Mass Balance Models of Ekman Transport and ... - University of Utah

GLOBAL BIOGEOCHEM!CAL CYCLES, VOL. 8, NO. 2, PAGES 165-177,JUNE 1994

Mass balance modelsof Ekman transport and nutrient fluxes in coastalupwelling zones Paul W. Jewell Departmentof Geologyand Geophysics,Universityof Utah, Salt Lake City

Abstract. The nutrientcyclesof coastalupwellingzonesare studiedwith simplemass balancemodelsof Ekmantransport,longshoretransport,surfaceproductivity,and dissolvedphosphorous.The modelsare constrained with datafrom the Peru, northwestAfrica, andOregonupwellingsystems.The onshore-offshore massbalancemodel agreeswith publishedEkmantransport,surfaceproductivity,andnutrientdataas well ashypothesized nutrientf-ratiosfor highlyproductivecoastalsettings.The onshore-offshore modelsuggests that increased primaryproductivityin glacial-era coastalupwellingzoneswasnot a linear functionof Ekmantransport,but insteadwas probablydependent on the physicalandchemicaldynamicsof a specificsetting. In the Peruupwellingsystem,longshoreequatorwardsurfacecurrentsandpolewardundercurrentsproducepositivesurfacenutrientgradientsin the equatorwarddirectionand relativelyconstantgradientsin subsurface waters. Longshorenutrientgradientsoff northwestAfrica are positivein the equatorwarddirectionfor both surfaceand subsurfacewaters. Theseobservations are consistent with the conceptualmodelof surface andsubsurface currentswhichare movingtowardthe equatorandcontinuallybeing upgradedby the offshoreflux of nutrients. The northwestAfrica andPeru data are not consistent with the longshorenutrientmodelof Redfieldet al. (1963).

Introduction

creasedlow-latitude upwelling during a glacial climate may have produced lower atmosphericCO2. Boyle [1986] presentsa box model which suggeststhat increasedlow-latitudeupwelling loweredatmosphericCO2 by-25 parts per million by volume. The rain ratio theory [Berger and Keir, 1984] holds that increasedEkman transportduringglacial timesdecreased atmospheric CO2 due to the very high organiccarbon/carbonate carbon ratioswhich are typicalof upwellingzones. $arntheinet al. [1988] summarizepalcoproductivity relationshipsof the low- and mid-latitudeoceansand how they may have influencedatmosphericCO2. Discussionsof low-latitude upwelling during glacial periodsimplicitly assumethat strongerzonal winds increasedEkman transportalongcoastalmarginsor Ekman pumpingnear the equator. This in turn increasedsurface productivitytherebydecreasingatmospheric CO2. All of these assumptionsmerit examinationwithin the framework of modem meteorologicaland oceanographic data. For instance, although it is recognized that stronger winds generally cause higher Ekman transport [e.g., Smith, 1981; Lentz, 1992], there are dynamicsituations suchas the westerncoastof Australiawhere equatorward at. [ 1963]. A considerable body of literaturealso addresses the winds resultin little or no Ekman transport[Smithet al., r,,1,•that,,,,w•n•,o may haveplayedin globalprocesses. 1991]. On the other hand, the commonlyassumedlink betweenstrongEkmantransportandhigh surfaceproducNewell et al. [1978] were the first to suggestthat intivity has not been critically examinedwith data from Copyright1994by theAmericanGeophysical Union modemcoastalupwellingzones. The presentstudywasundertakento providea quantitaPapernumber94GB00097 tive understandingof the relationshipsbetweenEkman 0886-6236/94/94GB-00097510.00 transport, longshorecurrents, surfaceproductivity, and

Understanding exchanges of nutrientsand oxygenbetweenvariouspartsof the oceanand the atmosphere has occupiedthe imaginationof oceanographers and earth scientistsfor decades[e.g., Broecker,1974; Garrelsand Perry, 1974; Holland, 1978; Berneret al., 1983]. Mass balancemodelsof biogeochemical constituents haveproven to be effectivetools for understanding theseexchanges in both modem and ancient settings[e.g. $armiento et al., 1988; Shaffer, 1989]. Othermodelshaveaddressed the causeof atmosphericCO2 changeswithin the context of circulationin the ocean [e.g. Enneverand McElroy, 1985; Toggweilerand $armiento, 1985; Wenk and $iegenthaler,1985]. While thesemodelshave beenuseful in testing various scenariosof world oceanprocesses, they have (almostby necessity)neglectedthe detailsof the marginsof oceanbasins. Coastalupwellingzonesare amongthe mostbiologicallyproductiveareasof theocean [Koblents-Mishke, 1970;Berger, 1989] andthusare critical to understanding thesemarginalsettings. Very few biogeochemical modelsof coastalupwellinghave been attempted.A notableexception is thework of RedfieMet

166

JEWELL: COASTAL UPWELLING MODELS

nutrientconcentrations in coastalupwellingzones. The The equatorward windsin a coastalupwellingsystem procedureinvolves constructingsimple, mass balance producea surfaceshearstresswhichis balancedby Comodelsof idealizeAcoastalupwelling. Chemicaldata riolis forces. The resultingflow is perpendicular to the basesfrom well-studiedmodemoceanographic settings wind direction. The total amountof masstransport then constrain results of the mass balance models. The (termedEkman transport)is equalto x/f wherex is the studyis meantto addressseveralquestions.(1) Are the wind shearstressandf is the Coriolisparameter.Surface relationships betweenproductivity,Ekmantransport,and Ekmantransportoccursin a relativelyshallow(orderof nutrientflux consistent betweenupwellingzonesin wide- tens of meters)zone near the surface. The thicknessof ly differentsettings? (2) Can the longshoremodel of the Ekmanlayeris a functionof thedensitystratification, and the shearstressin the fluid RedfieMet al. [1963] be validatedby field dataandused the Coriolisparameter, to explainthedevelopment of oxygendepletionandnutri- andcanvary significantlyfrom oneupwellingsystemto the next [Smith, 1981; Lentz, 1992]. ent enrichmentin water over continentalshelves?(3) What implications mightthesesimplemassbalancemodTheoffshoremovement of waterdueto Ekmantransport is balanced by the onshoretransportof waterbeneaththe els havefor upwellingsystems in the geologicpast? Alwithin though not specificallyaddressedby this study, the surfaceEkmanlayer. Upwelledwateris generated procedureis applicableto equatorialupwellingas well as 100-300 m of the surface[Bowden,1983; Pond and Pickcoastalupwelling. ard, 1983] and generallyflows upwardover the continental shelf and/or

shelf break.

The

effect of the

upwellingprocessis to causeisopycnalsto intersectthe water surfaceadjacentto a landmass(Figure 1). The Models of Modern CoastalUpwelling dense,nutrient-richwaterenhances surfaceproductivity The generalphysical,biological,and chemicalpro- adjacent to thecoast.The levelof surfaceproductivity is cesses in moderncoastalupwellingzonesareagreedupon usuallygreatestnear the coastlineand decreases seaward. by mostresearchers [e.g. Richards,1981]. Upwelling The highestconcentration of biologicallyproduced mateoccursalongthe easternbordersof oceanbasins(i.e., the rials suchas chlorophyllis locatedseawardof the zonein westernedgesof continents)where winds are equator- which isopycnal surfaces intersect the water surface ward. Theseequatorward geostrophic windsare the re[Friebertshauser et al., 1975; Hafferty et al., 1978] sult of quasi-stationary, high-pressure atmospheric cells (Figure1). Theenhanced surface productivity of upwellin the middleof the ocean. In the modemocean,coastal ing zonesincreases biologicalactivityat other trophic upwellingis most intensealong the coastsof northwest levelsas well as the flux of organicmatterfrom the surAfrica, southwestAfrica, Peru-Chile, and the west coast facephoticzone. of the United States. Seasonal upwellingoccursin variA fractionof organicmatterproducedat the surface ous parts of the Indian Ocean. Palcogeographic recon- sinks throughthe photic zone and is remineralizedat structions in conjunction with the placement of depthby a varietyof oxidants. Beneaththephoticzone atmospherichigh- and low-pressurecells have allowed of upwellingsystems,oxygen,and morerarely, nitrate zonesof coastalupwellingto be pinpointedin palcogeo- are consumed. In rare instancesin the Pacific, sulfate graphicreconstructions [e.g. Parrish, 1982]. reduction hasbeenobserved [Dugdaleet al., 1977]. Baroffshore ,,•_••productive ._• IayerEkman zonephotic

,•. •

100

...::?•

200

i

300

i

I

I

I•"•-..,

I

I

I

I

I

I

•ii!:'

I

PO4 (/• mot/L) 100

200

300

Figure 1. Cross-sectional hydrographic plotsnear 15øSoff of the Perucoast. From stations363-377, JOINT II expedition,May 5-6, 1977 [Haffertyet al., 1978].

JEWELL: COASTAL UPWELLING MODELS

ber and Smith [1981] attributethis to strongpoleward undercurrents which, accordingto the longshorenutrient modelof RedfieMet al. [1963], lead to enhancedsubsurfacenutrientenrichmentand oxygendepletion.

167

entsis not evaluatedin this study,and Ekmantransportin the box model is assumed to be constant.

The subsurfacecoastal(sc) zone receiveswater from the subsurface offshore(so) zone (Figure 2). Although the specificsof individualupwellingsystems vary, the sc Model Formulation zoneprobablyextendsfrom 30-200 m depth,depending on locationof the upwellingrelative to the continental Actual coastalupwellingzonesare a combinationof shelf. The productivephotic zone (pzp) receiveswater both longshoreand onshore-offshore processes.In this from the underlyingsc box and is separatedfrom it by study,simplemodelsof both dimensions havebeenconupward curvingisopycnalsurfaces(Figure 1). The pzp stmctedin order to highlightthe characteristic processes box probably extendsfrom the surfaceto 20-50 m depth. of each. The highestbiologicalproductivities of coastalupwelling Onshore-offshore model. Onshore-offshore coastal zones take place in the pzp box. upwelling is parameterizedas a two-dimensionalbox Estimatesof primaryproductivityin upwellingphotic modelwhich representsthe generalphysicaland chemical zones are relativelynumerous.Much lessis knownabout featuresof upwellingzones(Figures1 and 2). Two offhow much of this productivityleavesthe photic zone shoreboxes are connectedto onshoreboxes (Figure 2). [e.g. Bergeret al., 1989]. Here the amountof primary The offshore boxes serve as boundary conditionsfor variablesof the nearshoreboxes. The boundarybetween productivityP exportedto the scbox is multipliedby the nearshoreand offshoreboxesis arbitrary. In reality there factor 0•. The flux of organicmaterialto water below by 0%. The is a continuousgradationof chemicalpropertiesin the the scbox and to the sedimentsis represented remaining fraction of biogenic particles produced in the cross-shelfdirection(Figure 1). The box model repreproductive photic zone (0•) is advected offshore. sentspropertygradients suchasprimaryproductivity and

Modelprimaryproductivity P hasunitsof molesm'• s'•

nutrient concentration as discrete values.

The four boxesform a continuous loopwhichexchanges and mustbe convertedto valuescommonlyreportedfor productivity (usuallyg (2 m'2yr'•). Assuming a water via Ekman transport,T (Figure 2). Ekmanvolume surface

C:P ratioof 106, it canbe foundthat1 g C m-2yr'• = 2.5 x 10's mmolesP m'2 s'•. Multiplyingthis figureby the distanceof shoreline. Ekmantransportcanbe calculated as (1) surfacewind shearstressdividedby the Coriolis width of theproductivephoticzonegivesthe net biogenic parameterand water density,(2) offshorevelocityinte- particleflux for the model. The pzp box canhavean arbitrarywidth. WidthschogratedoverthesurfaceEkmanlayerdepth,or (3) onshore to published,spatiallyavermasstransportbelowthe surfaceEkmanlayer. Compari- senfor this studycorrespond aged values of surface productivity (Table 1). The sonof thesemethodsfor calculatingEkmantransportin a arbitrary width does not change the model resultssince varietyof settingsis givenSmith[1981] andLentz[1992]. the units employed in the model (i.e., cross-shelf nutrient Integrationover the surfaceEkmanlayer is employedin this paperbecause,unl•e wind shearcalculations, water gradientsand productivity)are internallyconsistent.For transportis mostdirectlyrelatedto nutrientconcentra- instance,a wider pzp box will leadto highervaluesof P

transport hasunitsof m2/s,thatis, volumeflux perunit

tions of the box model. Furthermore, surface offshore

transportis morecoherentthanonshoretransportwhich often interactswith the bottom boundarylayer of the continental shelf.

which in turn makes computed cross-shelfproperty differenceshigher. Massbalanceequationswhich relatethe flux of chemical components betweenthe modelboxesare constructed in a fashion similar

to those outlined

in other studies

Onshore-offshore gradientsin wind shearmagnitudeare [e.g., Broecker,1974]. In theory,a numberof chemical documentedin many coastalupwellingregions[e.g. Batracers could be used in a mass balance model such as this kun and Nelson, 1991] and may accountfor the observed of the commonnutrients onshore-offshore variability in surfaceEkman transport one. In reality, only databases and oxygenare sufficientlydetailedto providethe comrates [Lentz, 1992]. The influenceof wind stressgradi-

Offshore EkrnonLover !.,T IProduct•ve Photic Zone./ff T// / mhermocl ins.......... ,•

.

.

-•-- •-......... '

•

•t

(.•uDsurtoce

/ •oostol

/ /

•one/

•nsn• )c over a,'cP • Csc ) / •x •

F•ure 2. BoxmodelrepresenUtion of coasUlupwe•g. Symbolsare expla•

• thetext.

168

JEWELL.' COASTAL

UPWELLING

MODELS

Table 1. Descriptionof VariablesUsedto Constrainthe UpwellingBox Model Peru

Location

NorthwestAfrica

15øS

Oceanographicexpedition

Date

JOINT

21ø40'N

II

JOINT

Oregon

45 ø N

I

CUE II

March-May 1977

March-April 1974

July-August,1973

20, 40 •

50 b

Variable

Primaryproductivity,gC m'2yr4

1400 •

720 b

250 ½

Depth of surfaceEkmanlayer, m

30 a

35 a

20 a

0.82 + 1.48 a 0.71 + 0.96 ß

1.45 + 1.84 a 0.82 + 0.791

90 ½

40 ½

Width of productivezone, km

SurfaceEkmantransport, m2/s

Depth of onshoreflow, m

0.65+ 1.09 0.25 + 0.41 60 ½

' Packard et al. [1983].

• Huntsman andBarber[1977]. c Walsh[1981]. Reportedvaluedoesnotcorrespond to theJuly-August 1973timeperiod. dSmith[1981]. cLentz[1992];calculatedfromthickness of the surfacemixedlayerplusa transitionthatis half thethicknessof the surfacemixedlayer.

tBaden_Dangon etal. [1986] parisons necessary to constrain models of coastal upwelling.

A simplemodelof biologicalparticleflux usingPOnas the mastervariableis employedin this study. The multiple nitrogencompounds presentin somecoastalupwelling environments(e.g., NO3, NO2, NHn) make this a more difficult elementto model. Oxygencouldalso be modeledif exchangewith the atmosphere were takeninto account. Modeling oxygen concentrations,however, is complicatedby nitrate reductionin the deep water of low-oxygensystems. The massbalanceequationfor POnin the pzp box is (Figure 2): P0 4, t,zt,- P0 4, sc

p

=--- T

Ax

(1)

The POn massbalanceequationin subsurface coastal

classicpaperby Redfieldet al. [1963]. Theseauthors arguedthat the longshoredistributionof nutrientsin a two-layer coastalsystemcould be modeledas a balance betweenhorizontaladvection,vertical diffusion, and nutrientfluxes. In the upperlayer thesebalancesare [RedfieMet al. 1963,equation(5a)]:

dNv=-R+AdN

w.f

Vuis themeanvelocityof theupperlayer,R is lossof the

nutrientN by photosynthesis, h is depthof the upper layer,andA is verticaleddydiffusivity.They coordinate is assumed to be in the longshore direction. A similar expression can be constructed for the lower layer [RedfieMet al., 1963,equation(Sb)]:

Vr. dNr=R- AdN

ay

(so) box is: P04, sc- P04, so Ax

=

P' Otsc T

(2)

Longshoremodel. The longshorevariabilityof nutrient elementsin coastalsettingswas consideredin the

(3)

haz

(4)

If one assumes that the upperand lower layersare of equal thicknessand that the loss of nutrients due to

photosynthesis in the upperlayeris exactlybalanced by respiration in the lowerlayer, thenthesetwo equations canbe combinedto yield:

JEWELL: COASTAL UPWELLING MODELS

dN•

-

Vv dNv

(5)

169

the surfaceEkman layer, longshoretransportis dominated by the polewardChile-PeruUndercurrentwhich is the sourcefor most of the upwelledwater [e.g. Brockmann et al., 1980]. This current is derived from the Pacific Equatorial Undercurrent. Equatorial upwelling producesnutrient-rich,oxygen-deficientwater which in turn influencesintermediate-depth water characteristics

Equation(5) showsthat if the upper and lower velocity have oppositesigns (a commonsituationin upwelling zonesin which equatorward,wind-drivensurfacecurrents overlie polewardundercurrents) then the horizontalnutrient gradientsin the two layers will both decreasein the from 5ø-16øS. The well-studied area near 15øS is characterized by high surfaceproductivities,bottom velocities equatorwarddirection[RedfieMet al. 1963]. which are closeto zero, and organiccarbon-richbottom More detailedversionsof (3) and (4) canbe derivedby sediments[Suesset al., 1987]. The continentalshelf near assumingthat the depthsand biogenicfluxes in the two 15øSis relativelynarrow (-20 km wide). layersare not equal: OffshoreEkman transportalonga transectat 15øSwas dNv

hr. Vv . ---- =-hr.

Rv +A

dN

(6)

calculated to be 0.82 __+1.48 m2/sby Smith[1981] and 0.71 __+0.96 m2/sby Lentz[1992] from datacollected

during the JOINT-II oceanographicexpedition in the springof 1977. Onshoremasstransportis roughly50% dNœ AdN (7) higherthanoffshoretransport[Smith, 1981]. The difference is due to entrainmentof a portion of the onshore flow in a longshoredirection. At 15øS,offshoreflow Equations(6) and (7) canbe combinedto yield: extendsfrom the surfaceto 30 m depth,and onshoreflow existsbetween30-120 m depth(Figure 3). Equatorward dNr hvVv dNv (hvRv- hrRD (8) flow is observedto depths of 30 m and the poleward dy hr Vr dy hr Vr undercurrentat depthsbetween30 and 120 m (Figure 3). In otherwords, the pzp and sc compartments of both the By assumingthat (1) Tv = hv Vvand Tt•= hLVLrepresent onshore-offshoreand longshorebox models have the two-dimensional,longshoretransportin the upper and samedepths. lower layers,respectively,and (2) only a fractionof surPrimaryproductivityhasbeenmeasuredthroughoutthe faceproductivity({x•) is remineralized in the lowerlayer, Peruvianupwelling system. A meanprimary productiv(8) can be recast into the sameterms used in the crossity of 400-700g C m'2yr"for the entiresystemfrom shelfbox model(equations1 and2): 5ø-15øS hasbeensuggested [$uesset al., 1987]. Produe-

h•.V•.-• =h•.R•- dz

l•P04, sc

TU l•P04, pzp (hr. P-hr. Ctsc . P)

Ay =-'•-•'

Ay -

T•

tivitiesare consideredto be highernear 15øS. Using the JOINT II data, Packardet al. [ 1983] calculatean average

(9)

productivity of- 1400g C m'2yr 4 for theinner40 km of the upwelling zone and somewhatless than half of that for the next 40 km seaward.

Field settings Constraintson the physicaland chemicaldynamicsof the cross-shelf and longshoremodelsare considered within the contextof data from three field settings:the Peru system near 15øS, the northwest Africa system at 21ø40'N, and the Oregon systemat 45øN. The use of thesesettingsis motivatedby severalconsiderations.All three localities were studied exhaustively in the mid-1970sby researchgroupswhich usedsimilar techniquesto determinesurfaceproductivity,Ekman transport, and nutrientconcentrations.A certaindegreeof confidencecan thereforebe drawn in the comparisons of the three settings. On the other hand, the three systems have different source-waterchemistryand bathymetryand thusrepresentthe significantcontrasts whichcanoccurin

coastalupwelling. Finally, published summaries of Ekman transportand surfaceproductivityover periodsof approximately tw'o months are available, thereby eliminatingmuchof the high-frequencynoiseinherentin short-termcoastalupwellingstudies. Peru. Coastalupwelling occursoff the coastof Peru between6ø and 17ø S in response to the southeast trade winds of the low-latitude, southernhemisphere. Below

Northwest Africa. Along the northwestcoastof Africa, upwellingoccursthroughoutthe yearbetween20øand 25ø N latitude and seasonallyover much wider latitudes [Futterer, 1983]. The data usedhere is from a transectat 21ø 40' N near Cape Blanc collectedduring the JOINT I expeditionin the springof 1974 [Barber, 1977]. In this area, the continentalshelf is -50 km wide. The highest valuesof surfaceproductivityoccurover the shelf. Althoughcirculationalong the shelf marginis spatiallyand temporallycomplex, the entire zone is characterizedby an equatorwardsurface current (the Canary current). Poleward

undercurrents

occur over the inner continental

slope. Polewardundercurrents over the continentalshelf are negligibleand the entire flow regimeover the continental shelf is 'equatorward [Smith, 1981] [Figure 3]. Offshoreflow occursto a depthof-35 m, and onshore velocity occursbetween35 m and 75 m depth [Smith, 1981]. Thesezonescorrespond to the p• andscportions cro•-•ad, andlongshore•x -'- • • Huntsmanand Barber [ 1977] reportsurfaceproductivi-

tiesof 1-3g C m':d4 overthecontinental shelf. An averageprimaryproductivity of 720 g C m'2yr 4 overa shelf width of 50 km is assumedfor this study. MinaJ et al. [1986] argue that the productivityvaluesof Huntsman

170

JEWELL:

COASTAL

OREGON

5 JULY

• •0

NORTHWEST

TO 28 AUGUST

I0 MARCH-6

1973

ONSHORE

OFFSHORE 0

©ee•l I , , ,

22 MARCH-IO

ONSHORE

40

c• 80

•e

60

1

0NSHORE

0 ' ' t'e• ' ' ' ', .

ßß

40

,- , •

20

1977

20

60

i'

' •0cm/sec •0' 3'0 -10 10203 080 -20 -10

lOO 20 -10

MAY

OFFSHORE

20

Iß

-

PERU 1974

0"•I ' ' '

ß

60

MODELS

AFRICA APRIL

OFFSHORE

• 4O •.

UPWELLING

era/see

100

0

10

20

cm/sec

POLEWARD 0

•

i

20

'

I

,

• 40 c•

'',

EQUATORWARD

ß •e

0 20

ß

i

•0

'•

'ß

c• 80

, I

•oo_ -•o

ß 0I

POLEWARDEQUATORWARD

• 20

I 30

40

, e•

,-

0

I

40

•

60

•

-10

I 10

',

0

POLEWARD EQUATORWARD

40

10

,lpI

20

30

40

cm/sec

-

leap e

ß ,,

60

80 -

ß

100_

ß I

era/see

- 20

-

I

I

-

I

_

I

I ß1

- 10

0

I

I

10

I 20

cm/sec

Figure3. Summaryof longshore andoffshorecurrentmeterdatafor northwest Africa,Peru,andOregon [Smith, 1981].

and Barber [1977] are lower than typical productivities for northwestAfrica. The Huntsmanand Barberproductivity valuesare usedherebecausethey correspond to the sametime periodof other measuredproperties(i.e., Ekman transport and dissolvednutrient concentrations). Publishedvaluesof surfaceEkman transportduring the JOINT I projectare higherthanthoseof the Peru upwelling system(Table 1). Oregon. The Oregonupwellingsystemwas analyzed in detailduringthe CoastalUpwellingEcosystems(CUE II) programin 1973 (Table 1). Upwelling is mostintense betweenlate May andAugust[Smalland Menzies,1981]. Ekman transportis weakerthanthat of the Peru and Oregon systems(Figure 3, Table 1). There is a very strong longshoreequatorialjet and relatively weak poleward undercurrent(Figure 3). Depthsof thejet and undercurrent do not correspondto the depthsof onshoreand offshoretransport. Oregon upwelling productivityand nutrientdatabases for the period in which Ekman transportanalyseswere conductedare not as comprehensive as they are for Peru

PO4, so-PO4,t,q,

P- (1- asc)

Ax

T

(lO)

The solutionprocedurefor (10) involvescalculatingthe amountof phosphateconsumedduring the Ekman transport loop for a number of stationsfrom the JOINT I, JOINT II, and CUE II datasets. The availabilityof Ekman transport,surface productivity, and nutrient data make (10) overdetermined. Extensivecurrent meter and

geochemical data allows Ekman transport[Smith, 1981; Lentz, 1992] and nutrient gradients(this study) to be determinedwith some degreeof statisticalconfidence. Surfaceproductivity,on the otherhand, is generallyreported as a singlevalue [e.g., Walsh, 1981; Packardet al., 1983] althoughHuntsmanandBarber [ 1977]reporta rangeof surfaceproductivitiesfor northwestAfrica. A possibleinconsistency in the cross-shelfbox model is the choice of box model width.

The width of the time-

averaged,nearshoresurfaceproductivityfor Peru is 40 km [Packard et al., 1983], while the currentmeter mooring usedto calculateEkmantransportwas 15-20km from and northwest Africa. Much of the nutrient data collected during July and August 1973 was at relatively shallow shore[Smith, 1981; Lentz, 1992] (Figure 4). The Peru depths(Figure 3) [Stevenson et aL, 1975]. Furthermore, systemcross-shelfbox model was thereforeevaluatedfor both 20-kin and 40-kin widths. Onshore-offshore nutrian averagesurfaceproductivityfor the July-August,1973 ent gradients in the Oregon system were evaluatedat a time period has not been published. A generalvalue of variety of stations (and hence box model because 250 g C m'2y&reported by Walsh[ 1981]for theOregon the nutrient database was so much smallerwidths), than that of the systemis usedhere. Peru and northwestAfrica systems. The irregularly spacedOregon nutrient data was normalizedto a 20 km Model results box model width for the sake of comparisonwith the Cross-shelfprocesses.Equations(1) and (2) canbe currentmeterstation(Figure4). The amount of phosphateconsumedin the Ekman combined to yield an expressionwhich reflects the amountof phosphateconsumedduring the Ekman trans- transport loop(i.e., POn,,o - POn,•) showsconsiderable variability for all three upwellingsystems(Figures5, 6, port loop of the box model(Figure 2):

JEWELL: COASTALUPWELLING MODELS

171

22ON

"x

\

•

' :'.'..

\

ß

.,

•

',

.:::..

.,oos':'\...I

"-, :-•'•....

,'

'•.. I

•.

..•!

•

lSo:•o'L -"' / o •/ '

•

-, x.:•: xL-•: ' \".... '-,,\ •'.:...... "\-"'•'• 25km -•,,, •c-.• •::

/

21000 ' .20 o

". - - ',,I"•.-'" ' ' -

,'

0"

I

''

!"'::'•:' "iJ

sow

,,

!!..':'.'.... '?.

!:.9'

'

18ø w

45ON _ 50ø

, ,..-':.

30ø

-,-,...

I

,,

1000m loom

',

'

• •'"?'

':" ',

=.,o•o.,

........x. •--• •x •"' 90' 80ø70ø

I

\x

I$øEI

I

17.30'

I

'".

•'.

17"00'

16"30'

.....::..:

"m ,OOm 50m •?: I

200

/e

fi:

ß //

/'

/ I "

•5'p ,J."• I // / / !•::-• J ,,o.,•o.,•. ,•o.w/ , I•

/

124030 '

/

I

124øW

Figure4. Location of stations forphosphate data(circles) andcurrent meterdata(triangles) usedin the onshore-offshore boxmodelanalysis.Opencirclesfor Perureferto theinnershelfanalysis (15-20km offshore) whereas solidcirclesrepresent outershelf(40-50km)analysis.

lO

lO

0.46-+ .38

Peru -20km

Peru-40 km 0.84_+.34

o0.0 •////////////////• •-///..•////'•. • 0.5 1.0 K".• 1.5 2.0 00.0 0.5 1.0 APO 4 (p.mol/L )

, 1.5

, 2.0

PO4(p. mol/L)

Figure5. Histogram ofcalculated phosphate depletions in theEkman transport loopforthePeruupwelling system.

172

JEWELL: COASTAL UPWELLING

10

•orthwest

Africo

MODELS

Oregon - 20 k m

O.85 -+.59

0.21_+.10

øo.o

0.5

1.0

1.5

2.0

APO4(P. mol/L) Figure 7. Histogramof calculated phosphate depletions in the Ekman transportloop for the northwestAfrica upwelling systems.

00.0

0.2

0.4

PO4(mol/L)

(0.86 __+0.59 •tmol/L), despitehaving relativelylow surfaceproductivity(Figure7, Table 1). Figure 6. Histogramof calculated phosphate depletions in the The degreeto which thesegeochemicaldata match Ekman transportloop for the northwestAfrica upwelling productivity andEkmantransport data(Table1) depends systems. on thevaluesof eta,thatis theamountof surfaceproductivity which is recycledin the shallowcoastallayer (Figure2). Choosing0t• = 0.5 givesreasonable results and 7). The Peru Ekmantransportloop consumes 0.46 valuesof Ekmantransport in the Peruand _.• 0.38 and 0.84 4- 0.34 pmol/L of POnfor the 20-km for published and 40-km cross-shelf box model widths, respectively northwestAfricasystems,whereaset• = 0.25 givesthe (Figure 5). The two-fold increasein the box modelwidth bestresultsfor Oregon(Table2). The poorestfit beand commensurate increase in calculated nutrient contweenmodeland datais clearlythe Oregonsystem,alsumptionsuggeststhat the cross-shelfbox model is con- thoughas statedpreviously,the Oregonnutrientdataare sistent. The phosphate consumed in the northwestAfrica not as goodas the Peru and northwestAfrica data. Two importantpointsemergefrom this analysisof system(0.21 ñ 0.10 pmol/L for a 50 km wide model)is significantlylessthanthat consumed in the Perusystem productivity,Ekmantransport andphosphate datain the (Figure6). The Oregonsystemconsumes a relatively cross-shelf domainof modernupwellingsystems.(1) For high amountof nutrientsin the Ekmantransportloop a givenvalueof 0•, the massbalancerelationships be-

Table 2. Calculated SurfaceEkmanTransport Rates,T, in thePeru,NorthwestAfrica,and OregonUpwellingSystems Peru, 20-km Width

T, published

Peru, 40-km Width

NorthwestAfrica, 50-km Width

Oregon, Variable Width

0.82 q- 1.48"

0.82 q- 1.48"

1.45 q- 1.84"

0.65 q, 1.09"

0.71 q- 0.96 •'

0.71 q- 0.96 •'

0.82 q- 0.79 ½

0.25 q- 0.41 •'

T, calculated, o•:

0.75

0.83

2.13

0.07

T, calculated, 0t:

1.10

1.24

3.19

0.11

' Smith [1981].

bLentz[1992];calculated fromthickness ofthesurfac• mixed layerplusa transition thatishalfthethicknessof the surfacemixedlayer. cBaden-Dangonet al. [1986].

JEWELL: COASTAl., UPWELLING

tweenEkmantransport,surfaceproductivity,and nutrient consumption yield consistent resultsfor threewell-studied field areas with distinctly different bathymetriesand sourcewater chemistry. In the studiesof Smith [1981] and Lentz [ 1992] as well as the box model calculationsof Ekman transportusing nutrientand surfaceproductivity data(Table 2), northwestAfrica hasthe highest,Peru the next highest, and Oregon the lowest Ekman transport rates. (2) The northwestAfrica upwelling systemhas relatively low productivity and consumesthe smallest amount of nutrientsdespitehaving the highest Ekman transportrate of the threesystems. The fractionof productivityremineralizedin the scbox (oc•) plus the fractionescapingthe Ekman loop (%) is crudelyequalto thef ratio, that is, the ratio of new productionto total production. For the Peru system,Suess [1980] estimatedthat 18% of primary productivitysinks below 110 m. This roughlycorresponds to oq of the box

14030

MODELS

ß

15000 '

ß

15 ø 30'

model(Figure2). If so, thenthef ratioof thebox model is (% + 0c•) or 0.43-0.68 (Table 2). Thesevaluesare consistent with f ratiosreportedfor upwellingsystemsin general[Epply, 1989]. Longshore processes. As suggested by RedfieMet al. [ 1963] and shownin (5), the longshoregradientsof nutrients in the upper and lower layers of coastalupwelling systemsshouldhave the samesign if the longshorevelocities are in the oppositedirection. This is the casefor the Peru upwelling system(Figure 3). On the other hand, if the flow of both layersis equatorward(the case for northwestAfrica; Figure 3), then longshoregradients of nutrientswill have the oppositesign. This simple model has been testedby plotting longshorephosphate gradientsfrom the JOINT I and II data sets(Figure 8). The depthof the upper and lower layers are assumedto correspondto the pzp and sc boxes of the onshoreoffshoremodel (Figure 3). Analysisof longshorenutrient gradientswasnot conductedfor the Oregonupwelling systembecausestationswith suitable nutrient data are relativelyfew [Stevenson et al., 1975]. Suitablestations are also irregularly spacedand the maximum distance betweenthem relatively small (< 20 kin; Figure 4). In the Peru upwellingsystem,the upper (pzp) layer shows a small equatorwardnutrient increase,whereas nutrient contentsin the lower remain relatively constant (Figure 9). The causeof theseobservations doesnot lend itself to an easyexplanation. The fact that the longshore upperand lower layer velocitieson the Peruvianshelfare in oppositedirectionsmeansthat the first term on the right-handside of (9) is positive. In order for the lefthand term of (9) to be closeto zero (Figure 9), then the secondterm on the right-handsidemustalsobe positive. This can only occurif (h,•)(oc•) > hu. (Note that TLis negative.) This is the caseoff the coastof Peru, where the depth of the polewardflowing lower layer (hi) is th___re• times that of the equatorwardflowing upper layer (hu)(Figure3), yet oc•appearsto be > 0.33 (Table 2). Phosphatedata from northwestAfrica increasesin the equatorwarddirection for both upper and lower layers (Figure 10). Sincethe upperand lower velocitiesover

173

I

I

76 c w

75030 '

/ 22000

,/

75000

../

ß

/.::.

'

/

./

I

/.': '-.

/

I

!:': ':

/

/

/

j""•j

I

\

21030 '

-•-'

ß ,'

'""

/

'.:-:

I •

).:;.: /'";;'

,

ß 21000

ß

/

/

/

/

•

I

I • • I

18oW

I

/

I

17030 '

•

I' :! •:.•

•:..!. 'N:"..

t:.•

¾.4'.

-• ,

'•:':

'

17000 '

'

'::)'.

16030 '

Figure 8. Locationof stationsfor phosphatedata usedin the longshoremodelanalysis.

(Figure 3), the model of Redfieldet al. [1963] suggests that nutrient gradientsin the upper and lower layers shouldhaveoppositesigns(equation(5)). The discrepancy betweenthe observedphosphatedata (Figure 10) and the RedfieMet al. [1963] modelarisesfrom the fact that onshore export of nutrients is not accountedfor in (3)-(9). Nutrientspumpedonshoreby Ekman transport enter the loop of photosynthesis and remineralizationand are subsequentlyrecycledas both upper and lower layers get advectedequatorward. The net result is a "nutrient trap" of progressivelyhigher nutrient concentrations in

boththeupperandlowerlayers6f theupwelling system

the shelf of northwestAfrica are both equatorward (Figure 10).

174

JEWELL: COASTAL UPWELLING

MODELS

1.0

2.0

1.0

8 I o

0

50

100

I 50

I 100

150

krn

1.0

2.0

ß I

1.0 0

I

50

EQURTORWRRD •

I

100

150

POLEW/•RD

Yigm'e 9. Phosphateversuslongshoredistancedata for the upper (pzp) and lower (so) portions of the Peru upwelling system.

Discussion

Glacial-InterglacialPaleoclimateScenarios The discoverythat glacial atmospheres had considerably lessCO2 thanthe interglacialatmospheres hasled to numeroustheoriesconcerningthe couplingof the atmospheric,oceanic,and terrestrialcarbonreservoirsduring the Pleistocene[e.g., Keir, 1989]. Carboncyclingsee-

ß{ ß

0.5

ß

ß

I

I

I

0

50

100

EQURTORWRRD •POLEWRRD Figure 10. Phosphateversuslongshoredistancedata for the upper (pzp) and lower (se) portionsof the northwestAfrica upwellingsystem.

the Peru systemappearsto havehigherproductivitythan a similarportion of the northwestAfrica system,despite havinglower Ekmantransport. Oregonappearsto representan intermediatecase. Ekman transportand productivity are relativelylow, yet nutrientutilizationis high. The CUE II, JOINT I, and JOINT II data setsfrom the

northwestAfrica, Peru, and Oregonupwelling systems representlong-term(2-3 month) recordsof Ekman transport and surfaceproductivity(Table t). The lack of a narios related to the ocean can be divided into two direct correlationbetweenEkman transportand surface groups:(1) alterationof the verticalflux of high-latitude productivity on short timescaleshas also been noted. nutrients,which in mm changesdeepwaternutrientand Differencesin surfaceproductivitybetweenperiodsof oxygenconcentrations and(2) changingthe flux of nutri- weak and strongupwellingwithin the samefield setting (Oregon) is discussedby Small and Menzies [1981]. ents between the continentalshelf and the deep ocean. Within the lattergroupof theoriesis the suggestion that Theseauthorsnote that weak physicalforcings(i.e. small Ekmantransportrates)often lead to surfaceproductivity low-latitudeupwellingincreased as a resultof moreintensegeostrophiccirculationin the atmosphere. This which is twice as high as that associatedwith strong second scenario is mostdirectlyrelatedto theresultsof physicalforcings.SmallandMenzies[1981]alsoemphasize thatproductivitiesoff the Oregoncoastexhibitcomthis paper. Equation(t0) containsthreevariables:Ekman trans- plex spatial and temporal differenceswhich are not port, primaryproductivity,and the amountof nutrients directlyrelatedto physicalforcings. Similar short-termvariationsin productivityhavebeen removedduring transportthroughthe Ekman loop. A observed off northwest Africa [Huntsman and Barber, plot of (t0) showshow thesethreevariablesarerelatedin 1977]. These authorsnote that strongwinds lead to a a nonlinearmanner(Figure 11). Data from modemupwelling systemssuggestthat no direct relationshipbe- deep mixed layer which tendsto inhibit photosynthesis tween the three variables can be established. The dueto light limitation. The role of suspended materialin environments also tends to lower northwestAfrica systemis characterized by a relatively nearshore photosynthesis. high Ekmantransportrateandlow nutrientconsumption, In summary,the simpleworking modelof increased whereasthe Peru systemhasrelativelylow Ekmaa transEkmantransportleadingto increased surfaceproductivity port andhigh nutrientconsumption.The inner40 km of

JEWELL: COASTAL UPWELLING MODELS 2.5

175

the form of calcareousplankton which produce rather than consumeCO2 during the growth of their tests. For this reason,the rain ratio hypothesisof Bergerand Keir [1984] in its simplest form should be viewed with caution.

Coastal zone anoxia and nutrient enrichment

1.5 '•

ß

Conditionsleadingto anoxiaand sulfate-reducing waters in coastaland near-coastalsettingshave been the subjectof considerable research. Density-stratifiedenvironmentssuchas fjords and marginalmarinebasinsare typicallyconsidered to be the mostsusceptible to anoxia [e.g., Tysonand Pearson, 1990]. Anoxia and sulfatereductionhave beendocumented in someopenoceanand coastalupwellingsettings[e.g. Goering, 1968; Codispoti and Richards, 1976; Dugdale et al., 1977]. The longshoremodelof Redfieldet al. [1963] (equations(4)-(6)) suggeststhat strongdifferentialadvectionof the surface

3000

N

ooo

1.0

1000

0.5

and undercurrents

can lead

to nutrient

enrichment

or

oxygendepletionin the lower layersof coastalupwelling zones. Barber and Smith [1981] used this model to explain the occurrenceof sulfatereductionin the Peru upAP04(p.mol/L) welling systemat 15øSin 1976. Figure11. Plotof primaryproductivity asa functionof Ekman The dataand modelspresentedhere (equations(6)-(9)) transportT and nutrientut'Qization withinthe Ekmantransport suggestthat a simple longshorenutrient model doesnot loop for assumedaverageprimaryproductivity.Shadedareas adequatelydescribethe distributionof nutrientsand oxygen in coastalupwellingzones. In a settingwith differrepresentobservednutrientdepletions(_• one standarddeviation) and a star representsvalue of averageoffshoreEkman ential longshoreadvection(Peru), subsurfacenutrient gradientsare very small (Figure 9). On the other hand, transport. P, Peru;NA, northwestAfrica; and O, Oregon. both surfaceand longshorenutrientgradientsincreasein the directionof longshoreflow off the coastof northwest and organiccarbonflux to the sedimentsduring the last Africa (Figure 10). These results suggestthat coastal glacialmaximumis not supported by the detailsof either zone anoxiaand nutrientenrichmentis muchmore likely short-termor long-term data from modem upwelling to be dependenton cross-shelfEkman transportand/or systems. An alternativeworking model might be that surfaceproductivityratherthanlongshoreadvection. high Ekmantransportratescausenutrientsto moverapidly throughthe coastalzonewheretheir incorporationinto Acknowledgments.Lou Codispotiprovided copies of the the food chain is not complete. Less intenseEkman JOINT I and II data, and Larry Small providea copy of the transportmay allow phytoplanktonto consumenutrients CUE II data. Acknowledgmentis madeto the donorsof The moreefficientlyas well as allowingzooplanktoncommu- Petroleum Research Fund, administeredby the American nities to becomeestablished.The resultis high produc- ChemicalSociety,for supportof thisresearch. tivity and organiccarbonfluxesto the underlyingwater. Anotherpossibilityis that the nutrient contentof water References feed•g the upwellingzone influencesthe productivity. For instance,the relativelynutrient-poorSouthAtlantic Baden-Dangon,A. R. F., K. H. Brink, and R. L. Smith, On the dynamic structureof the midshelfwater column off intermediate waterleadsto low productivityoff northwest 0

I

.25

I

.50

I

.75

1.00

1.25

Africa, whereas the nutrient-rich Peru-Chile Undercurrent

northwest Africa, ContinentalShelf Res., 5, 629-644,

1986. leadsto highproductivity in thePeruupwellingsystem. The massbalancemodelpresentedhere suggeststhat Bakun, A., and C. S. Nelson, The seasonalcyclesof windstresscurl in subtropicaleasternboundarycurrents,J. increased Ekmantransportdoesnot necessarily producea Phys. Oceanogr.,21, 1815-1834, 1991. linear increasein primaryproductivitywithin an upwelling zone. IncreasedEkmantransportduringthe last gla- Barber, R. T., The JOINT I expeditionof the CoastalUpwelling EcosystemsAnalysisProgram,Deep ,YeaRes., 24, cial maximum would certainly have brought more 1-6, 1977. intermediate-depth water into coastalupwellingphotic zones. A commensurate increasein surfaceproductivity Barber,R. T., and R. L. Smith,Coastalupwellingecosystems,

may have been dependenton the nutrient concentration in Analysisof Marine Ecosystems, editedby A. R. Longhurst,pp. 31-68, Academic,SanDiego, Calif., 1981. of the upwelledwater(Figure 11). Furthermore,higher Ekmantransportmay havetransported a substantial frac- Berger, W. H., Global maps of ocean productivity, in Productivityin the Ocean:Presentand Past, editedby W. tion of upwellednutrientsthroughthe highly productive coastalzoneandout into the openocean. Any increased H. Berger, V. S. Smetacek,and G. Wefer, pp. 429-454, JohnWiley, New York, 1989. productivityin the openoceanwouldlikely havebeenin

176

JEWELL: COASTAL UPWELLING MODELS

Berger, W. H., and R. S, Keir, Glacial-Holocene changesin atmospheric CO2and the deep-searecord,in ClimateProcessesand ClimateSensitivity,Geophys.Monogr. $er., vol. 29, editedby J. E. Hansenand T. Takahashi,pp. 337-351, AGU, Washington,D.C., 1984.

Hafferty, A. J,, L. A. Codispoti,andA. Huyer, JOINT-II, RN Melville LegsI, II, and IV; R/V IselinLeg II bottledata, March-May, 1977, Data Rep. 45, 779 pp., Dept. of Oceanogr., Univ. of Wash., Seattle,1978. Holland,H. D., The Chemistryof theAtmosphere and Oceans, Berger,W. H., V. S. Smetacek, andG. Wefer,Oceanproduc351 pp, Wiley-Interseiehee, New York, 1978, tivity andpalcoproductivity - An overview,in Productivity Humsman, S. A., and R. T. Barber, •ary productionoff in the Ocean.'Presentand Past,editedby W. H. Berger, northwestAfrica' The relationshipto wind and nutrient V. S. Smetacek,and G. Wefer, pp. 1-34, John Wiley, conditions,Deep SeaRes., 24, 25-33, 1977. New York, 1989. Keir, R. S., Palcoproduction and atmosphericCO2 basedon Berner, R. A., A. C. Lasaga, and R. M. Garrels, The oceanmodeling,in Productivityin the Ocean:Presentand carbonate-silicate g•chemical cycle and its effect on atPast, editedby W. H. Berger, V. S. Smetaeek,and G. mospheric carbondioxideoverthepast100m•ion years, Wefer, pp. 395-406, JohnWiley, New York, 1989. Am. J. $ci., 283, 641-683, !983. Koblents-Mishke,O. I., V. V. Volkovinsky,and Y. G. KabaBowden,K. F., PhysicalOceanography of CoastalWaters,302 nova, Planktonp•ary productionof the world ocean, in pp., Horwook, West Sussex,Great Britain, 1983. ScientificExplorationof the SouthPacific, editedby W. Boyle,E. A., Deepoceancirculation,preformednutrients,and Wooster,pp. 183-193, Nat. Acad. of Sci., Washington,D. atmosphericcarbon dioxide: theories and evidence from oceanic sediments, in Mesozoic and Cenozoic Oceans,

Geodyn.$er., vol. 15, editedby K. J. Hsu, pp. 49-59, AGU, Washington,D.C., 1986. Brockmann,C., E. Fahrbaeh,A. Huyer, andR. L. Smith,The polewardundercurrentalong the Peru coast:5 to 15øS., Deep SeaRes., 27, 847-856, 1980. Broecker,W. S., ChemicalOceanography, 214 pp., Harcourt, Brace, and Jovanovich,New York, 1974.

Codispoti,L. A., and F. A. Richards,An analysisof the horizontalregimeof denitrificationin the easterntropicalnorth Pacific,Limnol.Oceanogr.,21,379-388, 1976. Dugdale, R. C., J. J. Goering,R. T. Barber, R. L. Smith, and T. T. Packard,Denitrifieationand hydrogensulfidein the Peru upwelling system, Deep Sea Res., 24, 601-608, 1977.

Ennever, F. K., and M. B. McElroy, Changesin atmospheric CO2: Factors regulatingthe glacial to interglacialtransition, in The Carbon Cycleand AtmosphericCOz' Natural Variations,4rchean to Present, Geophys.Monogr. Ser., vol. 32, edited by E. T. Sundquistand W. S. Broecker, pp. 154-162, AGU, Washington,D.C., 1985. Epply, R. W., New production:history,methods,problems,in Productivityin the Ocean.'Presentand Past, editedby W.

H. Berger, V. S. Smetaeek,and G. Wefer, pp. 85-97, JohnWiley, New York, 1989. Friebertshauser, M. A., L. A. Codispoti,D. D. Bishop,G. E. Friederich, and A. A. Wethagen, JOINT-I, RN Atlantis hydrographicstationdata, March-May, 1974, Data Rep. 18, 243 pp., Dept. of Oceanogr.,Univ. of Wash., Seattle, 1975.

Futterer, D. K., The modern upwellingrecord off northwest Africa, in CoastalUpwelling:Its sedimentaryrecord, Part

B: Sedimentary Recordsof AncientCoastalUpwelling, edited by J. Thiede and E. Suess,pp. 105-I21, Plenum, New York, 1983. Garrels, R. M., and E. A. Perry, Jr., CyClingof carbon,sulfur, and oxygenthroughgeologictime, in The Sea, vol. 5, edited by E. D. Goldberg, pp. 303-336, WileyInterscience, New York, 1974.

Goering,J. J., Denitrifieationin the oxygenminimumlayer of the easterntropicalPacificOcean,Deep Sea Res., 15, 157-164, 1968.

C., 1970.

Lentz, S. J., The surfaceboundarylayer in coastalupwelling regions,J. Phys. Oceanogr.,22, 1517-1539, 1992.

Minas, H. J., M. Minas, and T. T. Packard,Productivity in upwellingareasdeducedfrom hydrographicand chemical fields,Limnol. Oceanogr. , 31, 1182-1206,1986. Newell, R. E., A. R. Navato,andJ. Hsiung,Long-termglobal sea surface temperaturefluctuationsand their possible influenceon atmosphericCO2 concentrations, Pure Appl. Geophys.,116, 351-371, 1978. Packard,T. T., P. C. Garfield, and L. A. Codispoti,Oxygen consumption and denitrifieationbelow the Peruvianupwe•g, in Coastal Upwelling.'Its sedimentaryrecord, Part A: Responses of the SedimentaryRegimeto Present CoastalUpwelling,editedby E. Suessand J. Thiede, pp. 147-173, Plenum, New York, 1983.

Parrish,J. T., Upwellingandpetroleumsourcebedswith reference to the Paleozoic, Am. Assoc. Petrol. Geol. Bull, 66, 750-774, 1982.

Pond,S., andG. L. Pickard,Introductory DynamicalOceanography, 329 pp., Pergamon,New York, 1983. Redfield, A. C., B. H. Ketehum, and F. A. Richards, The influenceof organismson the composition of seawater,in The Sea, vol. 2, editedby M. N. Hill, pp. 26-77, WileyInterscience,New York, 1963.

Richards,F. A. (Ed.), CoastalUpwelling,CoastalEstuarine $ci. $er., vol. 1,529 pp., AGU, Washington,D.C., 1981. Sarmiento,J. L., T. D. Herbert, andJ. R. Toggweiler,Causes of anoxiain the world ocean,GlobalBiogeochem. Cycles, 2, 115-128, 1988.

Sarnthein,M., K. Winn, J.-C. Duplessy,and M. R. Fontugne, Globalvariationsof surfaceoceanproductivityin low and

mid latitudes:Influenceon CO• reservoirsof the deep oceanand atmosphereduringthe last 21,000 years, Paleoceanography,3, 361-399, 1988.

Shaffer, G., A modelof biogeochemical cyclingof phosphorous,nitrogen,oxygen,and sulfurin the ocean:One step toward a global climate model, J. Geophys.Res., 94, 1979-2004, 1989.

Small, L. F., and D. W, Menzies,Patternsof primaryproductNity and biomass h a coastalupwellingregion,DeepSea Res., 28, 123-149, 1981.

Smith,R. L., Descriptions and comparisons of specificupwell-

•EWELL: COASTAL UPWELLING MODELS

ing systems,in Coastalupwelling,CoastalEstuarineSci. Ser., vol. 1, editedby F. A. •ehards, pp. 107-118, AGU, Washington,D.C., 1981.

Smith,R. L., A. Huyer, J. S. Godfrey,andJ. A. Church,The

177

Tyson, R. V., and T. H. Pearson,Modem and ancientcontinental shelf anoxia: an overview, in Modem and ancient continentalshelf anoxia, Spec. Publ. Geol. Soc. London 58, 1-24, 1990.

Walsh, J. J., Shelf-seaecosystems,in Analysisof Marine EcoLeeuwin Current of western Australia, J. Phys. Oceasystems,editedby A. R. Longhurst,pp. 159-195, Acanogr., 21,323-345, 1991. demic,SanDiego,Calif.,1981. Stevenson,M., D. Menzies,and L. Small, Physical/biological Wenk, T., and U. Siegenthaler,The high-latitudeoceanas a measurements off the Oregoncoast,Data Report 17, 110

pp., Dept. of Oceanogr.,Univ. of Wash.,Seattle,1975. Suess,E., Particulateorganiccarbonflux in the oceans- surface productivityand oxygen utilization, Nature, 288, 260-263, 1980.

Suess,E., L. D. Kulm, and J. S. KillingIcy, Coastalupwelling anda historyof organic-richmudstone depositionoff Peru,

controlof atmospheric CO2, in The CarbonCycleandAtmosphericCO•' Natural VariationsArchean to Present, Geophys.Monogr.Ser., vol. 32, editedby E. T. Sundquist and W. S. Broecker, pp. 185-194, AGU, Washington, D.C., 1985.

P.W. Jewel!, Department of Geology and Geophysics, University of Utah, Salt Lake City, UT 84112. (e-mail: Toggweiler,J. R., andJ. L. Sarmiento,Glacialto interglacial pwjewe!l•mines.utah.edu) changes in atmospheric carbondioxide:The criticalroleof oceansurfacewater in highlatitudes,in The CarbonCycle and AtmosphericCO•' Natural VariationsArchean to Present,Geophys.Monogr. Ser., vol. 32, editedby E. T. Sundquist andW. S. Broecker,pp. 163-184,AGU, Wash- (Received September9, 1993; revisedJanuary3, 1994;

in Marinepetroleumsourcerocks,Spec.Publ Geol. Soc.

London, 26, 181-197, 1987.

ington,D.C., 1985.

acceptedJanuary12, 1994.)