The Arctic ocean-ice system studied by contamination m.odelling

Annals cifGlaciology 25 1997 © Internationa l Glaciological Society

The Arctic ocean-ice system studied by contamination m.odelling LARISSA NAZARENKO,I ,2 TESSA SO U,I MI C HAEL EBY,3 GR EG HOLLOWAy l

IInslill/le cif Ocean Sciences, 9860 IVesl Saanich Road, Po. Bo.\' 6000, Sidne)" Brilish Columbia V8L 4B2, Canada 2Institule cif Northern EcoLogy Problems, 14a Femnan Slreel, ApaliO' 184200, Rllssia 3 School cif Earth and Ocean Sciences, University cif Victoria, Po. Bor 1700, , 'icLoria, B ritish Columbia VBTI' 3P5, Canada

ABSTRACT

The Arctic represe nts a relatively pristine frontier that is vuln erable to pollution. Substances origin ating at mid latitudes are transported to the Arctic by atm ospheric processes, ocean currents and riYers. These pollutants can acc umul ate in the Arctic environment Testing of nucl ear weapons, dumping of waste a nd operation of ships, a nd nuclear p ower pl a nts also pose threats. To inves tigate possible pollutant pathways we used a multi-level primitive-equ a ti on ocea n mod el, coupl ed to a dynamic- thermodynamic sea-i ce model. Coupling included conser va tive tra nsfer of mome ntum, heat a nd fresh water. Atmospheric rorc ing (wind stress, te mpera ture, humidit y, radiation a nd heat fresh-w a ter Ouxes ) was supplied by datasets or bulk formulae. Open late ra l-bou ndary co nditi ons for the ocea n model were suppli ed either by d atase ts (temperature a nd sa linity) or from a large r-sca le ocean model (momentum ). Two integratio ns were compa red - one used a centred-diITerence ad\'ecti on sch eme a nd \'isco us damping, while the oth er used a better represe nta ti on of an advec tion sc heme and a sub-g rid sca lc edd y parameterization. Tracer simu lations showed (I) the importance of good representa ti on of num erica l advec tion, a nd (2) th e role of edd y interacting with sea-n oo r topography (the neptune effec t).

INTRODUCTION

MODEL DESCRIPTION

For und ersta ndin g the Arctic ice- ocea n system it is necessar y to develop co upled Ice- ocean models that ca n re produce climato logical ch a nges in appropriate ways. Th e occ urrence of ice aITects th e fo rcing of ocean circulation . Likewise, heat a nd salt tra nsport in th e ocean helps determ ine the long-term th ickness a nd ex tent of sea ice. Because most oceanic heat ente rs the Arctic from the Atlantic O cean, co rrect represe nta tion of Atlantic water circul a tio n is partic ul a rl y important. Obse rved radionuc lid e distributions from European fuel reprocessing provide a n o ppo rtunit y to eva luate differe nt circulation schemes for thi s water. \\re explore h ow th e represe ntation of sub-g rid edd ies a nd th e improved numer ica l simulation of ad \'ection lead to a tracer distribution that is more consistent with that which is obse rved (Livingston a nd other , 1982a, b). Soluble rad ionuclides, from th e Sellafield nuclear-power stati on in north ern England, a rc carried northwards from the Irish Sea, a round the wes t coast of Scotland, and into th e North Sea (LiYingsto n a nd others, 1982a, b ), mix ing with additi onal comp onents from La H ague (Francc1. Then they are transported nort hwards in the ~o rweg i an Coasta l Current (Holm and others, 1983) before sp litting oIT northern Norway. On c branch nows eas tward s imo the Barents Sea, whi le the other becomes part of the \Ves t Spitsbergen C urrent, em er ing Nansen Basin through Fram Strait (Smith a nd others, 1990). A p ortion then returns southwa rd in the East Greenland Current (Aarkrog a nd others, 1983; Livingston a nd others, 1985).

A multi -IC'\'CI primitive-equation model (Bryan, 1969; Cox, 1989; Paca nowsk i, 1995) was coupled to a dynamic- th ermod yna mi c sea-i ce a nd snow model (Hibler, 1979; Parkinso n and Was hingto n, 1979), using a sphe ric al coord inate system . The ocea n was speci fi ed with three \'e locity comp onents, potential temperatures and sa linity, while the sea ice was specifi ed with ice \'elocity, thickness and compactness. The sea-i ce model suppli ed the heat, sa lt and moment um excha nge [or the ocean surface, whereas th e ocean model suppli ed the cu rrent and heat-trans fer data for the ice model. The atmospheric forci ng (month Iy cli matological fields of wind stress, surface-air temperature, humidit y, radiation a nd heat Duxes ) came from datasets or bulk formulae (Trenberth and oth ers, 1989: Schubert and others, 1993; C hapman and ot hers, 199+), Th e study cowred the Arctic Ocean a nd northern parts of the Greenland and No rwegian Seas. Along the ope n boundaries with the Atlantic Ocean a nd Br ring Sea, wc used mom entum boundary conditi o ns deri\'Cd from a larger-sca le ocea n model. Le\'it us' (1982) temperature and salinity d ata were appli ed for in11 0w with free conditi o ns se t for the outflow of tracers. To a\'oid the singu larity eITect at th e Pole, wc employed Eby and H oll oway's (1994) grid rotation. The horizontal resolution of the model is 55 km, with 29 le\'e ls in the \,(,rtical , co\'Cring a to tal depth of +350 m, and expanding in thickness from 10 m a t th e surface to 290 m in the deep ocean. The configuration of the basin is defined by the depth field from ETOP05 (1986), with co rrection nea r the North Pole. 17

Nazarenko and others: Antic ocean- ice system contamination modelling The modcl includes pa rameter ization of the interaction of th e bottom topogr aphy a nd sub-g rid sea le eddi es th at ac t as a dri ving force (the "neptune" effee t) in ocea n circul ation (H oll oway, 1992; AI\,a rez a nd o th ers, 1994). Wc prese lll the res ults of two numerical experim ents: a cel1lered-differenee advectio n (CDA ) scheme with co nsta nt diffusion without neptun e, a nd a flux- corrected transport (FCT) scheme (a fter G erd es a nd others, 1991) with neptune pa ra meterizatio n. To limit the num erical di spersion by centered-difference ad\'ecti on, a n explicit diffusion larger than ca n be a llributed to physical mixing was added (Rood , 1987). Fo r the eentered-difference advecti on scheme, the hori zonta l - 1 respec. I d 1' ffuSlOns . a n d \'Crtlca were 10')- a n d ')_ x 10 ." m 2 s, ti\'C ly. From initi a l co nditi ons of zero \'eloc ity, no ice, a nd a nnu a l mea n Levitus (1982) temperature a nd salinity, integrati o n proceeded under m onthl y-va rying (arcing until quasi-equ il ibr ium was atta ined in th e upper a nd intermedia te depths. Th e accelerati on method (Bryan, 1984) was used with thi s time-dep endent forcing. Test cases using different accelerati on r ati os showed o nl y a weak dependence o n the accelerati on factor. After qu as i-equilibrium was attain ed, a consta lll trace r source was introduced inlO the No rwegian Coa ta l C urrent, as a n ideali zed E uropea n source of reprocessing radionuclides. The m odcl were th en integrated for a furth er 15 yea rs.

RESU LTS Arctic O cean

Surface-wa ter circulati o n in th e Arctic O cea n is la rgely understood from studi es of sea-ice d rift. Prominent long-

term features of this a rc th e a nticyclonic Beaufort G yre, which occ upies mos t of the Ca nad a Basin, a nd the Transpola r Drift th at fl ows ac ross the Pole toward s Fram Stra it. Th e coupled ice- ocean model q ua litati vely reproduces these features. Sub-surface circulation (42.5 111 ) shows the Beaufort G yre with only modest d ifferences between CDA a nd FCT (Fi g. I). At thi s depth, water motion in the experiments does not p roduce lh e Tra nsp ola r Drift. H owever, ice m otion (not show n) is nearly th e same for both a nd is co nsistent with the observed Transpola r Drift. Th e No rwegian Coastal C urrent (:\TCC), which foll ows the western a nd northern coast of Aforway before fl ow ing into th e Ba rents Sea, splits into two m a in branches: one turns eas tward s and co ntinues as a coas ta l cu rrent, the other tu r ns northwa rd s before subdividing into sm a ll er bra nches. Fi gure I shows d ilTerences in th e Ba rellls Sea circul ati on generated by the two ex periments. Two m ain currents exchange water through Fram Strait. T he ' '''est Spitsbergen C urrent (\'VSC ), an extension of th e NCC, fl ows through Fram Stra it, off th e west coast of Spitsbe rgen, carr ying wa rm, salty water into the Arctic O cea n. Its continua tion is co mplex, as the current appears to split into two or more bra nche ' (Perkin a nd Lewis, 1984·; Aagaard a nd others, 1987). For the C DA case with visco us d a m ping (Fig. la ), the WSC was not seen, but with FCT a nd neptun e (Fi g. lb ) th e nea r-surface vVSC does appear. Th e seco nd m ain current is the Eas t Greenl and C urrent (EGC) th at exits from the Arctic O cean. Fram Strait a pp ears to be a region of pronounced reeircul ation of Atl a ntic water, much of it j oining the south wa rd-m oving EG G. Th e differences between th e two cases a rc elearl y seen at il1lerm edi ate depth s (497.5 111 , Fi g. 2). With FCTa nd neptune

~

max 14cms - 1

· 15

o

15

max 14cms - 1

-10

-15

o

15

X

Fig. I. Velocities at 42.5 m.JoT (a) CDA witizout neptune, and ( b) FCT with neptli ne.

18

Fig. 2. Velocities at 497.5 m.for ( Cl ) CDA without ne/Jillne, and ( b) FCT with nejJtllne.

Nazarenko and others: Arctic ocean- ice system contamination modeLLing (Fig. 2 b), the cyelonic circu lati on of Atlantic-derived water is domin ated by se\·eral gyres steered by top ographic features, such as the G a kkel, Lomonosov a nd Alpha-l\1end eleev ridges. The FCT a nd neptun e case a lso shows the exiting EGC at intermedi ate depths, a feature a bsent in the CDA visco us-da mping case ( Fig. 2a ). >-

Tran s port of con tam i n a n ts

6.0'10 - 13

To im·estigate potenti a l rad ionucl ide p ath ways in th e A rctic O cea n, wc included an idealized so urce, co nsisting of an assig ned (co nsta nt) co nce ntrati on on th e open bou nda ry from th e NCC, to represent the European reprocess ing contaminants as rep on ed by K ershaw and Baxter (1995). Fig ure 3 shows th e ve rticall y integrated tracer burden aft er 15 yea rs. Th e CDAwithout neptune (Fig. 3a ) shows a diffu se co nta min ation field because of a la rge ex plicit d iffusion a nd unreali sti c circul ation. Th e FCT case with neptun e is m ore rea li sti c (Fig. 3b ). Some contamin ants enter th e Ba rents Sea via th e NCC, whil e the res t Oow northwa rd s with the '''Se. In Fram Strait, there is a p a rti a l rec ircul ation southwa rd s in the EG C (Figs 1,2 a nd 3b ) whil e th e north wa rd bra nch sink s a nd follows the sh elf break in th e Na nsen Basin as a boundar y current. Thi s current is relati vely strong, increasing with depth through the Atl antic laye r into deeper water, and moving co unter to th e surface fl ow (Figs 1,2 a nd 3 b). It takes contamina nts along the E urasian slop e a nd r eturns a long the Lom onosov R idge (Fig. 3b ). C onta min ati on spreading to the Canad a Bas in is carri ed by a co unter-cl ockwise flow along the contin enta l m arg in opposite to th e surface fl ow (Fig. 2).

9 .0 -10- 13

>-

Fig. 3. Vertically integrated tracer burden. Units are burden per unit slI1foce area, normalized by the source release Jar (a) CDA without neptune, and (b) FCT with neptlln e.

122. 5

i"

~

11 22. 5 .s: ~

g-

o 1997. 5

3 122. 5

(a) ~

_ _- L_ _~_ _~_ _~~_ _- L_ _~_ _~_ _~~_ _~_ _~

10

4487.5

20

122 .5

4 97. 5 ~

i"

~

1122.5 .s

~

15.

19 97.5

o'"

31 22.5

~

_ _- L_ _~_ _~_ _~_ _~~_ _~_ _~_ _~_ _~_ _~~

10

448 7. 5

20

X

Fig. 4. Tracer concentration and veloci£y normaL to the section r = 0 in Fig. I. Shading indica tesfiow "Ollt " of the /Jage. Contours are tmcer concentmtion relative to the sOll rceJar (a) CDA wilizollt neptune, and ( b) FCT wilh neptllne. 19

Nazarenko and others: Arctic ocean- ice s..)lstem contamination modelling

>-

o

-15

o

3.0'10- 13

6.0'10- 13

15

X 9.0'10- 13

>-

-15

o

15

Fig. 5. Verticalfy integrated tracer burden.for (a) FCT williollt neptune, and (b) CDA witiz neptllne.

In the Ba rents Sea, co nta mina ti on fl ows a long th e coast of Norway (Fig. 3). M ost co nta minants spread along the western coast of Novaya Zem lya (Fig. 2). Smaller bra nches of the NCC carry co nta mina tion to oth er pa rts of the Barents and Kara Seas. l\10d ifi ed At la ntic wa ter in the Ba rents Sea ex its along the eastern fl a nk of the St Anna Trough, carr ying conta min a nts into th e Arctic Basin. Pa rt of the \VSC a lso turns into th e Barents Sea a long the western flank of th is trough. Figure 4 shows conta min ation co nce ntration a nd velocit y norm a l to a vert ica l sec ti ons along Y = 0.0 (through th e North Pole). Shaded a reas denote norma l velocity "ou t" of th e page (negative), whil e blank areas denote velocity "into" the page (positive). Near X = 7.4 co nta mina nts are brought into the Arctic O cean from Fram Strait by th e intermediate-dept h current. l\1uch less contam in ati on leaves the Arctic with the outflow at X = 5.0 (Fi g. 4·b) where veloc iti es become negative. The model with FCT and neptune produces the returning fl ow a long the Lom onosov Ridge at X = 2.0 (Fi g. 4 b).

20

The distribution of co ntamination from the CDA without neptun e model is shOlVn in Fig ure 4a. The distributi on is dominated by the expli cit diffusion, necessary to maintain num erical stabi li ty, th at makes the co ntam ina nts spread in a n unrealistic flow fi eld (K ershaw and Baxter, 1995). Because improve ments h ave resulted from the use of FCTwith neptune, the question a ri ses as to how much each modification would contribute separatel y. Figure 5 has the same format as Figure 3, except that Figure 5a is based on FCT without neptune a nd Fig ure 5b uses CDA with neptune. Most of the overall improvem en t (Fig. 3b) res ults from ineluding th e neptun e effect (on acco unt of modified circulation patterns). Th e FCT is, noneth eless, helpful in retaining sh arper property contrasts in the co ntamin a nt field a nd in the tem perat ure a nd salinity field s. Finally, a word of"eauti on. Our comparisons were based upo n the GFDL ocea n model (PacanolVski , 1995) which is gridded on fixed z. levels. Mode l results are also dependen t on th e und erlying architecture, which is sometimes poorly understood. \lVe do not know how the apparent progress

.\(l.::.arenko and a/h ers: Arclic ocean- ice system conlamill a/ion mode/ling m ade here m ay ca rr y O\'e r to oth er m odel s such as th e sig ma-coo rdin ate (H akkinen a nd IVIe 11 or, 1992) or isopyc na l-layer (H oll a nd, 199'1) model s.

SUMMARY

Cox. 'd . D. 1989.. \n i d~a li z~d model of th e \\'orid ocea n. Pa rt I. The g loba lsca le water masses . .7 I'I!J". Orl'Gllogr., 19, 1730- 1752. Eby. M . a nd G. Ho ll owa),. 199 1. G rid tra nsl'o rm atio n lor inco rporating th c Arct ic in a gloha l ocea n mode l. Clilllale DJ'Il., 10+- 5,2-" 2+7. ETOPO :i. 1986. Clobal j ' x j 'de/III! alld elevalion. Bo ul der, C O . li.s. Department of Commr rce. Code E GC3. Gcrdes. R ., C. Koberi c a nclJ \\·ill ebra ncl. 1991. The inOuence of numerica l ach TCl io ll sc he m es o n the res ults

A multi-level primitive-equ ati on model , coupled to a dynamic- therm od ynamic sea-ice model, was used to ex plore the di stribution of radi onuclides from idea li zed Europ ea n reprocessing so urces. The resea rch foc used on determining the possible effec ts of th e pa rameteri zati on for unresolved eddi es, a nd a n ad vection sc heme on conta mi na ti o n distribuli on in the Arctic O cea n. Th e model with sm a ll mixing a nd improved physics (FCT with neptune) simul ated a more rea li stic circul a ti on a nd co nta mina tion distribution tha n the ce ntered-difference ad\'ecti on sch eme with co nsta nt dif~ fu sio n (CDA withoul neptune).

ACKNOWLEDGEMENTS This research has been supported in p a rt by th e Pa nel on E nergy R esea rch a nd D evelopment (C a nad a ) under proj ec t 6B2027 a nd by the Office of Naval Resea rch und er g ra nt NOOOl '~- 96 -1 - 0 5 18.

REFERENCES Aagaa rd, K .. A.

1~1c" ' ik

a nd S. R . I-lillm a n. 1987. Th r West Spit sbe rgr n

C urre nL: cli spo:-; ili on a n d Wa L(T m ass lra nsform al io ll .

.7

(;eo/)/~l'J.

Hrs.,

92 14,3778 37!H. . \ a rkrog. A., 1-1 . Da hl gaa rd, H. H a nse n a nd E. I-Iolm . 1983. Rad iocaesilllll from Scllaficld d TI lIe nts in Grren la nd waters. . \ alllre. 304 (5921. +9- 51. Ah'a rez .. \ ., j. Ti nto r!·. G. Hollowa)', 1\1. Eb)' a nd]. 1\1. Bec kers. 199+. EfTrrt of topographic stress o n the circul ati on in the wes tern 1\lcd iterra nea n. ]. Geo/JI!)·s. Res., 99 (Rl.16.053 16,06+. Br)'a n. K . 1969.. \ num erica ll11c thod l'or th e stud), oflhe ci rc ul at io n of the wori d ocea n . .7 Com/ml. PI!)·s.. 14.666 671 Brya n. K . 19B+. .\ c("d era ting lh e cO Il \'(' rgc llce to eq uili b riu I11 of ocea n cl i m" te mockl s. .7 I'I,I'J. Oreall~~I:, 17 (71,970 985.

C ha pm a n, \\'. L.. \\'. .J. Welch , K . P Bow ma n, j. Sacks a nd J E. \\·a lsh. 199 l. Arct ic sea ice va ri ab ility: model se nsiti"iti es a nd a l11ultidecada l simul ati o n. ]. GfO/J/!l's. Re, .. 99 ( C l), 91'1 '1:~5. c

or ocea n ge ne ra l c irc ulati o n tllnc1 c1 s.

CLimale D.1'1I., 5, 211 226. H'ikkincn. S. a nd G. L. 1\1ellor. 1992. tll odeling the seaso na l va ri a bili ty ora co upled . \ rctic ice- occa n system.]' Ceoplrp. Re.l.. 97 1C I2 . 20.285 20.30+. I-li bler, \\'. D. . Ill. 1979. A d ynam ic ther mod yna mi c sea ice mode l. .J P/~)',. Oreallagl: . 9 (7 .81:)- 8+6. Ho ll and. D. i\1. 199-1-. ~um cri ca l simul ati on of .\r('lic sea-ice and ocean circu \'lli o l1 . ()(('allogm/)/~) '. 7, 272ft Ho ll oway, C. 1992. Re prese nting topop;raphic stress 10 1' large-scale ocea n

model s. .J PI!)'.I. O{fallo,~': ' 22. 1033 10+6. I-I o lm. E .. B. R. R . Persson. L. H a ll sl:ld ius. A. ,\ a rk rog a nd 1-1. Da hl gaa rd. 1983. Radiocacs ium a nd tra nsura nium elements in the Greenla nd a nd Bare nts Seas. Orenllolagira . Irln. 6. ~ :i7 +62. h ershaIV. PJ. " lid A.J. Baxte r. 1995. T he lransfer of reproressing ",,,s tes fm lll north-west Euro pe to th e A rclic. Dee/I -Sea R"s.. SI'/'. 11, 42 (6), 1+13 1++8. Le,·itus. S. 1982. Ciilllalological alias qf Ihe world omlll. Rockl'ill e. ,\ ID, U.S. Depa rt ment of CommeJ'ce. :\m iona l O cea nic a nd Atmos ph eric :\ d ministrat io n. 1:'\0. \i\ Profess iona l Paper 13. Li"i ngston, H . D.. \ '.T. BOlI'c n a nd S. L. Kupcrm an. 1982a. R ad ion uc1 ides from \\'indscalc d ischa rges. I: non-equili briu m tracer experiments in hig h latitudl' ocea nog raphy..J .I/arille Res., 40. 2.'i3 272. Lil' ingston, H . D.. V. T Bo\\'c n a nd S. L. "-upcr m" n. 1982b. Rad ion uclides from \\'indscalc disc harges, 11 : thei r d ispe rsio n in Scott ish a nd :'\OI'wegia n coasta l circu latio n . .7 .Ilari,"' Re.>., 1·0.1227 12.18. Li,·i ngston. 1-1 . D.. j. H . S,,·ift a nd 1-1 . C. O stlllnd. 1985. ,\rtificia l radionud idt's trace r supply to the Denm ark Stra it OIwll o\\' between 1972 and 1987. ]. Ceopl!),!. Res., 90 , C+), 697 1- 6982. Pacanowski , R. C. 1 99~ . .1/om2 ",as .~lIide alld referellre lIIalllla/. leHioll /. Princc to n, NJ. :"JOAA. G FDI . O ccan G roup. r'lc'chnica l Report 3.) Pa rkinson. C. L . a nd \\'. 'd . Was hinglO n. 1979.. \ la rge-sca le numerica l model of sea ice . .J Cropl!)';. Res., 84 C l . 31 1- 337. Perkin, R . B. and E. L. Lewis. 198+. :\1 ix ing in the \ \'est Spi tsbe rge n current. ]. PI!)'s. Oceallogr.. 14, 1315- 1325. Rood. R. B. 1987. :"Jumeri ca l ack ecti o n algorithm a nd th cir rolc in a tmospheric tra nsport a nd chcmi str y models. Rer'. (;eophl's., 25,71- 100. Sc hubert. S. D.. R. B. Rood a ndJ P r.~c ndtn e r. 1993. An assimila ted datase t fo r eart h sc ience applical ions. BIIII. • 1111 . . IIelenro/. Sal'.. 74 12.2331 23+2. Sm ith . .J. K . K. AI. Ellis a nd E. P. .l ones. 1990. Carsi um 137 tra nsport in to lhe .\rel ic Ocea n th ro ug h Fram Stra it..J Cea/)I!)',. Res.. 95 121.1693 1701. Ti'c nbc rth , K. E., J G. 01 50n a nd \ \'. G. L a rge. 1989.. 1global oreall u'illd slre.I.' climalalo!!,)' bIlled 011 EC \l11 F alla!)'se.1. Bo ulder, c O , Nati ona l Celller for Atm os ph eric Rcsea rch. I :'\C A R ' lcc hnica l Notc TN-338+ STR .)

21