Reconstructing the origin and trajectory of ... - Wiley Online Library

JOURNAL OF GEOPHYSICALRESEARCH,VOL. 102,NO. C6, PAGES12,575-12,586, JUNE 15, 1997

Reconstructingthe origin and trajectory of drifting Arctic seaice

S.L.Pfirman, • R.Colony, 2'3D.Niirnberg, 4H.Eicken, 5andI. Rigor 6 Abstract. RecentstudieshaveindicatedthatdriftingArctic seaice playsan importantrole in the redistributionof sedimentsandcontaminants. Here we presenta methodto reconstructthe backwardtrajectoryof seaice from its samplinglocationin theEurasianArcticto its possiblesiteof origin on the shelf,basedon historicaldrift datafrom theInternationalArcticBuoy Program.This methodis verifiedby showingthatoriginsderivedfromthebackwardtrajectories aregenerallyconsistentwith otherindicators,suchascomparison of thepredictedbackwardtrajectorieswith known buoydriftsandmatchingtheclaymineralogyof sediments sampledfromthe seaice with thatof the seafloorin the predictedshelfsourceregions.The trajectoriesarethenusedto identifyregions wheresediment-laden ice is exportedto theTranspolarDrift Stream:from theNew SiberianIslands andtheCentralKaraPlateau.Calculationof forwardtrajectories showsthatthe Kara Seais a major contributorof ice to the BarentsSeaandthe southernlimb of theTranspolarDrift Stream. Introduction

Althoughstudiesshowthat centralArctic seaice originatesin a numberof different sourceareas[Colonyand Thomdike, 1985], seaice in the Arctic has typically been consideredas an undistinguishedaggregationof floes.Knowingthe sourceof individualice floesonly becameimportantwhenexpeditionsto the centralArctic rediscoveredthe large quantitiesof sedimentand other material incorporatedin the seaice originallynotedby Nansen[ 1897] and attemptedto derivetheir origin [e.g., Pfirman et al., 1989a;Abelmann, 1992; Niimberg et al., 1994]. Using the meanice drift pattern [Colony and Thomdike, 1985], clay mineralogy, and biological assemblages,the wide, shallow, Siberian shelves,and particularlythe Laptev Sea [Zakharov,1966], were identified as the probable source region of much of the sediment-ladenice observedin the EurasianArctic [Pfirman et al., 1989a; Wollenburg, 1993; Niimberg et al., 1994; Eicken et al., 1997]. Once advectedoff the shelf,the ice is transportedin the TranspolarDrift Stream,mostly exiting the centralArctic Basin severalyearslater throughFram Strait (Figure 1). Becauseof the generallyminor contributionfrom wind-blownlithogenicsediment[Pfirmanet al., 1989b], ice floes containinga large sedimentload can be assumed to have formed originally on a shallow shelf, probably in water depthslessthan 50 m where suspension freezing,includingfrazil and anchorice formation,canmostefficientlyentrainseafloorsediments [Reimnitz et al., 1992, 1993a,b]. However, asidefrom these generalconclusions,until now therewas no way to determinethe

actualtrajectoryof an individualice floe from the time that it left the shelf to the time that it was sampled.And, for ice that doesnot

containsediment, you canonly surmiseits historyfromthe general ice drift patternandlaboriousanalysisof physical,chemical,

andbiological characteristics [•rnason,1985;Abelmann, 1992]. Becauseof seasonal, interannual, and decadalvariabilityin ice drift patterns[McLarenet al., 1987;MysakandManak, 1989;Serrezeet al., 1989;Kotchetov et al., 1994;Zakharov,1994],generalized informationon ice drift is not reliablein derivingsource regionsfor individual ice floes.

In addition,there is a growingawarenessthat ice may play a role in the redistributionof contaminantsin the Arctic [Weeks,

1994; Pfirman et al., 1995a; Chemyaket al., 1996]. Widespread distributionof pollutedArctic haze,dumpingof radioactivewaste on the Siberian shelf, contaminationof thYArctic watershed,and

observations of highlevelsof heavymetalsandorganochlorines in someArctic people,polar bears,whales,and seabirdshave led to concern about transportof contaminantsacrossthe Arctic Basin [Raatz, 1991; Muir et al., 1992; Dewailly et al., 1993; Yablokovet al., 1993], by atmospheric,oceanic, and sea ice conduits.Sea

ice may be contaminatedby atmosphericdepositionand/or by entrainmentfrom the marine environment.Depositionof atmosphericpollutants[Melnikov, 1991] transportedin Arctic haze is likely to increasethe contaminantload of the snowand ice [Pfirman et al., 1995a]. Drifting ice may entrainorganochlorines from the surfacemicrolayer [Gaul, 1989]. In addition,ice formed in the shallow shelf seasoften incorporatesfine-grainedsedimentand organic material [Pfirman et al., 1989a]. Becausecontaminants, suchas organochlorines and metals,includingradionuclides,are knownto sorbpreferentiallyon fine-grainedmaterial[Stummand Morgan, 1981], if the ice forms in regionspollutedby rivers or

dumping,the entrainedsedimentis likely to be contaminated. The

1Barnard College, Columbia University, NewYork few data availabledo indeedindicatethat particle-ladensea ice 2International Arctic Climate System Study Office, Oslo, Norway contains elevated levels of some contaminants relative to the 3Also atPolarScience Center, University ofWashington, Seattle underlyingoceanwater [Campbelland Yeats,1982; Gaul, 1989; 4GEOMARResearch Centerfor MarineGeosciences, Christian AlbrechtsUniversity,Kiel, Germany

5Alfred-Wegener Institute forPolarandMarine Research, Bremerhaven,Germany

6polar Science Center, University ofWashington, Seattle

Meeseet al., 1997]. However,unlessthereis a way to determine the trajectoryand origin of the ice, it is difficult to ascertainthe sourceof any contaminationfound.

Here we show that backwardtrajectoriescalculatedfrom archives of icemotionandobserved pressure fieldscanbeusedto

Copyright1997 by theAmericanGeophysicalUnion

reconstructthe drift path of individual ice floes with some confi-

Papernumber96JC03980.

denceovera periodof severalyears.If mineralogical or other information is alsoavailable, the originandtrajectory are con-

0148-0227/97/96J C-03980509.00

strainedwith greatercertainty. 12,575

12,576

PFIRMAN ET AL.: RECONSTRUCTING THE ORIGIN AND TRAJECTORY OF ARCTIC SEA ICE

Finally, we also take advantageof the relationshipbetweenthe geostrophicwinds and the ice motion and use this informationin the analysis.The surfacegeostrophicwind is definedas

thewinddrift factor, G'(x,t) = G(x,t)- Gm(x)is thedeparture of the monthlywind field fromthe climaticmonthlymean,C(x) is the climaticmonthlymeanfieldof oceancurrents,andU*(x,t) is themonthlyfieldof ice motiondeparture fromthesumof simple

G = (Gx,Gy) = 1/(pf) (dp/dy,dp/dx)

wind-drivenresponseandoceancurrentadvection. The optimal estimateof ice motion is thusdefinedas the deviation from the guessfield of the monthlyobservations of ice veloc-

where (dp/dy,dp/dx) is the horizontalgradientof the sea level pressurefield, p is the density of the air, andf is the Coriolis parameter.As a first approximation,the surfacegeostrophic wind is determinedby the horizontalgradientof the sealevel pressure. In the mean,a stronghigh persistsover the Arctic Basin,creating what is commonly known as the Beaufort Gyre of seaice motion. In contrast,however, periods of cyclonic ice motion have been notedduring late summerand early autumn[McLaren et al., 1987; Serreze et al., 1989].

ity. In datasparseareasandmonths,theoptimalestimate at a grid point reducesto the first-guessestimateof ice motion. Since the

long-termmean ice motionreflectsroughlyequalcontributions from the meanoceancurrentsand geostrophic winds,and in the shortterm, the geostrophicwinds explainmore than 70% of the varianceof ice motion [Thorndikeand Colony,1982], in the absence of observations, the firstguessis still a goodapproximation of the ice motion.

Using the samedata from the Arctic Buoy Programfrom 1979 to 1994, Walshet al. [1995] found oktadal changesin the mean fields of pressure.Since 1988, the BeaufortHigh has decreased and shifted.As seenfrom the ice motion equationgiven below, interannualchangesin the pressurefield imply corresponding changesin the wind forcingof seaice. While the supplementof the sealevel pressuredatamakeslittle differencein areasof densebuoy coverage,it doesimprovethe analysisduringtimeswhenthe buoyarrayhaswanedto fewerthan 10 buoysand in areasof the basinwhich havebeenpoorlypopulated by buoys. The InternationalArctic Buoy Programhasdeployedbuoyson ice for trackingby satellitesince1979. Becauseobserveddrift trajectories are available only for the central Arctic, the backward calculationof trajectoriesis most certain in the central basin.As the trajectories approach the shelf break and continue on the shelves,they are often not constrainedby buoydata. However,we chose to continue the trajectoriesto the fast ice border which is located on the shelves, because this is where much of the ice

Given a samplinglocationand date, we estimatethe individual

backwardtrajectories by integrating themonthlyaveraged motion vectors.The firstpositionon thetrajectoryis the samplelocation. The nextpositionbackwardin time is foundby interpolating the velocity from one monthlyice motion field to the next. Forward

trajectories fromthesamplelocationcanalsobecalculated by this method.

Trajectory Verification We testedthe accuracyof the backwardtrajectoryanalysisin twoways:(1) by backtrackingknownbuoytrajectories and(2) by comparingthe compositionof sedimentsand diatomssampled fromtheice with thatof thepredictedsourceareas. The averagelife spanof a buoyis lessthan2 years.Usingthe lastreportedpositionof a buoy,whichwasindividuallynotincorporated into the analysis of the mean fields of ice motion, we

appliedthebackwardtrajectoryanalysisto predicttheknowndrift

for severalbuoys.For example,Figure3 showstheactualtrajec-

forms. It is clear that the final part of the backwardtrajectory tory of buoy 9372 in 1992-1994 as a solid line and the estimated betweenthe shelfbreakandthe fastice borderis subjectto greater error than in the central basin.

Ice motion

is calculated

for the entire Arctic

for each month

usingtechniquesof optimalinterpolation[Thomdikeand Colony, 1982; Gandin, 1963] as follows:

Uint(X,t) = Uguess(X,t ) +AT[ Uobs(X,t )_Uguess(X,t )] whereUint(X,t)is theoptimalestimateof icevelocityat thelocation x, for the month,t; A are the weightschosento minimize the vari-

anceof theinterpolation error;Uobs(X,t ) is the observedmonthly

velocity; andUguess(X,t) isthefirst-guess estimate oficemotion. The weights,A, are chosento minimize the varianceof the interpolationerror by solvingthe linear equation: A=M.S

whereM is the covariancematrix between observedmonthly pairsof ice velocityandS are the covariancevectorsbetweenthe observedmonthlyice velocitiesand the interpolationgrid point. The covariance matrix, M, and the covariance vectors, S, are

filled using perpendicular and parallel correlations functions

[Thomdike,1986] for monthlyobservations of ice velocityshown in Figure 2. As previouslystated,the correlationlengthscalefor monthly observationsof ice motion is 1400 km.

Thefirst-guess estimate, Ugues s,isdefined as

Figure3. Exampleof reconstructed (dottedline)andactualtra-

jectories (solidline)ofbuoydriftback836days. Forthisexample

Uguess(X,t) = Am(x)G(x,t) + C(x)+ U*(x,t) whereAm(x) is the climatic monthly mean spatial pattern of

thereconstructed drift deviated fromtheactualdriftby 204 km after2 years.Typicaldeviationsfrom the reconstructed driftsfrom

theactual trajectories werefound tobe100kmyear -1.

PFIRMAN ET AL.: RECONSTRUCTING THE ORIGIN AND TRAJECTORY OF ARCTIC SEA ICE

12,577

backwardtrajectoryfrom the analysisas a dottedline. Two years berg, 1972; 13bllenburg,1993]. To the eastand west of theseriver after, the difference between actual and predicted position is inputs, smectiteconcentrations decreasein seafloorsedimentsof 204 km. We estimatethat the uncertaintyof points on the back- both the Kara and Laptev Seas, althoughthere are some local

wardtrajectories grows atabout 100kmyr--1. We comparedthe characteristics of sedimentsampledfrom the ice (duringexpeditionsin 1987, 1988, 1989, 1990, and 1991) with those obtained from the seafloorin the predicted sourceregion [Silverberg,1972; Wollenburg,1993; Dethleffet al., 1993; N//rnberget al., 1994]. Eachtrajectorywastruncatedat the fastice border on the shelf, becausethe coastalpolynya along the seaward edgeof the fast ice is a major regionof ice production[Zakharov, 1966; Dethleffet al., 1993]. Also, the polynyais typicallylocated in a water depthof about20 to 50 m, which is shallowenoughto allow for entrainmentof seafloorsedimentsthrough suspension freezing [Reimnitzet al., 1992, 1993a]. In the processof suspensionfreezing,which occursin shallowice-free seasduringstorms, strongmixing resuspends sediment,andrapid freezingsupercools the water column. Growing ice can thus come into contactwith sedimentparticlesandorganisms eitherin the watercolumn(frazil

smectite sourcesin the southwesternKara Sea [Shelekova et al., 1995]. As smectite concentrationsdecrease,concentrationsof illite

ice) or at the seafloor(anchorice), bond to them, and raise them to the ice sheetswhich ultimatelyform at the surface[Reimnitzet al.,

increase(Figure 4). Once the ice is trackedback to the Kara and Laptev Seas, the steep mineralogicalgradientsin smectiteand illite aid in delineating specific sedimententrainmentlocations. The other two clays, chlorite and kaolinite, do not have nearly as muchregionalvariationalongthis part of the Siberianmargin. Clay mineralogyin the East SiberianSea is relativelyuniform, not allowing differentiationof sourceregion [Wollenburg,1993] without recourseto other data, e.g., heavy mineral assemblages [Silverberg,1972; Naugler et al., 1974; Darby et al., 1989; Stein et al., 1994]. Relatively few trajectoriestrackbackto this region,a result which supportsother studiesindicatingthat the East Siberian Sea is not a major sourceof sea ice to the EurasianBasin [Zakharov, 1966], at least during the years representedin this study. Note that severalsamplestrackedbackto about157øEfit poorly (Figure 4). This can be explainedbasedon the uncertainty(grow-

1992].

ingatapproximately 100kmyr-4inthecentral Arctic) intheback

Comparisonof the clay mineralogyof the Siberian shelf calculationand the lack of buoy drift data on the shelves.Looking seafloors and sea ice sediments shows that the individual backward at the reconstructed trajectories(Plate 1), it appearsthat if the tratrajectoriesgenerallyfit with characteristics of knownsourceareas jectorieshad been shiftedslightlyto the west,they wouldhavetereither at the terminationat the fast ice borderor alongtheir path on the shelves(Plate 1 and Figure 4). Some of the scatteris due to measurement precisionandmethodologicaldifferencesin calculation of clay mineralogy.Unfortunately,dataon precisionandinterlaboratory comparisonsare not available. According to D.A. Darby (personalcommunication,1996), while the valuescan differ by as much as 40% dependingon the method,mountingtechnique, and amount of material analyzed,the error on averageis

minated to the west of the New Siberian Islands, rather than to the east in the East Siberian

Sea. The result of this shift would be a

muchbetterfit with the seafloormineralogy.Similarly,two trajectories that we tracked back to the fast ice border in the western

LaptevSea (approximately110øE)actuallypassoverseafloorsedimentswith similar compositionin the easternLaptevSea nearthe New SiberianIslands.It is likely that the sedimentwith intermediate levels of smectite was entrained there, rather than in the west-

ern Laptev Sea. The trajectoryanalysiscan be usedto estimatethe age of sedi20% couldbe reportedas 16-24% by variousinvestigators. For comparisonwith the individualtrajectories,we also show ment-ladenice floes.Most sea ice with significantsedimentload the backwardtrajectories(Plate 2) calculatedusing the averaged has entrained resuspendedseabed sediments[Pfirman et al., ice drift (Figure 1). The temporallyspecifictrajectories(Plate 1) 1989a, b; Reimnitz et al., 1992, 1993a, b]. Therefore we can conclearly do a better job of matchingthe sourcemineralogy,thus cludethatthe floe originatedon the shelf.Assumingthatthe ice is demonstrating the significanceof seasonaland annualvariationsin advectedoff the shelf after forming in the coastalpolynya/flaw ice drift. However,the calculationsusingthe meando illustratethe lead, the approximateageof sediment-ladenfloescanbe estimated possibilityat timesfor transportby ice from the Kara to the Laptev from the time of samplingto the "termination"of the backward SeathroughVilkitsky Strait southof SevernayaZemlya (discussed trajectoryat the fast ice border.The averageage of seaice tracked also by Pavlov and Pfirman [1995] and Pfirman et al. [1995b, back to the fast ice borderis 35 months,or 2.9 years(Figure4). The age of seaice without sedimentsis not as clear:the ice could 1997]. Smectite,a clay mineralformedin part by weatheringof basalt, have formed on a shelf and not entrained sediments, or it could showsthe most variability in sedimentson the Siberian shelves. haveformedwithin the centralArctic Basinoverdeepwater. Analysisof biogenicremains(Plate 3) supportsthe trajectory The prime sourceof smectiteto the Siberianseasappearsto be that showthat ice with very high smectiteconcenweatheringof the floodbasaltsandtuffscoveringthe CentralSibe- reconstructions rian Plateau [Zolotukhin and Al'mukhamedov,1988; Wahsnerand trations(up to 63%) may haveoriginatedin the southernKara Sea, Shelekhova,1994]. (Naidu et al. [1995] similarly attributedthe offshoreof the Yeniseyand Ob' Rivers.The percentplanktonic occurrenceof expandableclay mineralsin seafloorsedimentsof freshwaterdiatom speciessampledfrom the ice reflectsboth envithe Bering Sea to weatheringof andesitic/basalticrocks on the ronmentalconditionsduring ice formation and later modification Kamchatka and Alaskan Peninsulas.)The Central Siberian Plateau [Abelmann,1992]. Samplesfrom severalof thesefloeswith high containedan unusuallyhigh concentration separatesthe watershedsof the Kara and Laptev Seas.Yenisey smectiteconcentration River sediments,fed by westerly drainagefrom this region, con- of freshwaterplanktonic diatom species(Plate 3). Most of the tain the highestsmectitelevel, comprisingup to 70% of the clay other sea ice sampleshad a significantfraction of benthic marine mineral assemblagein the confluenceregion of the Yenisey and species,similarto the speciesreportedfrom shallowwaterareasof Ob' Rivers of the Kara Sea [Wahsnerand Shelekhova,1994]. To theSiberian shelves [Abelmann, 1992].However, thetwosamples the east, in the Laptev Sea, the KhatangaRiver also drainsthe thattrackedbackto the KaraSeahad70%).Abelmann[1992]interpreted thisunusualassemblage as decreasingto less than 50% offshoreof the river mouth [Silver- representing formationin a regionwith large freshwaterinflux, +20 % of the actual value. This means that a smectite value of

12,578

PFIRMAN ET AL.' RECONSTRUCTINGTHE ORIGINAND TRAJECTORYOF ARCTIC SEAICE 90 ø E 1979 - 1994 10 cm/s =

Kara i

i

'Sea•'"•'l: ) \

!

\

!

-,.

East

,

!

Barents

Sea-'

.' - '813•N'

', o

180 ø

90 ø W

Figure1. MeanfieldoficedriftintheArcticOcean derived frombuoydriftfrom1979to 1994.Velocities areindicatedbyarrows (seescaleonfigure). Numbered linesindicate theaverage number of yearsrequired foricein this locationto exit theArctic throughFram Strait.

Calculation of Backward Trajectories As observedby Nansen [1902], Zubov [1943], and others,sea ice motion in the Arctic Basin is primarily forcedby local surface winds and oceancurrents.Nansenobservedthat compactice tends to drift about 35ø to the right of and at 2% of the surfacewind speed.At lower ice concentrations, the ice drift can reach3-4% of the surfacewind speed.On daily timescales,Thorndikeand Colony [1982] found typical winter valuesfor the multiplier IAI of 0.008 and a turning angle theta of-18 ø and determinedthat the geostrophicwinds accountfor 70% of the varianceof the ice

Thorndikeand Colony[ 1982].This longercorrelation lengthscale allows for better estimatesof the monthly fields than the daily fields.Giventhatice velocityis a vector,two autocovariance func-

tionsarerequiredto describethelongitudinal andtransverse relationships. Estimates of theseautocovariance functions werefound by resolvingthevelocitiesof buoypairsintocomponents perpendicularandparallelto thelinesconnecting thebuoysandcorrelating the resultingcomponents.

motion.

To produceanalyzedfieldsof ice motion,Thorndikeand Colony [1982] used techniquesof optimal interpolation[Gandin, 1963] which require the specificationof certain statisticsinherentin the data, suchas the mean field of ice motion (Figure 1), the variance

II

of the individual observations about this mean, and correlation

functionswhich quantify the statisticalrelationshipbetweenseparate observations.For this studywe extendthe work of Thorndike and Colony [ 1982] and apply the techniquesof optimalinterpolation to monthlyratherthan to daily observations of ice drift. The advantagesof usingmonthlydata are that the measurement 0 500 1000 1500 2000 error of ice velocitiesis greatly reducedand the correlationlength SeparationDistance scale (Figure 2) between observationsis much longer.Since the error in the positionof a buoyby satelliteis typicallyabout300 m Figure 2. Observedcorrelationsbetweenmonthly ice velocity [ArgosIncorporated,1996], the approximateerrorin monthlyice components as a functionof distance.The dotsshowthe average velocities is correlationsof the parallelcomponentsof severalhundredpairsof buoysin 200-km bins.The trianglesshowthe perpendicular correlations of these same pairs. The solid lines show the estimated (2x m = 0.02cms monthly parallel and perpendicularcorrelationfunctionsfor ice eu = 30d motion.For comparison,the correlationfunctionsfor daily obserFor the monthly observationswe found a correlationlength vationsare shownas shortdashedlines [Thomdikeand Colony, scaleof 1400 km (Figure 2) as comparedto 900 km found by 1982]. i

J

3002 ) 2

-!

i

i

i

i

PFIRMAN ET AL.' RECONSTRUCTING THE ORIGIN AND TRAJECTORY OF ARCTIC SEA ICE

I /

12,579

12,580

PFIRMAN ET AL.' RECONSTRUCTING THE ORIGIN AND TRAJECTORY OF ARCTIC SEA ICE

!!

i I

XC > \

-

V

_

PFIRMAN ET AL.: RECONSTRUCTING THE ORIGIN AND TRAJECTORY OF ARCTIC SEA ICE

I !

! I

t ! I

! I

I

1 I

I

I I

I

I

I I I

I 1 I I

I

I I

! t

I I i I I

I

! I I

I t

I

t

t 1 t

! !

t-

\ \

12,581

12,582

PFIRMAN ET AL.: RECONSTRUCTING THE ORIGIN AND TRAJECTORY OF ARCTIC SEA ICE

I !

%

%

I

•0 I t I

t I

! i

1.

/

PFIRMAN ET AL.' RECONSTRUCTING THE ORIGIN AND TRAJECTORY OF ARCTIC SEA ICE

•ø o o O

12,583

12,584

PFIRMAN ETAL.:RECONSTRUCTING THEORIGIN ANDTRAJECTORY OFARCTIC SEAICE LaptevSea

ment is redistributedon the surface of floes by meltwater and winds, resultingin unevenconcentrationseven on one floe. As a result, sedimentconcentrationdata are quite variable and cannot be used for detailed analysis,althoughwe believe that, in aggregate and spanning5 ordersof magnitude(Plate 4), they do representgeneraltrends.

EastSiberianSea

::::::::::::::::

::::::::::::::::::: :.:.:.:.:.:,:.:.:.:

2•3

::::::::::::::::::

35

Earlier

::::::::::::::::::

,:.:.:.:.:.:.:.:.

x x xi!!iiiiiiiiii!•;• • xXx• xx

:::::::::::::::::: :::::::::::::::::: ::::::::::::::::::

::::::::::::::::::: ::::::::::::::::::: :::::::::::::::::::

12"O•i"g]tude

:.:.:.:.:.:.:.:.:..

::::::::::::::::::

?:iii::?:iiiii!iii • :::::::::::::::::

. X

:::::::::::::::::::::::::: x •

• 41::::::::::::::::::::::::::::::::::::

•3

studies indicated

that most of the sediment-laden

ice

sampledin theEurasianArctic camefrom the LaptevSea[Wollenburg, 1993;Niirnberget al., 1994;Steinet al., 1994;Eickenet al., 1997). The backwardtrajectoriessupportthis in showingthat sea ice with the greatestsedimentconcentrationscrossedthe shelf edgeprimarily in the vicinity of the New SiberianIslandswhich protrudesinto the Arctic Basin (Plate 4), separatingthe broad, shallowshelvesof the Laptev and the East SiberianSeas.In addition, this studyalsoidentifiesthe north-centralto easternKara Sea as a secondarysourcefor sediment-ladenice that movesinto the southernpart of the EurasianArctic. These partsof the Kara Sea are shallow,perhapsallowing sedimententrainment,and there is an offshoremovementof ice resultingin its export to the Arctic Basin.As pointedout by otherinvestigators [Vize, 1937;Niirnberg et al., 1995], the southwestern Kara Sea doesnot appearto be an

ii:::.•".•i:iiii?•

::::::::::::::::::::::::: X

active source area for ice to the Arctic Ocean.

::::::::::::::::::::::::::

Anothercharacteristicof seaice sedimentis that it is primarily fine-grained,generallyclayey silt or silty clay (up to 75% silt and 120 140 160 180 90% clay [Niirnberget al., 1994]). Comparisonof thegrainsizeof seafloorsedimentsrelative to the sedimentsin the pack ice indiLongitude catesthat sedimentincorporatedin the pack ice is finer-grained Figure 4. Concentrations of the clay mineralssmectiteandillite than seafloorsedimentsin the sourcearea [Wollenburg,1993; in seafloorsediments(crosses)are plottedagainstthe longitudeof with the reconstructedorigin for sea ice samples(numbersindicate Niirnberget al., 1994] (Plate5). This differenceis consistent freezing as the main mode of sedimententrainment duration(in months)from the samplelocationto the fast ice bor- suspension der). If the fit were perfect,the two shouldmatchwithin the error [Reimnitzet al., 1993b]. Fine sedimentparticlesremainsuspended of the clay mineralogies(approximately+20 % of the actualvalue in the water column longer than coarseand are more likely to be (D.A. Darby, personalcommunication,1996). Concentrations of incorporated into growingice; conversely, fine sediments are winthe otherclay minerals,chloriteand kaolinite,do not vary signifi- nowed from the seabed.Size fractionationis important,because cantlyin the LaptevandEastSiberianseas.Clay mineralogydata most contaminantsof concernin the Arctic, organochlorines and are from Wollenburg[ 1993] andNiirnberget al. [ 1994]. heavy metals, including radionuclides,sorb preferentiallyonto :::::::::::::::::: ::::::::::::::::::

'

i

!

,

i

fine-grained material[Stummand Morgan, 1981].Thereforesediment-ladenseaice is likely to haveconcentrations of contaminants greaterthan thoseof shelf sediments,which are "diluted"with mostlikely in the vicinity of river discharge,suchas the Ob' and/ higherproportions of sand[Pfirmanet al., 1995b]. or YeniseyRivers. Forward trajectoriescan also be calculatedto identify regions influencedby seaice from different sourceareas.Simulationsof Discussion Lagrangiandrifterswere startedalongthe mid-September multiyear ice edge each month from 1979 to 1994 and were allowed to Once verified, we can use the backwardtrajectoriesto understandbetterthe characteristics of entrainedmaterial.For example, drift forward in time as long as the drifter remainedwithin the sedimentconcentrations might providean indicationof sourcesof boundsof our velocity fields, shownon Figure 5 as a rectangular sediment-laden ice. Note, however, that the sediment concentra- box, or 12 years, whichevercame first (Figure 5). Note that, as tion data used in this study reflect both the amountof sediment with the backwardtrajectoryanalysis(Plate 1), trajectorieswere entrained as well as the amount of ice ablated during drift. The not allowedto passthroughislands.Ice from the Kara Seahasa samplesusedfor analysisof sedimentconcentration weretypically stronginfluenceon the Laptev Sea, BarentsSea, Svalbard,the obtainedby scrapingsediment-richice from theupperseveralcen- southernportionof the TranspolarDrift Stream,andeasternFram Strait. Ice from the Laptev Sea is mostly advectedthroughFram timeters of the floe. Data are not available on total sediment load per unit area of ice, which would havebeena bettermeasureof Straitandto a lesserdegreeinto the BarentsSea.Ice from theEast total sedimententrainment.Most of the ice sampledwas several SiberianSeais eitheradvectedthroughFram Straitor is caughtup yearsold (Figure 4), and about40 cm of ice will be lost from the in the BeaufortGyre and is transportedalongthe northernNorth American coast. floe surfaceeach year [Romenov,1993]. As the ice ablates,sediment distributedthroughoutthe original sectionof ice collapses into a sediment-richlayer on the ice surfacewhich may represent Conclusions much of the sediment load carded by the ice [Pfirmen et el., This studyshowsthat usingan analysisof backwardtrajectories 1990; Wollenburg,1993;Niirnberget el., 1994].Anothersourceof uncertaintyis the amountof materiallost to the underlyingwater calculatedindividuallyfor eachseaice sampleand availablemincolumnduring drift by runoff, brine drainage,andotherprocesses eralogicalandbiologicaldata,the sourceof Arctic seaice, aswell [Pfirmenet el., 1990; Reimnit• et el., 1993b].Furthermore,sedi- as its trajectoryand possibleage, i.e., time elapsedsinceleaving

PFIRMAN ET AL.: RECONSTRUCTING THE ORIGIN AND TRAJECTORYOF ARCTIC SEA ICE

12,585

in ice, its likely areaof release,possibleuptakeby ice-associated organisms, andentranceto theArcticecosystem. For example,this study showsthat sea ice from the Kara Sea contributessignificantlyto the ice coverof the BarentsSeaandthe so•athern partof the EurasianBasin.The Kara Seahasthe potentialto be contaminatedby agriculturalandindustrialchemicals:it liesjust northof theindustrialcomplexof Nori!sk,Russia,anda mainatmospheric transportconduitfor Eurasiaasourcesof Arctic haze passesover the Kara Sea [Raatz,1991;Shaw,1995].Riversdischarging to the Kara Seadraina largewatershedwith a varietyof potentialcontaminant sources[Pavlov and Pfirman, 1995]. The amount and fate of material entrainedin sea ice formed in this region are importantto determineby futurestudies. Acknowledgments.This researchwas supportedby the U.S. Office of Naval Research, the U.S. National Science Foundation, and, under the direction of J. Thiede (GEOMAR, ResearchCenter

for Marine Geosciences, Kiel, Germany),the GermanBundesministerium ftir Forschungund Technologie.We thank Terry Tucker for his helpful review of the manuscript. References

Abelmann, A., Diatom assemblagesin Arctic sea ice--Indicator for ice drift pathways,Deep SeaRes.,Part A, 39, specialissue,S525S538, 1992.

Argos Incorporated,Argos Users Handbook, Landover, Maryland, 1996.

Arnason, B., Theuseof deuterium to tracetheoriginof drifting ice, Rit FiskideildarJ. Mar. Res.Inst. Reykjavik,10, 85-89, 1985. Campbell, J.A., and P.A. Yeats,The distributionof manganese,iron, nickel, copperand cadmiumin the watersof Baffin Bay and the CanadianArctic Archipelago,Oceanol.Acta, 4(2), 161-168, 1982. Chernyak, S.M., C.P. Rice, and L.L. McConnell, Evidence of currently-usedpesticidesin air, ice, fog, seawaterand surfacemicrolayer in the Bering and Chukchi Seas, Mar. Pollut. Bull., 32(5), 410-419,

1996.

Colony,R., and A.S. Thorndike,Sea ice motion as a drunkard'swalk, J. Geophys.Res.,90, 965-974, 1985. Darby, D.A., A.S. Naidu, T.C. Mowatt, and G. Jones,Sedimentcomposition and sedimentaryprocessesin the Arctic Ocean, in The Arctic Seas,editedby Yvonne Herman,pp. 657-720, Van Nostrand Reinhold, New York, 1989.

Dethleff, D., D. Ntirnberg, E. Reimnitz, M. Saarso, and Y.P. Savchenko,East SiberianArctic regionexpedition'92: Its significance for Arctic sea ice formation and transpolarsedimentflux, Ber. Polarforschung,120, 1-44, 1993. Dewailly, W., P. Ayotte, S. Bruneau, C. Lalibertd, D.C.G. Muir, and Figure5. Thisfigureshowsthepossible seasof originof samples R.J. Norstrom, Inuit exposure to organochlorinesthrough the takenfromseaice in theArcticBasin.Simulations of Lagrangian aquatic food chain in Arctic Qu6bec,Environ. Health Perspect., drifterswerestartedalongthe mid-September multiyearice edge 101(7), 618-620, 1993. each month from 1979 to 1994 and were allowed to drift forward Eicken, H., E. Reimnitz, V. Alexandrov, T. Martin, H. Kassens, and in time as long as the drifter remainedwithin the boundsof our T. Viehoff, Sea-iceprocesses in the Laptev Sea and their imporvelocityfields.The top panelshowsexportfromtheKara Sea,the tance for export of sediments,Cont. Shelf Res., 17(2), 205-233,

middlepanelshowsexportfrom the LaptevSea,and the bottom 1997. panelshowsexportfrom the East SiberianSea. Gandin, L.S., The Objective Analysis of Meteorological Fields, Hydrometeorol. Publ. House, St. Petersburg, Russia, 1963.

(Englishtranslation,Is. Programfor Sci. Trans.,Jerusalem,1965.) Gaul, H., Organochlorinecompoundsin water and seaice of the Eurothe fast ice border,can be distinguished in the EurasianArctic. pean Arctic Sea, in Proceedingsof the Conferenceof the Comitd Althoughthe methodhasbeendemonstrated and verifiedusing ArctiqueInternationalon the Global Significanceof the Transport sediment-laden ice, trajectoriescanalsobe estimatedfor ice withand Accumulationof PolychlorinatedHydrocarbonsin the Arctic, out sediment;it is just moredifficultto verify their accuracy. editedby F. RootsandR. Shearer,pp. 18-22, Plenum,New York,

Usingreconstructed backwardandforwardtrajectories, we can 1989. identifyregionsof sediment entrainment, mainpathways of sedi- Kotchetov,S.V., I.Y. Kulakov,V.K. Kurajov,L.A. Timokhov,andY.A.

ment-ladenice, and regionsof possiblereleaseof sedimentsfrom Vanda,Hydrometeorological Regimeof the LaptevSea, Arct. and the ice, thus contributingto understanding Arctic sedimentflux. Antarct.Res.Inst., St. Petersburg, Russia,1994. Similarly,we canassess thepotentialfor contaminant entrainment McLaren, A.S., M.C. Serreze,and M.G. Barry, Seasonalvariationsof

12,586

PFIRMAN ET AL.: RECONSTRUCTINGTHE ORIGINAND TRAJECTORYOF ARCTIC SEAICE

sea ice motion in the Canada Basin and their implications,Geophys.Res.Lett., 14, 1123-1126, 1987. Meese, D., E. Reimnitz, W. Tucker,A. Gow, J. Bischof,andD. Darby, Evidence for Cesium-137 transportby Arctic sea ice, Sci. Total Environ.,in press,1997. Melnikov, S.A., Report on heavy metals,StateArct. Environ.Rep., 2, pp. 82-153, Arct. Cent., Univ. of Lapland, Rovaniemi,Finland,

Romanov,I.P., TheIce Coverof theArcticOcean.,211 pp.,Arct. and Antarct.Res.Inst., St. Petersburg, Russia,1993. Sen'eze,M.C., R.G. Barry, and A.S. McLaren, Seasonalvariationsin sea ice motion and effects on sea ice concentration in the Canada

Basin,J. Geophys.Res.,94, 10,955-10,970, 1989. Shaw,G.E., The Arctic haze phenomenon,Bull. Am. Meteorol. Soc., 76(12), 2403-2413, 1995.

1991.

Shelekova,E.S., M.A. Levitan, Y.A. Pavlids,D. Ntirnberg,and M. Muir, D.C.G., R. Wagemann, B.T. Hargrave, D.J. Thomas, D.B. Wahsner,Clay mineral distributionin surfacesedimentsof the Peakall, and R.J. Norstrom,Arctic marine ecosystemcontaminaSouthwesternKara Sea, Oceanology,35(3), 403-406, 1995. tion, Sci. Total Environ., 122, 75-134, 1992. Silverberg,N., Sedimentologyof the surfacesedimentsof the East Mysak, L.A., and D.K. Manak,Arctic sea-iceextentand anomalies, Siberianand Laptev Seas,Ph.D. thesis,185 pp, Univ. of Wash., 1972. 1953-1984, Atmos. Ocean, 27(2), 376-405, 1989. Naidu, A.S., M.W. Han, T.C. Mowatt, andW. Wajda, Clay mineralsas Stein,R., H. Grobe, and M. Wahsner,Organiccarbon,carbonate,and indicatorsof sourcesof terrigenous sediments, theirtransportation clay mineraldistributions in easternCentralArctic Oceansurface sediments,Mar. Geol., 1!9, 269-285, 1994. anddeposition: BeringBasin,Russian-Alaskan Arctic,Mar. Geol., 127, 87-104, 1995. Stumm, W., and J.J. Morgan, Aquatic Chemistry,John Wiley, New York, 1981. Nansen,F., Farthest North, Archibald Constable,London, England, 1897. Thorndike, A.S., Kinematics of sea ice, in Air-Sea-Ice Interactions, editedby N. Untersteiner,pp. 489-549, Plenum,New York, 1986. Nansen,F., NorwegianNorth Polar Expedition1893-1896,Scientific Results, vol. 3, The Oceanographyof the North Pole Basin, Thorndike,A.S., and R. Colony, Sea ice motion in responseto geostrophicwinds,J. Geophys.Res.,87, 5845-5852, 1982. 427 pp.,Longmans,Green,Toronto,Ont.,Canada,1902. Naugler,F.P.,N. Silverberg, andJ.S.Creager,in MarineGeologyand Vize, V.J., Ice drift from the Kara Sea into the Greenland Sea, Prob. Arck., 1,103-116, 1937. Oceanographyof the Arctic Seas,editedby Y. Herman,pp. 191Wahsner,M., and E.S. Shelekhova,Clay-mineraldistributionin Arctic 210, Springer-Verlag, New York, 1974. deepseaand shelf surfacesediments(abstract),Greifswald.Geol. Ntirnberg,D., I. Wollenburg,D. Dethleff, H. Eicken, H. Kassens, Beitr., A(2), 234, 1994. T. Letzig,E. Reimnitz,andJ. Thiede,Sediments in Arcticseaice: Implicationsfor entrainment,transportand release,Mar. Geol., Walsh, J. E., W. L. Chapman,and T. L. Shy, Recentdecreaseof sea 119, 185-214, 1994. level pressurein the CentralArctic Basin,J. Clim., 9, 480-486, 1995. Ntirnberg,D., M.A. Levitan,J.A. Pavlidis,andE.S. Shelekhova, Distributionof clay mineralsin surfacesedimentsfrom the eastern Weeks, W., Possible roles of sea ice in the transportof hazardous

Barents and south-western Kara Seas, Geol. Rundsch., 84, 665-

material, Arct. Res. U.S., 8, 34-52, 1994.

Wollenburg,I., Sedimenttransportby Arctic seaice: The recentsediment load of lithogenic and biogenic material, Ber. PolarPavlov,V., and S.L. Pfirman,Hydrographicstructureandvariabilityof forschung,127, 1-159, 1993. the Kara Sea: Implicationsfor pollutantdistribution,Deep Sea Res., Part II, 42(6), 1369-1390, 1995. Yablokov,A.V., V.K. Karasev,V.M. Rumyantsev,M.Y. Kokeyev,O.I. Pfirman, S.L., J.-C. Gascard,I. Wollenburg,P. Mudie, and A. AbelPetrov,V.N. Lytsov,A.F. Yemelyanenkov, andP.M. Rubtsov,Facts mann, Particle-laden Eurasian Arctic sea ice: Observationsfrom and ProblemsRelatedto RadioactiveWasteDisposalin SeasAdjaJulyandAugust1987,Polar Res.,7, 59-66,1989a. centto the Territoryof theRussianFederation(alsoknownas"The Pfirman,S.L., I. Wollenburg,J. Thiede,and M.A. Lange,Lithogenic YablokovReport"),72 pp., Off. of the Pres.of theRuss.Fed.,Moscow, 1993 sedimenton Arctic pack ice: Potentialaeolianflux and contribution to deepsea sediments,in Paleoclimatology and Paleometeo- Zakharov,V.F., The role of flaw leadsoff the edgeof fast ice in the rology: Modem and Past Patterns of Global Atmospheric hydrologicalandice regimeof the LaptevSea,Oceanology, 6(1), 815-821, 1966. Transport,NATOASI Ser.C, vol. 282, editedby M. LeinenandM. Zakharov, V.F., On the character of cause and effect relationships Sarnthein,pp. 463-493, KluwerAcad.,Norwell, Mass., 1989b. betweensea ice and thermal conditionsof the atmosphere,Ber. Pfirman,S., M.A. Lange,I. Wollenburg,andP. Schlosser, Seaice characteristicsand the role of sedimentinclusionsin deep-seadeposiPolarforschung,144, 33-43, 1994. tion: Arctic-Antarcticcomparisons,in GeologicalHistory of the Zolotukhin,V.V., andA.I. Al'mukhamedov,Trapsof the SiberianplatPolar Oceans: Arctic VersusAntarctic, edited by U. Bleil and form, in ContinentalFloodBasalts,editedby J.D. Macdougall,pp. 273-310, Kluwer Acad., Norwell, Mass., 1988. J. Thiede,pp. 187-211, Kluwer Acad.,Norwell, Mass., 1990. Pfirman, S.L., H. Eicken, D. Bauch, and W. Weeks, The potential Zubov, N.N., Arctic Ice (in Russian),Glavsevmorputi, Moscow, 1943. (Englishtranslation,Transl. 103, U.S. Navy Electr.Lab., transportof pollutantsby Arctic sea ice, Sci. TotalEnviron.,159, SanDiego,Calif., 1943.) 129-146, 1995a. Pfirman,S.L., J. Koegler,andBriceAnselme,Coastalenvironments of the westernKara and easternBarentsSeas,Deep Sea Res., Part II, R. Colony,International Arctic ClimateSystemStudyOffice,Post42(6), 1391-1412, 1995b. Pfirman,S.L., J.W. KOgeler,and I. Rigor,Potentialrapidtransportof boks5072 Majorstua,N-0301 Oslo,Norway H. Eicken,Alfred-Wegener Institutefor PolarandMarineResearch, contaminantsfrom the Kara Sea, Sci. Total Environ., in press, 682, 1995.

1997.

Columbusstrasse, Postfach 120161,Bremerhaven, D-27515,Germany

Raatz,W.E., The climatologyandmeteorologyof Arctic air pollution, Do Ntirnberg,GEOMAR ResearchCenterfor Marine Geosciences, in Pollution of the Arctic Atmosphere,edited by W.T. Sturges, ChristianAlbrechtsUniversity,Wischhofstrasse 1-3, D-24148 Kiel, pp. 13-42, Elsevier,New York, 1991. Germany Reimnitz, E., L. MarincovichJr., M. McCormick, and W.M. Briggs, SoPfirman,BarnardCollege,ColumbiaUniversity, 3009Broadway, Suspension freezingof bottomsedimentandbiotain theNorthwest NewYork,NY, 10027.(e-mail:[email protected]) Passageand implicationsfor Arctic Oceansedimentation, Can.J. I. Rigor,PolarScienceCenter,Universityof Washington, 1013NE Earth Sci., 29, 693-703, 1992.

Reimnitz, E., M. McCormick, K. McDougall, andE. Brouwers,Sedi-

40thStreet,Seattle, WA,98105(e-mail:[email protected])

mentexportby ice raftingfrom a coastalpolynya,ArcticAlaska, U.S.A.,Arct. Alp. Res.,25(2), 83-98, 1993a. Reimnitz,E., P.W. Barnes,andW.S. Weber,Particulatematterin ice of theBeaufortGyre,J. Glaciol.,39(131), 186-198, 1993b.

(Received May 3, 1995;revisedNovember 26, 1996; acceptedDecember2, 1996.)