra-Americas Sea and the Atlantic Ocean, 1984 -1996

Elizabeth Johns,W. DouglasWilson, and Robert L. Molinari Atlantic Oceanographicand MeteorologicalLaboratory, National Oceanicand AtmosphericAdministration, Miami, Florida

Abstract. Shipboardacoustic Doppler currentprofiler observationsof the velocityin the upper 200 rn of the water columncollected during 1984-1996 usingthe National Oceanic and AtmosphericAdministration R/V MalcolmBaldrige are usedto examinethe velocity structureand transportin the passagesbetween the Atlantic Ocean and the Intra- AmericasSea (IAS). Data were collectedduring 23 cruisesalong the followingsections: acrossthe Straitsof Florida, in the NorthwestProvidence Channel (NWPC), acrossthe northernpassages into the CaribbeanSea (Windward,Mona, and Anegada),across the easternCaribbean along 63030 ' W, therebyforming a closedquadrangle, and in the GrenadaPassage. The Florida Current, the easternCaribbean, and the Grenada Passage sharea similarmean velocitystructure characterized by high-velocity,surface-intensified flowswith strongvertical and horizontalshears. The northern Caribbeanpassages (NWPC, Windward,Mona, and Anegada)share a differentcommon mean velocity structure,with subsurfacevelocity maxima directed into the IAS, and surface-intensified counterflowsalong one side of each passage.On average,there is a transportbalance in the upper200 rn betweenwaters entering and exitingthe IAS, with the 16.5 _+2.4 Sv (1 Sv = 106m3/s) transport of theFlorida Current at 27øNcomprised of 0.4 _+0.8 Sv from the NWPC, 2.2 _ 1.5 Sv from the Windward Passage,2.8 _ 2.1 and 2.4 _ 2.8 Sv from the Mona and Anegadapassages, respectively, and 9.5 +_4.7 Sv acrossthe easternCaribbean, for a total of 17.3 Sv. The four passagesnorth of 17øN(from NWPC to AnegadaPassage) have a combinedtransport of 7.8 Sv, nearlyhalf of the transportof the Florida Current in the upper200 m. Of the 9.5 Sv flowingthrough the easternCaribbean between lløN and 17øN,4.9 _+2.6 Sv, or more than half, come from the Grenada Passage.This is significant to the subjectof cross-equatorialexchange of mass,heat, and salt, as the GrenadaPassage is where the highesttransport of watersoriginating in the southernhemisphere is thought to enter the Caribbean.

1. Introduction Caribbean and into the southeastern corner of the Gulf of Mexico and eventuallyexit via the Florida Current. The wind- The Intra-AmericasSea (IAS) is the semienclosedbody of driven circulationof the Caribbeanis dominatedby the Trade water comprisedof the CaribbeanSea, the Gulf of Mexico,and Winds, which have a strong seasonalcomponent and a wind the Straits of Florida, connected to the North Atlantic Ocean stresscurl field that respondsto the seasonalnorthward and by a numberof passagesbetween the islandsof the Bahamas, southwardmigration of the Intertropical ConvergenceZone Greater Antilles, and LesserAntilles (Figure 1). The circula- [Mayerand Weisburg,1993]. tion of the IAS is dominatedby the Caribbeanand Florida In addition to its strictly physicaloceanographic interest, currents.These currents form a major componentof the North attention has recentlybeen givento the interconnectednature Atlantic subtropicalgyre, leading directly into the Gulf Stream, of the marine and atmosphericenvironments of the IAS in and as suchare an importantconduit for mass,heat, and salt terms of pollutant dispersal,larval transports,the health of fluxes in the Atlantic. coral reefs,and commercialand recreationalfisheries [Roberts, The IAS is a fascinatingregion from a physicaloceano- 1997; Ogden,1997; Mooersand Maul, 1998]. It is becoming graphicviewpoint, with variabilityon all timescalesfrom syn- apparentthat only throughthe useof focusednumerical mod- opticto seasonal,interannual, and decadal,driven by wind and els of the circulationwhich can properly simulatethe atmo- thermohalineforcing. The deep circulationis dominatedby sphericand oceanicforcing and the ocean'sresponse on all sporadicinflow eventsover the sillsof the major Caribbean relevantspace and timescaleswill the modelsbe able to move passages[Sturges, 1970], while the upper layer circulationis a toward the nowcast/forecastcapability that could be used to throughflowsystem in whichinflows from the passages,includ- address the environmental and societal concerns critical to the ing recirculatingNorth Atlantic subtropicalgyre waters as well IAS. as watersoriginating in the SouthAtlantic, passthrough the However,the developmentof effectivenumerical circulation Copyright1999 by the AmericanGeophysical Union. modelsrequires high-quality and abundantobservational data Paper number 1999JC900235. for modelverification. For someparts of the IAS, suchas the 0148-0227/99/1999JC900235509.00 Florida Current, adequatedata exist for this purpose.The

25,805 25,806 JOHNS ET AL.: VELOCITY AND TRANSPORT IN THE IAS PASSAGES

estimateof the transportdistribution based on the limited data availableat that time. More recently,Wilson and Johns[1997] conducteda comprehensivestudy of the inflow to the eastern 2826 _:j•GUlfof Mexico •FC:,•"•-•, Atlantic Ocean Caribbean during the Windward Islands PassagesProgram. Using a cable-lowered acoustic Doppler current profiler uJ 22 (ADCP), they were able to obtain enoughdirect observations • 20 • • MP L...... of transportand velocitystructure in the southernpassages over several years to more closely examine the mean inflow • 16 and its variability. Caribbean Sea 14 The northern passagesto the IAS (Windward, Mona, and Anegada;Figure 1) are the least studied,and few direct mea- 12 surementsof their transporthave been obtained. Thus a key 10 motivationfor the presentstudy was to utilize direct velocity data collectedon a series of NOAA cruisesboth passing 90 88 86 84 82 80 78 76 74 72 70 68 66 64 62 60 58 throughand directed at the region between 1984 and 1996 to LONGITUDE addressthis deficiency.Direct velocity observationswere col- Figure 1. Bottom topography of the Intra-Americas Sea. lectedusing a hull-mountedADCP along a number of sections Transect locations are indicated as follows: FC, Florida Cur- acrossthe passagesof the IAS, in the easternCaribbean Sea, rent; NW, Northwest Providence Channel; WP, Windward and in the Straitsof Florida (Figure 1). Passage;GI, Great Inagua Passage;MP, Mona Passage;AP, Herein, the ADCP velocitydata from thesecruises are used Anegada Passage;CA, eastern Caribbean; and GP, Grenada to examinethe velocitystructure in the upper 200 m of the Passage. OBC indicates the location of the Old Bahama Channel. water column and to estimatetransports through the Carib- bean Sea and the Straitsof Florida. We beginwith a descrip- tion of the available observationsand data reduction, followed by an analysisof the velocityand transportresults for each Florida Current is probably the most studied current in the passage,and a discussionof their significanceboth compared world, from the pioneeringphysical oceanographic study of to earlier transportestimates and in terms of addressingsome Pillsbury[1890] which utilized the first current meter to mea- of the issues mentioned above. sure the transportof the Florida Current, to more sophisti- cated studiessuch as that by Richardsonand Schmitz[1965] and Schmitzand Richardson[1968] using dropsondes.More 2. Data and Methodology recently, NOAA's SubtropicalAtlantic Climate Study [Moli- The eight sectionsconsidered herein are located in the nari et al., 1985]used a combinationof shipboardobservations, Straits of Florida at 27øN and 26øN: the Northwest Providence moored current meters and pressuregauges, and a submarine Channel (NWPC), the Great Inagua Passage(between the cable to examine the velocity structureand variability of the Bahamianislands of Great Inaguaand Hispaniola),the Mona Florida Current at 27øNand to continuouslymonitor the trans- Passage,the AnegadaPassage, the easternCaribbean Sea, and port usingthe cable[Larsen and Sanford,1985]. The Subtrop- the Grenada Passage(Figure 1). Ideally, for estimatingthe ical Atlantic Climate Studywas begun in 1982, and the trans- transportbalance of the IAS, the Great Inagua sectionwould port monitoringcontinues to the presenttime. Resultsfrom insteadhave been located acrossthe Windward Passage,be- this studyinclude determination of a stablemean for the trans- tweenCuba and Haiti. However,this was not possiblebecause port of 32.3 +_3.2 Sv [Larsen,1992] and a better understanding of clearanceconstraints. Thus flow throughthe Great Inagua of its seasonaland interannualvariability [see, e.g., Molinari et sectionis comprisedof flow into or out of the CaribbeanSea al., 1985;Lee et al., 1985;Schott and Zantopp, 1985;Johns and through the Windward Passageand flow into or out of the Schott,1987; Larsen, 1992;Leaman et al., 1987]. Straitsof Florida throughthe Old BahamasChannel and can Suchis not the casefor the other passagesof the IAS and for only be used as a proxy for flow through the Windward Pas- the Caribbean Sea as a whole. Although the Caribbean has sage.The datesof the 23 cruisesused in this analysisand the been the subjectof a number of hydrographicstudies, few sectionsoccupied and transportsobtained during each passage direct observations of the circulation have been made. Earlier transect are listed in Table 1. historicalstudies used hydrographic and tracerdata to describe ShipboardADCP observationsare the primary data source the circulationusing either the "core"method [Wust, 1964] or for thisstudy. A shipboardADCP unit providesnearly contin- geostrophiccalculations [Gordon, 1967], or in the case of a uous coverageof upper ocean currents.The systemused for study by Roeromich[1981] using a combinationof the two most of the cruisesaboard the NOAA R/V Malcolm Baldrige techniquesin an inverse calculation. Studies using surface was a 150-kHz RD Instruments unit with a hull-mounted trans- driftersin the Caribbeanhave also been conducted [Molinari et ducer.Earlier ADCP units(an Ametek Straza115-kHz instru- al., 1981], revealing an ocean environmentthat is rich with mentwith the samehull configuration)were limited to -200 m eddies superimposedonto a mean throughflow.A review of in verticalcoverage, although more recentcruises yielded data the state of understandingof the Caribbean circulationas of down to a depth of 350-400 m. Resultsare thereforepre- the 1980sis givenby Kinder et al. [1985]. sentedonly for the upper 200 m of the water columnin order The few direct velocity and transport observationswhich to useall of the cruisedata consistently.Standard data acqui- have been made in the IAS passagesinclude the study of sitionand processingprocedures are describedby Wilsonand Schmitzand Richardson[1991]. They used data collected in Routt [1992], and additionaldetails about the ADCP system, 1970 by Stalcupand Metcalf [1972] to examinethe flow be- includinga discussionof navigationalreferencing and error tween the islands of the Lesser Antilles and arrived at an estimates,are describedby Wilsonet al. [1994].After referenc- JOHNS ET AL.' VELOCITY AND TRANSPORT IN THE IAS PASSAGES 25,807

Table 1. Cruise Dates, PassageOccupation and Transport

Cruise Dates F27 F26 NW GI MP AP CA GP

Aug. 27 to Sept. 6, 1984 ...... 3.0 ...... April 17 to May 17, 1985 18.8 ...... 3.0 ...... Aug. 13 to Sept. 6, 1985 13.1 ...... 1.0 ...... 14.2 --. Jan. 13 to Feb. 11, 1986 13.6 ...... 3.1 ...... 9.9 ... March 25 to April 22, 1986 ...... 1.4 ...... 9.7 ... July 15 to Aug. 8, 1986 18.3 ...... 0.5 .... 4.9 -11.8 ... Oct. 23 to Nov. 7, 1986 19.3 ...... 1.5 .... 8.4 -1.7 ... Oct. 23 to Nov. 7, 1986' 18.1 ...... March 11 to March 23, 1987 13.8 14.2 ...... Sept. 1 to Oct. 1, 1987 ...... Aug. 22 to Sept. 22, 1989 ...... 0.5 -5.2 -3.0 -2.4 ...... June 15 to July 11, 1990 15.5 ...... Sept. 7 to Oct. 10, 1990 .-. 14.5 -1.7 --. 0.1 ...... Jan. 8 to Feb. 6, 1991 ...... 1.4 -5.0 ...... 10.3 June 12 to July 5, 1991 ...... 4.0 Sept.8 to Sept.23, 1991 ...... 0.9 .... 6.5 Jan. 31 to Feb. 10, 1992 ... 13.7 0.4 ...... Aug. 3 to Sept. 1, 1992 ...... 0.6 ...... Aug. 3 to Sept. 1, 1992' ...... 0.5 ...... June 1 to June 29, 1993 ... 15.8 0.2 ...... 2.2 .... 3.6 Sept. 16 to Sept. 22, 1993 ... 16.3 -1.6 ...... March 28 to April 8, 1994 ... 17.6 -0.1 ...... 2.2 .... 1.6 July 12 to Aug. 1, 1994 -.. 15.9 0.1 ...... 0.3 .... 4.1 March 3 to March 20, 1996 18.2 ...... 4.5 -0.8 .... 5.2 July 15 to Aug. 2, 1996 16.4 .... 1.0 .... 1.7 -1.5 .... 3.8

Mean 16.5 15.4 -0.4 -2.2 -2.8 -2.4 -9.5 -4.9 Standard deviation +2.4 +- 1.4 +0.8 _+1.5 +2.1 +_2.8 _+4.7 _+2.6 Standard error _+0.8 _+0.5 _+0.3 +_0.5 _+0.9 _+0.9 _+2.1 +_0.9

F27, Florida Current at 27øN; F26, Florida Current at 26øN; NW, Northwest ProvidenceChannel; GI, Great Inagua Passage;MP, Mona Passage;AP, AnegadaPassage; CA, EasternCaribbean along 63.5øW; GP, GrenadaPassage. Transport measured in Sverdrups. *More than one occupationof transectwas made during cruise.

ing, 15-minaverages are computedand form the basisfor the 3.1. Florida Current at 27øN velocityresults presented below. The 27øN Florida Current transect is located between Palm Transportsfor the passageswere computedusing the fol- lowing methodology.Velocity data from individual cruises Beach, Florida, and Little Bahama Bank (Figure 2). This were projected onto standard (i.e., most often occupied) transectwas occupiedten times betweenApril 1985 and July transectlocations (requiring in most casesonly minor adjust- 1996 (Table 1). The averagesurface and 0-200 m velocities ments in position) and the perpendicularand along-transect shownin the vector mapsshow the asymmetricalnature of the componentsof velocitywere computed.This velocitywas grid- Florida Current, with a sharprise to the velocitymaximum on ded in 2-km bins of along-transectdistance and in 10-m depth the western side of the Straits, and a more gradual decrease bins,and the perpendicularvelocity component was then inte- acrossthe middle and easternside of the channel(Figure 2). gratedhorizontally and verticallyto producetransport in units The vertical sectionof northwardvelocity also showsthe asym- of Sverdrups(Sv, where 1 Sv: 106 m3/s).Average transports metry of the velocity structure,with the maximum velocity and standarddeviations were obtained by summingthe indi- found farther offshorewith depth. The surfacevelocity reaches vidual gridded transports and dividing by the number of a maximumof over 160 cm/s ---20km east of the sectionorigin transectoccupations. No attemptwas made to removethe tidal (Figure 2). The standarddeviation of the northwardflow is variabilityfrom the observations.However, given enoughoc- highest(_+50 cm/s) near the surfaceon the westernside of the cupationsfor each transect,this variabilityshould tend to av- current(---12 km from the origin) and is due to lateral mean- erage out. dering of the current core as well as variability of the core velocity.The larger standarddeviations on the easternside of 3. Results the Straitsare in part attributableto variable southwardflows The resultswill be presentedby passagefrom northwestto also found in an earlier study [Leaman and Molinari, 1987]. southeastacross the IAS in the form of vector maps of the Both the eastwardvelocity section and the average velocity surfaceand upper 200 m averageflow through each passage vector maps of Figure 2 show that the flow is convergingat superimposedupon the bottom topography,and vertical sec- 27øN;i.e., on average,there is a weak eastwardcomponent at tions of the mean and standard deviation of the northward and the onshoreside of the transect,with a stronger(5-10 cm/s) eastwardvelocity componentsfor each passage.Transports westwardcomponent on the offshoreside. Standard deviations perpendicularto the passagesare presentedfor all passage of the eastwardflow componentare in the _+5to 15 cm/srange, occupationsin Table 1, which includesthe mean, standard highest near the eastern side of the transect in the upper deviation,and standarderror for each passage. 50-100 m. The averagetransport at 27øN in the upper 200 rn 25,808 JOHNS ET AL.: VELOCITY AND TRANSPORT IN THE IAS PASSAGES

SurfaceVelocity Average0-200 m Velocity

24' 02 s• so 7'9 7i• '7• 70 82 $'• •0 79 70 77 7'0 LONGITUDE LONGITUDE

Northwardvelocity (cm/s) std dev 50- , /,// / 5o0

200--t:sh•h•,•x• ,t ,L ; ,,'••, 200 0 20 40 60 80 0 20 40 60 80

std dev 0 Eastwardvelocity (cm/s) 0

50 g5o 100 150 150

200 200 0 20 4O 60 80 20 40 60 80 Distance(km) Distance (km) Figure 2. The Florida Current at 27øN:Acoustic Doppler current profiler (ADCP) vector maps of the averagesurface velocity (left top) andthe average0-200 m velocity(right top). The velocityvectors are scaled suchthat 50 cm/s equals1 ø of latitude. Also shownare vertical sectionsof ADCP velocityin cm/sfor the northwardand eastwardvelocity components (left bottom)and their standarddeviations also in cm/s(right bottom). The orientationof the verticalsections is west to eastalong the transect.

from the 10 sectionoccupations is 16.5 _+2.4 Sv, with a range 200 m. The standard deviation of the northward velocity of 13.1Sv in August1985 to 19.3 Sv in October1986 (Table 1). reachesa maximumof _+25cm/s at a depthof 100 m about200 km along the section,near the high horizontal and vertical 3.2. Florida Current at 26øN shear zone of the onshoreedge of the current. The standard The 26øN Florida Current transect, located between Miami, deviationis also high (_+25 cm/s) near the surfaceon the Florida, and Bimini, Bahamas(Figure 3), was occupiedseven offshore side of the channel. The standard deviations of the timesbetween March 1987 and July 1994 (Table 1). The sur- northwardflow are generallysmaller at 26øN than at 27øN, face and 0-200 m averagevelocity vectors (Figure 3) are sim- indicatingthat the velocitystructure of the Florida Current at ilar in appearanceto the equivalentvectors at 27øN(Figure 2) 26øNis more consistentfrom one cruiseto the next.This may but showthat whereasat 27øNthe currentwas directed slightly be due to the local bottom topography,which constrainsthe west of due north, at 26øN the averageflow is directed at a current to a relativelynarrower channel at 26øN (Figure 3). bearingof ---10ø eastward of due north. The northwardvelocity The verticalsection of eastwardvelocity shows that at 26øNthe structure(Figure 3) showsa maximumsurface velocity mid- flow hasan eastwardcomponent rather than westward,as also channelof over 170 cm/s.A weak counterflowis presentbelow observedfrom the vectormaps (Figure 3). On the westernside 100 m on the westernedge of the section,at a depth of 100- of the channel, there is an area of westward flow at 50-200 m JOHNS ET AL.' VELOCITY AND TRANSPORT IN THE IAS PASSAGES 25,809

AverageSurface Velocity Average0-200 rn Velocity

29-

I--- 27

81 80 79 78 77 76 82 81 80 79 78 77 76 LONGITUDE LONGITUDE

Northwardvelocity (cm/s) std dev 0

E 50- 50 .c 100- 100

c3 150-

2OO 200 • • 0 20 40 60 80 0 20 40 60 80 Eastwardvelocity (cm/s) std dev

E 5O- 50 •_•.o .c 100- 100 • •ø'"--9

C3 150- 150

2OO 200 • , 0 20 40 60 80 0 20 40 60 80 Distance (km) Distance (km) Figure 3. Sameas Figure 2, but for the Florida Current at 26øN.

depth;i.e., the flowis divergenton averageat 26øNas opposed The velocitystructure is highlyvariable from cruiseto cruise, to the convergentflow at 27øN,with higherstandard deviations as evidencedby the standarddeviations which are generally at 26øN,particularly within the westwardvelocity area. The _+10 to 20 cm/s,i.e., higherthan the mean currents.The aver- averagetransport of the seven26øN transect occupations is age transportthrough the NWPC from the 10 transectoccu- 15.5 _+ 1.4 Sv, with a low of 13.7 Sv observedduring January pationsis -0.4 + 0.8 Sv (negativetransport is directedto the 1992and a highof 17.6Sv duringMarch 1994(Table 1). west), rangingfrom a maximuminflow toward the Florida Currentof -1.7 Sv duringSeptember 1990 to a minimumof 3.3. Northwest Providence Channel +0.5 Sv (eastwardflow, i.e. directedaway from the Florida The NorthwestProvidence Channel (NWPC) is locatedto Current) duringAugust 1992 (Table 1). the east of the Florida Current between Grand Bahama Island and the Great Bahama Bank (Figure 4). This transectwas 3.4. Great Inagua Passage occupied10 timesbetween August 1989 and July 1996 (Table The Great InaguaPassage is locatedbetween the Bahamian 1). The surfaceand 0-200 m averagevector maps show that islandof Great Inagua and the northwesterncoast of Haiti the flow is generallyto the weston average,with the strongest (Figure5). Here the prevailingcurrents are generallyto the westward flow found at the northern side of the channel, and southwest,toward the Caribbean,with the majority of trans- with a surface-intensified current reversal to the east located at port mostlikely flowingdirectly into the Caribbeanthrough the southernedge of the channelalong the BahamaBank the WindwardPassage between Cuba and Haiti. Thus we as- (Figure4). The verticalsection of averagevelocity (Figure 4) sumethat this passagecan be consideredas a proxy for the shows that there is a subsurface maximum in the westward flow throughthe WindwardPassage. Any westwardflow that velocity,with speedsgreater than 30 cm/slocated below 150 m. doesnot enter the WindwardPassage would insteadeventually 25,810 JOHNS ET AL.: VELOCITY AND TRANSPORT IN THE IAS PASSAGES

AverageSurface Velocity Average0-200 m Velocity

25-

24 81 80 79 78 81 80 79 78 LONGITUDE LONGITUDE

Northwardvelocity (cm/s) std dev '•'•"•...... ::':•' ' •.•"•' •.•...... ß...... '""""i

E 50- •-• .c 100- 100 cl 150-

2OO 200150 0 20 40 60 80 0 20 40 80 Eastwardvelocity (cm/s) std dev o E 50- 50 -: 100- 100

r• 150-

200150 2OO 0 20 40 60 80 0 20 40 60 80 Distance (kin) Distance(kin)

Figure 4. Same as Figure 2, but for the NorthwestProvidence Channel. The orientationof the vertical sectionsis from northeastto southwestalong the transect. join the Florida Currentvia the Old BahamaChannel (Figure to the south.The averagetransport through the Great Inagua 1). Transportalong this route has been estimatedat 1-2 Sv Passagein the upper 200 m is -2.2 _+1.5 Sv (negativetrans- [Atkinsonet al., 1995] and is highlyvariable. port is directedto the southwest,i.e., towardthe Caribbean), The Great Inagua Passagetransect was occupieda total of with a low of -0.5 observedduring July 1986 and a high of nine timesbetween August 1984 and January1991 (Table 1). -5.2 Sv duringAugust 1989 (Table 1). The averagevelocity was to the southwest(Figure 5), with a subsurfacewestward maximum of greaterthan 20 cm/slocated 3.5. Mona Passage midtransectbelow 100 m depth. On the south side of the The Mona Passageis locatedjust west of Puerto Rico, be- channel(along the coastof Haiti) the flow is reversedand tween Puerto Rico and the east coast of the Dominican Re- surfaceintensified, with an averagecounterflow out of the public,and was occupiedfive timesbetween August 1989 and Caribbeanof 10-15 cm/sin the upper 100 m (Figure5). Dur- July 1996. The average0-200 m flow is predominantlysouth- ing four of the nine crossingsthe countercurrentwas not ward, i.e., into the Caribbean,through this passage (Figure 6). presentalong the Haitian coast.During the other five cruises A shallow shoal that extends eastward from the northeast cor- where it was present,maximum velocities were 30-40 cm/s, ner of the Dominican Republic seemsto partially block the and greaterthan 60 cm/sduring July 1986 (not shown).The flow, causingthe current to split into two bands(Figure 6). standarddeviation of the eastwardvelocity component is less Similarly to the Great Inagua Passageand the NWPC, the than _+10-15 cm/s over the northern half of the transect, and southwardinflow showsa subsurfacemaximum, greater than higher(_+20-25 cm/s)over the southernhalf. The northward 20 cm/sat 200-m depth in midchannel.In addition,there is a velocitycomponent is weak (<5 cm/s)and directedprimarily persistentcountercurrent (i.e., out of the Caribbean)located JOHNS ET AL.: VELOCITY AND TRANSPORT IN THE IAS PASSAGES 25,811

AverageSurface Velocity Average0-200 rn Velocity

75 74 73 72 75 74 73 72 LONGITUDE LONGITUDE

Northwardvelocity (cm/s) std dev o

50- 100 150

200 ...... i...... 200 • 40 80 120 0 40 8'0 120 Eastwardvelocity (cm/s) std dev

50- 50 100- 1O0o 150- 150 200 200 0 40 80 120 0 40 80 120 Distance (km) Distance (km)

Figure 5. Sameas Figure 2, but for the Great InaguaPassage. The orientationof the verticalsections is from northwestto southeastalong the transect.

along the coastof Puerto Rico with averagespeeds over 10 tenseflow foundat the westside of the transectnear the Virgin cm/s,maximum at the surface(Figure 6). Cruise-to-cruisevari- Islands,where there is an averagemaximum southeastward ability is very high in this passage,with the standarddeviation speedof over 60 cm/s (Figure 7). This high averagespeed is 1,-,.-,•,-,1.. ,-,,-, .... ,4 1-, ..... + .... 1.. 1....:'•1.....•1•.'.•'• •t.. .-,1.... :_ • 111111 VClU•ItlCb UUbCl UUllll• highor higherthan the averagevelocity. The averagetransport transectoccupations, July 1986and October1986, when speeds throughthe Mona Passagein the upper200 m is -2.8 _+2.1 Sv were observed in excess of 110 cm/s. The vertical section shows directed into the Caribbean,with a high of -5.0 Sv observed a subsurfacesouthward velocity maximum of 10-15 cm/sin the duringJanuary 1991 and a low of +0.1 Sv directedout of the center of the channel and a very weak flow reversal at the Caribbeanobserved during September 1990 (Table 1). surface.Standard deviations are generallyin the z 10-20 cm/s range,with highervalues in the high-velocityregion near the 3.6. Anegada Passage Virgin Islands. The flow is strongly convergent over this The AnegadaPassage is locatedat the northeastcorner of transect,with an eastwardcomponent over the westernhalf of the Caribbean, between the Virgin Islands and Saba Bank the transectand a westwardcomponent over the easternside (Figure 7). Two standardtransects were occupiedacross this and with a tendencyalong the easternside to be alignedalong passage,with somewhatdifferent transectlengths and angles, the bottom topography(Figure 7). The averagetransport in but both generallycrossing the entire passage. the upper 200 m is -3.4 m 3.5 Sv directedinto the Caribbean, The first Anegada Passagetransect was occupiedfive times with a low of +0.9 Sv (outflowfrom the Caribbean)observed betweenJuly 1986 and June 1993 (Table 1). The flow is pre- during September1991 and a high of -8.4 Sv during October dominantlysouthward into the Caribbean,with the most in- 1986,indicating a very large transportrange (Table 1). 25,812 JOHNS ET AL.: VELOCITY AND TRANSPORT IN THE IAS PASSAGES

Average SurfaceVelocity Average 0-200 rn Velocity

6•8 67 66 69 08 6• LONGITUDE LONGITUDE

Northwardvelocity (cm/s) std dev o

E 50- 50 .c 100- lOO

c) 15o- 15o

2OO 200 40 80 120 o 40 80 120 Eastwardvelocity (cm/s) std dev _ o

E 50- 50- .c 100 lOO

C3 150- 15o

2OO 200 0 40 80 120 o 40 80 120 Distance (km) Distance (km)

Figure 6. Same as Figure 2, but for the Mona Passage.

The secondof the two Anegada Passagetransects was oc- transectwere made between August 1985 and October 1986 cupiedfour times betweenMarch 1994 and July 1996 (Table (Table 1). Thesedata havebeen discussedby Smithand Mor- 1). The flow was again southward,into the Caribbean,and /ison [1989] and Morrisonand Smith [1990].The averageflow weaker than the first Anegada transect,with counterflowsout is predominantlyto the west (Figure 9). Maximum average of the Caribbeanon the east side along SabaBank and on the surfacespeeds of up to 80 cm/s directed to the northwestare west side near the Virgin Islands(Figure 8). The verticalsec- found at the southern end of the transect, where most of the tion againshows a subsurfacemaximum midchannel. Standard flow originatesfrom the Grenada Passage.The averageflow deviationsare equivalentto the averagespeed, _+5-15 cm/s. hasthree major westwardbranches, at 11.5ø-13øN,13.5ø-15øN, The averagetransport in the upper 200 m is -1.2 _+0.8 Sv into and 16.5ø-17øN,with counterflowsbetween the branches(Fig- the Caribbeanthrough this transect(Table 1), and it was al- ure 9). Counterflowswere presenton each individualcruise ways directed into the Caribbean with a range of -0.3 Sv transect(not shown)but varied in latitude and number.The during July 1994 to -2.2 Sv during March 1994. If the values verticalsection of eastwardvelocity (Figure 9) showsthat these for the two Anegada Passagetransects are merged for trans- branchesof flow are all surface-intensified.The averagetransport computations,the averageupper 200 m transportfor the port above200 m for thesefive sectionsis -9.5 + 4.7 Sv to the nine occupationsis -2.4 _+2.8 Svinto the Caribbean(Table 1). west, with a minimum value of -1.7 Sv observed in October 1986 and a maximumof -14.2 Sv in August 1985 (Table 1). 3.7. Eastern Caribbean The eastern Caribbean sectionis oriented along the Aves 3.8. Grenada Passage Ridge at 63.5øW,from Saba Bank on the northern end to the Two different transect locations were used for the Grenada Venezuelanshelf on the southernend. Five occupationsof this Passage.The first runs from Grenada southeastto near the JOHNS ET AL.: VELOCITY AND TRANSPORT IN THE IAS PASSAGES 25,813

Average SurfaceVelocity Average 0-200 m Velocity

66 65 64 63 66 65 64 63

LONGITUDE LONGITUDE

Northwardvelocity (cm/s) std dev 50 50 100o " o, ..• 100ot ltt,!l (•-; 150 150 200 • • • • • • 200 , , • • • • 0 40 80 120 160 200 0 40 80 120 160 200 Eastwardvelocity (cm/s) std dev

E 50- 50 • ? •o, ' .c: 100- 0tl C3 150- 150 200 I 200 0 80 120 160 200 0 40 8'0 120 160 200 Distance (km) Distance (km)

Figure 7. Sameas Figure 2, but for the first AnegadaPassage transect.

coastof Tobago(Figure 10). It was occupiedthree timesbe- July1996 (Table 1). The averagevelocity is slightlysouth of due tween Januaryand September1991 (Table 1). The average west,reaching a maximumin the centerof thepassage of over90 0-200 m velocityvectors show strongest inflow to the Carib- cm/s(Figure 11). The verticalsection of eastwardvelocity shows bean on the Grenada side, with northwestward flow near To- that the highestvelocity is locatedin the centerof the passage, 1__ • [1'-?• ...... t/-i% P'-•l ...... 1 '• i _ ll ß viago !,ragutc .iv)ß tHu awl-ag• v•lucity sections5HOW LHHL [he directedinto the Caribbean,and a weakaverage subsurface coun- stronginflow to the Caribbeanthrough this sectionis surface- terflow is located at the southern end of the transect below 120 m. intensifiedand showsthat the northwestwardflow alongthe Strongercountercurrents are evident in each of the individual coastof Tobagois alsohighest at the surface(Figure 10). The cruisetransects (not shown),but theyvary in horizontallocation maximumaverage surface speed is over90 cm/sdirected to the and thusdo not createa strongmean countercurrent flow. The southeast,and the standarddeviations are alsohigh, +_15-30 standard deviation of the current reaches a maximum of over _40 cm/s over nearly the entire transect,highest at the northern cm/sin the centerof the passagenear the surface,owing to both sidenear Grenada,and alsoin the centerof the passage.The variabilityin maximumsurface current speed (which ranged from averageupper 200-m transportis -6.9 +_3.2 Sv directedinto 60 cm/sin March 1994to over 130 crn/sin March 1996)and the Caribbean(Table 1), with a low of -4.0 Sv in June 1991 lateralmeandering of thesurface current core. The averagetrans- and a high of -10.3 Sv in January1991. This averagevalue is port in the upper 200 m is -3.7 +_ 1.3 Sv, directedinto the likely to be an overestimate,as the very high transportob- Caribbean,with a low of -1.6 Sv in March 1994and a highof servedduring January 1991 appearsto be anomalous. -5.2 Sv in March 1996.If the eightvalues for the two transects The secondGrenada Passagetransect runs southwestfrom are combinedto yieldan averagetransport through the Grenada Grenada to the shallow shelf waters off Venezuela at ---62øW Passage,the observations give an averagetransport of -4.9 _+2.6 (Figure11) andwas occupied five times between June 1993 and Sv into the Caribbean(Table 1). 25,814 JOHNS ET AL.' VELOCITY AND TRANSPORT IN THE IAS PASSAGES

AverageSurface Velocity Average 0-200 rn Velocity

20 20

17 .... 66 65 64 63 66 65 64 63 LONGITUDE LONGITUDE

Northwardvelocity (cm/s) std dev 50 5o 100 2OO • 200 40 80 120 0 40 80 120

Eastwardvelocity (cm/s) std dev

E 50- o- .c 100- 100

r• 150- 150

200 200 0 40 80 120 0 40 80 120 Distance (km) Distance (km)

Figure 8. Sameas Figure 2, but for the secondAnegada Passage transect.

4. Discussion ble measurements[Larsen, 1992], acousticvelocity profiling [Leamanet al., 1987],and mooredcurrent meters [Lee et al., The observationsof velocity structure in and transport 1985].The shipboardADCP observationsreported herein add throughthe passagesdescribed above confirm many features little new information about such an intenselystudied ocean reportedby previousinvestigators but alsobring to light new elementsof theflow in thepassages, particularly in thenorth- currentbut do demonstratethe reliabilityof the techniqueand ernpassages (e.g., NWPC, Great Inagua, Mona, and Anegada) lend confidenceto the observationsin the other passagesthat wherefew previousobservations have been made. Interesting do not haveextensive previous and ongoingstudies with which featuresthat thesepassages have in commoninclude the pres- to compare. enceof a subsurfacevelocity maximum and persistentsurface- Severalstudies have been done previouslyin the Northwest intensifiedcounterflows that tend to be found alongone sideof ProvidenceChannel (NWPC). The first, by Richardsonand eachpassage. The FloridaCurrent, the easternCaribbean, and Finlen [1967],used dropsondes to directlymeasure transport. the GrenadaPassage also share a commonvelocity structure, They obtainedfour sectionsacross the NWPC duringMarch having high velocity,narrow surface-intensifiedflows with 20-22, 1966. They found westward(toward the Florida Cur- strongvertical and horizontalshears. rent) flow in the northernhalf of the channeland eastward flow in the southernhalf, just as we observedin the present 4.1. ComparisonWith Previous Studies study(Figure 4). They found maximumsurface velocities in The observations in the Florida Current at both 26øN and both directions of 60-80 cm/s, which are somewhat higher 27øN(Figures 2 and 3) are veryconsistent with previousstud- than our observations,but it mustbe kept in mind that theirs ies,in termsof velocitystructure and transport, as determined were taken during one limited time period and ours are an by dropsondes[Schmitz and Richardson,1968], submarine ca- average.Richardson and Finlen [1967] observeda transport JOHNS ET AL.: VELOCITY AND TRANSPORT IN THE IAS PASSAGES 25,815

Average Surface Velocity Average 0-200 rn Velocity

16 16

• lS

12 12

11 11

10 10 67 66 65 64 63 62 61 60 59 67 66 65 64 63 62 61 60 59

LONGITUDE LONGITUDE

Northwardvelocity (cm/s) std dev

20015o 0 200 400 600 0 2oo 4oo 6oo

Eastward velocity!(cm/s) std dev ø-k TMi/i\ I ! ii -I E 50-- 50 -: 100- IO0

Q 150- 150

2OO 0 2t )o 400 600 0 200 400 600 Distance (km) Distance (km) Figure 9. Same as Figure 2, but for the Caribbean transectalong 63.5øW.The orientation of the vertical sectionsis from north to south along the transect.

rangeover the full water columnof - 1.5 to -2.5 Sv toward the made in the NWPC usingthe Pegasusacoustic velocity profiler Florida Current, consistentwith our transports considering during three cruisesbetween November 1990 and September that our observationsare for the upper 200 m only and theirs 1991.They found an averagetransport addition to the Florida are for the whole channeldepth. An interestingresult discov- Current of - 1.2 Sv over the full water column, consistentwith ered during the present studyby comparingFlorida Current our observations.They also presentvelocity sections perpen- transport data from the submarinecable, which gives total dicular to and along the transectorientation, which can alsobe water columntransport, to the shipboardADCP data is that comparedwith our observations(at leastin the upper 200 m). due to the vertical shear structurethe upper 200 m contains Their resultsdiffer from ours, in that they observedinflow to ---50% of the total 800-m transport.Examination of full water the Florida Current only below 80-100 m depth, and outflow column transportsin the easternCaribbean passages[Wilson abovethat depth and on the northern half of the channel.We, and Johns,1997] yielded the same percentage,implying that on the other hand, observed inflow in the center of the channel there is a common baroclinic structure to these currents, and all the way to the surface, although as describedabove and thus a rough estimatecan be made of total transportby mul- shownin Figure 4, we note that the inflow velocity does in- tiplyingthe upper200-m transportsby a factor of 2. (Of course, creasewith depth, so at least the shear structuresof the two this is not alwaysvalid and mayvary for individualpassages as studiesagree. Leaman et al. [1995] alsoshow the cross-transect well as temporally.The presenceof counterflowsbelow 200 m componentof velocity,which is small (<15 cm/s),southward, is also a possibilitythat would causesuch extrapolation of the and confinedto the upper 50 to 100 m within the southernhalf transportto be invalid.) of the channel, in good agreementwith our observations. More recently,Leaman et al. [1995] reported observations We are not aware of any previousobservations in the Great 25,816 JOHNSET AL.: VELOCITY AND TRANSPORTIN THE IAS PASSAGES

Average SurfaceVelocity Average0-200 rn Velocity

GRENADA GRENADA

lO 62 61 60 63 62 61 60

LONGITUDE LONGITUDE

Northwardvelocity (cm/s) std dev 0 0 ' '

50- 5O

100 - 100

150 - 150

2OO 2OO 0 40 80 120 160 0 40 80 120 160 Eastwardvelocity (cm/s) std dev 0• I \• (; 'L?I lOO 50 100 15o 150

200 • I • I 200 0 40 80 120 160 0 40 80 120 160 Distance (km) Distance (km) Figure 10. Sameas Figure2, but for the first GrenadaPassage transect. The orientationof the vertical sectionsis from northwestto southeastalong the transect.

Inagua-HaitiPassage, and few studieshave been done in the velocityobservations, the role of the other northernpassages Windward Passage,even fewer with direct velocityobserva- was not consideredto be significant.On the contrary, our tions.Roemmich [1981], in his inversecalculation of the circu- observationsshow that the Mona and Anegada Passagesmay lation of the CaribbeanSea usinghydrographic data collected contributenearly the sameamount of transportas the Wind- between 1954 and 1974, concludedthat the flow through the ward Passage(Table 1). WindwardPassage above the 27.4 sigmatheta surface(the Very few direct velocityobservations have been made in approximatesill densityof the Florida Current)was -7 Sv, Mona Passage.An earlierstudy [Metcalfet al., 1977]conducted with -22 Svcoming from the easternCaribbean and the other using423 drift bottles released in thepassage (of whichonly 29 few Sv necessaryto balancethe Florida Current comingfrom were returned)concluded that flow therewas highly variable, the Old Bahama Channel and the NWPC. However, there is a with somebottles traveling out of the Caribbeaninto the At- greatdeal of variabilityin Great InaguaPassage (and presum- lantic, and others going in the oppositedirection into the ablyalso the Windward Passage), with our nine transects show- Caribbean,more in agreementwith our observationof an ing a low of -0.5 Sv and a highof -5.2 Sv, soRoemmich's, averageinflow. In a later study, a satellite-trackedsurface [1981]-7 Svcertainly falls within our observedtransport range drifter was releasednear Mona Passagein October 1980 and if we take into accountthe transportextrapolation which could trackedfor 6 months[Williams, 1986]. Again, a complexflow doubleour 200-m transports.Kinder et al. [1985],in a review patternwas observed,as the drifter remainednorth of the article on the circulation of the Caribbean, attributes - 10 Sv to Caribbeanfor the first 4 months,and then subsequentlyturned the WindwardPassage, based on a compilationof resultsby and traveledthrough Mona Passageinto the Caribbeanat an Worthington[1976], Roemmich [1981], and severalothers. In averagespeed of 38 cm/s.Williams [1986] estimated the trans- these studies,which were basedon hydrographicrather than port associatedwith the drifter motion to be -1 to -4 Sv. JOHNS ET AL.' VELOCITY AND TRANSPORT IN THE IAS PASSAGES 25,817

Average Surface Velocity Average 0-200 rn Velocity

GRENADA GRENADA

lO 10 64 63 62 61 64 63 62 61 LONGITUDE LONGITUDE

Northwardvelocity (cm/s) std dev

50 50 lOO

15o 20o 200 i I/• I I o 20 40 60 80 0 20 40 60 80

100

1O0 .•...•"• ß..• 150 200 • • 200 , • 0 20 40 60 80 0 20 40 60 80 Distance(km) Distance (km)

Figure 11. Same as Figure 2, but for the secondGrenada Passagetransect. The orientationof the vertical sectionsis from north to south along the transect.

These observations are consistent with our ADCP measure- passagestogether accountfor close to half of the total trans- ments,which alsoshow highly variable transport (Table 1). port of the Florida Current as will be discussedbelow. Similarly, few previous studies of the upper layer flow The data collectedin the easternCaribbean along 63.5øW in through Anegada Passage have been conducted. Metcalf 1985-1986have been describedin detail by Smithand Morrison [1976],using a combinationof water massanalysis and geostro- [1989] and Morrisonand Smith [1990]. The former describes phic velocity computationsas well as a few direct current ADCP data methods and comparisonswith acousticvelocity measurements,computed a transportinto the Caribbean of profiler observationsand with geostrophicobservations, and -1.4 Sv for the upper 200 m, well within our -2.4 __+2.8 Sv the latter discussesthe observationswith particular regard to averagetransport (Table 1). His conclusionwas that the An- seasonalvariability of the geostrophictransport. The small egadaPassage is probablymore significantin terms of deep differencesbetween our reported transportsfor this section inflowsto the Caribbean,which ultimately fill the deep interior (Table 1) andthose tabulated by Smithand Morrison [1989] are basins,rather than in upper layer transportwhen comparedto due to methodologydifferences such as smoothing,interpola- the higher transport of the southern passages.However, tion, gridding, etc. Morrison and Smith [1990] discussthe Schmitzand Richardson[1991] concluded,based primarily on bandednature of the flow across63.5øW, with three primary water massanalysis, that the AnegadaPassage may play a more bandsof westwardflow separatedby eastwardbands or regions important role and should be examined more closely.Our of little flow. They hypothesizethat the eastwardcurrents may observationsagree with this, showing that the transport either be a result of the spacingof the passagesof the Lesser throughAnegada Passageis similarin magnitudeto Windward Antilles or a geostrophicresponse to the westwardinput of and Mona passages,and furthermorethat the three northern high-salinity subtropical underwater through the passages. 25,818 JOHNS ET AL.' VELOCITY AND TRANSPORT IN THE IAS PASSAGES

30 They report a mean geostrophictransport for the five sections 16.5 of -18 Sv above a zero reference surface at 1000 m, consistent 28 Gulf of Mexico 0.5 Atlantic Ocean with the -9.5 Sv directly measuredtransport for the upper 26 200 m reported herein. They note that the seasonalvariability 24 observedis in agreementwith the regionalclimatic wind stress curl distributionof Hastenrathand Lamb [1977] and is also u.! 22 consistentwith Florida Current transport variability as mea- ::) 20 sured by cable [Larsen, 1992] for the same 1985-1986 time period with a phase lag of 90-100 days. Other estimatesof 16 transportthrough the eastern Caribbean includeRoemmich's 14 Caribbean Sea [1981]value of -22 Sv,Schmitz and Richardsoh's(1991) recal- 12 culationfrom Stalcupand Metcalfs (1972) data of -28.8 Sv, 10 and Schmitz and McCartney's [1993] estimatesof -23 Sv 8 through the eastern Caribbean. Gordon [1967] computed a 90 88 86 84 82 80 78 76 74 72 70 68 66 64 62 60 58 geostrophictransport of -26 Sv along 64ø30'W (which in- LONGITUDE cludesAnegada Passage),and Kinder et al. [1985] estimated the transport at -20 Sv. As listed in Table 1 for the upper Figure 12. Schematicrepresentation of the circulationof the 200 m, our observationsalong 63.5øW range from a high of Intra-AmericasSea in the upper 200 m basedon the observed -14.2 Sv in August 1985 to a low of only -1.7 Sv in October ADCP transports(Table 1). Approximateaverage transport is indicatedin Sverdrups. 1986 when there was a strong,wide eastwardcurrent present over much of the northern half of the section.Velocity structure for the individualcruises are shownby Smithand Morrison 4.2. Transport Balance [1989]. The Grenada Passagehas been the subjectof severalprevi- The transportvalues for the upper 200 m of the transects ousstudies [cf. Stalcupand Metcalf, 1972],most yielding results representa balanceon averagebetween inflow throughall of from singleshipboard occupations and thereforeprobably not the passagesfrom the Northwest Providence Channel to representativeof mean flow conditions.More recently,Wilson Grenada Passageand outflowthrough the Florida Current at and Johns[1997] have conducteda comprehensiveprogram to 27øN(Table 1). Thisresult is shownschematically in Figure12. The inflow sum is monitor the velocity structurein the passagesof the eastern Caribbean using a cable-loweredADCP in conjunctionwith 0.4 (NW) + 2.2 (GI) + 2.8 (Me) + 2.4 (AP) + 9.5 (CA) hydrographicobservations. The strongestflow of all the pas- sageswas found to be in GrenadaPassage. The averageveloc- = 17.3 Sv. ity structure there, obtained from seven transectscollected Note that GrenadaPassage is not addedto the sum,as any flow between December 1991 and July 1994 and shown in their throughthe GrenadaPassage would be includedin the eastern Figure 4, showedsurface-intensified westward flow with inflow Caribbeantransect (Figure 1). This inflow sum balancesthe speedsup to 60 cm/s in the center of the channel, and a 16.5 Sv observedupper 200-m outflow of the Florida Current counterflowlocated along the southernedge of the passageat at 27øN to within less than 1 Sv. This balance is shown sche- 100-250 m depth with a speedless than 10 cm/s.Their mean matically in Figure 12. It must be noted that although the transportfor the six transectswhich had full-depth coverage averagetransport is in balance,within any individualcruise the (someof whichwere includedin the presentstudy) was -4.7 _+ passagetransports do not balance as closely.This is due pri- 1.6 Svto the bottomof the passage,which is locatedat ---900-m marily to lack of synopticityof the measurements,which were depth.The observedvelocity structure described by Wilsonand typicallycollected over a period of two to three weeks.Trans- Johns[1997] is in good agreementwith that from the present port variabilityis undoubtedlyhigh over that time frame. In the study(Figures 10 and 11), but the transportis lower, as our Florida Current, for example,Larsen [1992] hasobserved very averagetransport through Grenada Passagewas -4.9 _+1.9 Sv large changesin transport(as high as 50% of the total trans- for the upper 200 m alone. Our value is -4.1 Sv if the anom- port) overmonth-to-month timescales. The synoptictransports alouslyhigh transportobserved during one particularcruise is do not necessarilyhave to be in balance anyway,as sea level discarded.There are several reasonsthat could explain this changescan occur.However, over time, givenenough individ- apparent disagreementin transport. Variability due to tides ual realizationsof eachpassage transport, a meaningfulaver- and interactionwith eddies impingingupon the islandsfrom agetransport for eachpassage can be obtainedand an average the east, coupledwith an insufficientnumber of observations transport balance achieved.Of course,Figure 12 is only a to obtain a stable mean, is possible.Also, only the last five of schematicof the averagecirculation. Drifter studies,for exam- the passageoccupations listed in Table 1 were colocatedwith ple, thosepresented by Molinariet al. [1981],always show that the Wilsonand Johns[1997] transect,and the mean of these there is a great deal of eddy variability, particularlyin the five is only -3.7 Sv. Finally, the transport structure of the regionof the easternCaribbean, and so at any giventime the GrenadaPassage reported by Wilsonand Johns [1997] includes flow field will be much more complicatedthan the schematic a mean outflowover muchof the passage,mostly below 200 m, portrayal.Nevertheless, certain features of the Caribbeancir- which would tend to reduce the total transport.More recent culationdo on averageresemble the schematicof Figure 12, calculationsof the averageGrenada Passagetransport from suchas the convergenceof the flow from the passagesto form the ongoingWilson and Johns[1997] program,which include the CaribbeanCurrent/Loop Current system. 10 cruisesthrough 1996,show the samefeatures in the velocity The differencebetween the averageupper 200-m transport structureand have a slightlyhigher mean transportof -5.6 Sv. at 27øN(16.5 Sv) and the transportat 26øN(15.4 Sv) is due to JOHNS ET AL.: VELOCITY AND TRANSPORT IN THE IAS PASSAGES 25,819

the addition of transport through the Northwest Providence and off the coast of northeastern Brazil. Similar observations Channel. The upper 200-m transportsthrough the Great In- couldin the future be easilymade from Volunteer Observing agua, Mona, and Anegada passagesare all roughly equal at Ships outfitted with ADCP units such as described herein, about -2.5 Sv each(Table 2) and summedtogether are 7.4 Sv, providinga cost-effectivecurrent monitoring tool. Monitoring nearlyhalf of the 26øNFlorida Current transport.Of the -9.5 the circulationof the IAS over a longtime periodshould prove Sv total for the easternCaribbean south of the AnegadaPas- enlighteningin termsof seasonal,interannual, and longer-term sage,Grenada Passage accounts for -4.9 Sv, or approximately variability.Further understandingof the circulationof the IAS half of that total. This is a significantfinding, as the Grenada and its interconnectednessin terms of fisheries,pollution, lar- Passageis where the greatestinput of water originatingin the val transports,etc., will also dependon focusedobservational southernhemisphere is likely to be found. A new programto programsthat incorporate satellite fields of sea surface tem- continuouslymonitor the transportthrough Grenada Passage perature, ocean color and altimetry with Lagrangian drifters, using a submarinecable similar to the one that has been suc- and shipboard-basedinterdisciplinary process studies. cessfullymonitoring the Florida Current transport is in the early stages(W. Johns,personal communication, 1998) and Acknowledgments. The authors thank the officers and crew of the shouldproduce enlightening results about the long-termmean R/V MalcolmBaldrige and our manyAOML colleagueswho assistedin and variability of the transportthere. the collectionof the data.Tom Lee and Bill Johnsof the Universityof Miami provided many helpful commentsand suggestionsthat im- 4.3. Interannual Variaiblity provedthe final manuscript.The ChapmanConference on the Circu- The earlier studiesof transport in the eastern Caribbean lation of the Intra-AmericasSea held in La Parguera,Puerto Rico, in January1995 providedinspiration for this study. discussedabove [Gordon, 1967; Stalcup and Metcalf, 1972;Ro- emmich, 1981; Schmitz and Richardson, 1991; Schmitz and Mc- Cartney,1993] all showedhigher transport than the observa- References tions reported herein. It is tempting to speculatethat these transportsindicate a trend toward lower transportthrough the Atkinson, L. P., T. Berger, P. Hamilton, E. Waddell, K. Leaman, and T. N. Lee, Current meter observationsin the Old Bahamas Channel, southernCaribbean passages over the pastseveral decades. As J. Geophys.Res., 100, 8555-8560, 1995. the Florida Current transport is known not to have varied Bourles, B., R. L. Molinari, E. Johns, W. D. Wilson, and K. D. Lea- significantlyover the sametime period [Larsen,1992], sucha man, Upper layer currentsin the westerntropical North Atlantic trend would imply that there may have been a shift in the (1989-1991), J. Geophys.Res., 104, 1347-1361,1998. Fratantoni,D. M., W. E. Johns,and T. L. Townsend,Rings of the prevailinglarge-scale circulation pattern, with a larger propor- North Brazil Current: Their structure and behavior inferred from tion of the flow now enteringthe Caribbeanthrough the north- observationsand a numerical simulation,J. Geophys.Res., 100, ern passagesthan in the 1970s. 10,633-10,654, 1995. The available observations are too limited to allow more Gordon,A. L., Circulationof the CaribbeanSea, J. Geophys.Res., 72, than speculationon interannualvariability, however.This is 6207-6223, 1967. Hastenrath,S., Decadal-scalechanges of the circulationin the tropical alsowhy we do not addressthe subjectof seasonalvariability of Atlantic sectorassociated with Sahel drought,Int. J. Climatol.,10, the flow throughthe passages.The large range in transport 459-472, 1990. throughthe variousCaribbean transects (Table 1) showsthe Hastenrath,S., and P. Lamb, ClimateAtlas of the TropicalAtlantic and hazardsin basinglong-term mean estimatesof transport on EasternPacific Oceans, 110 pp., Univ. of Wisc.Press, Madison, 1977. only a few realizations.However, indicationsfrom long time Johns,W. E., and F. Schott,Meandering and transportvariations of the Florida Current, J. Phys.Oceanogr., 17, 1128-1147, 1987. seriesclimate data are that there havebeen somevery notice- Johns,W. E., T. N. Lee, F. A. Schott,R. J. Zantopp, and R. H. Evans, able changesin the tropicaland subtropicalNorth Atlantic on The North Brazil Currentretroflection: Seasonal structure and eddy the same decadal timescale,including changes in large-scale variability,J. Geophys.Res., 95, 22,103-22,120, 1990. wind patterns,sea surfacetemperature, hurricane frequency, Kinder, T. H., G. W. Heburn, and A. W. Green, Someaspects of the Caribbean circulation, Mar. Geol., 68, 25-52, 1985. droughtconditions in Africa, and changesin ENSO strength Landsea, C. W., W. M. Gray, P. W. Mielke Jr., and K. J. Berry, and frequency[Hastenrath, 1990; Landsea et al., 1992]. Long-termvariations of West Sahelianmonsoon rainfall and intense Schmitzand Richardson[1991] arguethat the transportinto U.S. landfallinghurricanes, J. Clim., 5, 1528-1534, 1992. the southernCaribbean includes a large southernhemisphere Larsen,J. C., Transport and heat flux of the Florida Current at 27øN componentrequired to maintain a thermohalineoverturning derivedfrom cross-streamvoltages and profiling data: theory and observations,Philos. Trans. R. Soc. London, Set. A, 338, 169-236, circulationof -15 Svin the Atlantic.If the lowertransports we 1992. observein the easternCaribbean also include correspondingly Larsen,J. C., and T. Sanford,Florida Current volumetransports from lesssouthern hemisphere water, alternativepathways outside voltagemeasurements, Science, 227, 302-304, 1985. the Caribbean would be required for the cross-equatorial Leaman, K. D., and R. L. Molinari, Topographicmodification of the Florida Current by Little Bahamaand Great BahamaBanks, J. Phys. flows. Most likely possibilitiesinclude North Brazil Current Oceanogr.,17, 1724-1736, 1987. Rings [Johnset al., 1990;Fratantoni et al., 1995],which may at Leaman, K. D., R. L. Molinari, and P.S. Vertes, Structure and vari- timesdrift northwestwardinstead of impingingupon the island abilityof the Florida Currentat 27øN:April 1982-July1984, J. Phys. chain as they often do, and interaction between the North Oceanogr.,17, 565-583, 1987. Equatorial Countercurrentand the North Equatorial Current Leaman, K. D., P.S. Vertes, L. P. Atkinson, T. N. Lee, P. Hamilton, and E. Waddell, Transport, potential vorticity, and current/ [Mayer and Weisberg,1993; Bourleset al., 1999]. Additional temperature structure acrossNorthwest Providence and Santaren analyses,particularly involving water masstracer analysis,and Channelsand the Florida Currentoff Cay Sal Bank,J. Geophys.Res., numerical models forced by realisticwinds for the decadesin 100, 8561-8569, 1995. questionwill likely shed more light on these issues. Lee, T. N., F. Schott,and R. Zantopp,Florida Current:Low-frequency variability of the Florida Current as observedwith moored current Finally, many of the observationspresented above were ob- meter stationsduring April 1982-June1983, Science, 227, 298-301, tained along ship trackswhile the researchvessel was "dead- 1985. heading" to or from various researchstudy areas in the IAS Mayer, D. A., and R. H. Weisberg,A descriptionof COADS surface 25,820 JOHNS ET AL.: VELOCITY AND TRANSPORT IN THE IAS PASSAGES

meteorologicalfields and the implied Sverdruptransports for the Schmitz,W. J., Jr., and W. S. Richardson,On the transport of the Atlantic Ocean from 30øSto 60øN,J. Phys.Oceanogr., 23, 2201-2221, Florida Current, Deep Sea Res., 15, 679-693, 1968. 1993. Schott,F., and R. Zantopp,Florida Current:Seasonal and interannual Metcalf, W. G., Caribbean-Atlantic water exchangethrough the An- variability,Science, 227, 308-311, 1985. egada-JungfernPassage, J. Geophys.Res., 81, 6401-6409, 1976. Smith, O. P., and J. M. Morrison, Shipboardacoustic Doppler current Metcalf, W. G., M. C. Stalcup,and D. K. Atwood, Mona Passagedrift profilingin the easternCaribbean Sea, 1985-1986,J. Geophys.Res., bottle study,Bull. Mar. Sci., 27, 586-591, 1977. 94, 9713-9719, 1989. Molinari, R. L., M. Spillane,I. Brooks,D. Atwood, and C. Duckett, Stalcup, M. C., and W. G. Metcalf, Current measurementsin the Surfacecurrents in the CaribbeanSea as deducedfrom Lagrangian passagesof the LesserAntilles, J. Geophys.Res., 77, 1032-1049, observations,J. Geophys.Res., 86, 6537-6542, 1981. 1972. Molinari, R. L., et al., SubtropicalAtlantic climatestudies: Introduc- Sturges,W., Observationsof deepwater renewalin the CaribbeanSea, tion, Science,227, 292-295, 1985. J. Geophys.Res., 75, 7602-7610, 1970. Mooers, C. N. K., and G. A. Maul, Intra-Americas Sea circulation, in Williams,W. G., Lagrangianmeasurements of surfacecirculation near The Sea, vol. 11, The Global Coastal Ocean: Ideas and Observations Puerto Rico, J. Phys. Oceanogr.,16, 1715-1719, 1986. on Progressin the Studyof the Seas,edited by A. R. Robinsonand Wilson, W. D., and W. E. Johns,Velocity structureand transportin K. H. Brink, pp. 183-208, John Wiley, New York, 1998. the Windward IslandsPassages, Deep Sea Res., 44, 487-520, 1997. Morrison, J. M., and O. P. Smith, Geostrophictransport variability Wilson, W. D., and J. A. Routt, Shipboardacoustic Doppler current along the Aves Ridge in the eastern CaribbeanSea during 1985- profilerdata collectedduring the SubtropicalAtlantic ClimateStud- 1986, J. Geophys.Res., 95, 699-710, 1990. ies (STACS) project (1989-1990), NOAA Tech. Memo., ERL Ogden, J. C., Marine managerslook upstreamfor connections,Sci- A OML- 71, 1992. ence, 278, 1414-1415, 1997. Wilson, W. D., E. Johns,and R. L. Molinari, Upper layer circulation Pillsbury,J. E., The Gulf Stream--A descriptionof the methodsem- in the westerntropical North Atlantic Oceanduring August 1989, J. ployed in the investigation,and the resultsof the research,report Geophys.Res., 99, 22,513-22,523, 1994. Appendix10, pp. 461-620, U.S. Coastand GeodeticSurvey, Silver Worthington,L. V., On the North Atlantic Circulation,120 pp., Johns Spring,Md., 1890. Hopkins Press,Baltimore, Md., 1976. Richardson,W. S., and W. J. SchmitzJr., A techniquefor the direct Wust, G., Stratificationand Circulationin theAntillean-Caribbean Ba- measurementof transportwith applicationto the Straitsof Florida, sins.I. Spreadingand Mixing of the Water TypesWith an Oceano- J. Mar. Res., 23, 172-185, 1965. graphicAtlas, 201 pp., ColumbiaUniv. Press,Palisades, New York, Richardson, W. S., and J. R. Finlen, The transport of Northwest 1964. ProvidenceChannel, Deep Sea Res., 14, 361-367, 1967. Roberts, C. M., Connectivityand managementof Caribbean coral E. Johns,R. L. Molinari, and W. D. Wilson,Atlantic Oceanographic reefs, Science,278, 1454-1457, 1997. and MeteorologicalLaboratory, NOAA, 4301 RickenbackerCause- Roemmich, D., Circulation of the Caribbean Sea: A well-resolved way, Miami, FL 33149. ([email protected]) inverseproblem, J. Geophys.Res., 86, 7993-8005, 1981. Schmitz, W. J., Jr., and M. S. McCartney, On the North Atlantic circulation,Rev. Geophys.,31, 29-49, 1993. Schmitz,W. J., Jr., and P. L. Richardson, On the sourcesof the Florida (ReceivedDecember 20, 1998;revised July 8, 1999; Current, Deep Sea Res., Suppl.,38, S379-S409, 1991. acceptedJuly 27, 1999.)