Thermohaline Steps Induced By Melting of the Erebus ... - CiteSeerX

0 downloads 0 Views 924KB Size Report
Jul 20, 1981 - account for much of the Glacier Tongue thinning by basal melting. ... McMurdo Sound (Figure 1), including three near the end of 1974] at all ... near the end of Erebus Glacier Tongue, which at that time projected ..... cent, less-energetic ones. .... year-round temperature measurements to 550 m depths [Tres-.



RESEARCH, VOL. 86, NO. C7, PAGES 6547-6555, JULY 20, 1981

ThermohalineStepsInduced By Melting of the ErebusGlacier Tongue S.S.


Lamont-DohertyGeologicalObservatory,ColumbiaUniversity,Palisades,New York 10964 H.


Departmentof AppliedMathematicsand TheoreticalPhysics Universityof Cambridge,England G.


Environment Canada, Ottawa, Ontario



ScottPolar ResearchInstitute, Universityof Cambridge,England A vertically stable,step-likethermohalinestructureis observedthroughouta continuous,400 m conductivity-temperature-depth (CTD) profile taken near the Erebus Glarer Tongue, McMurdo Sound, Antarctica. The pattern is best developedbetween the sea surfaceand 250 m depth, the interval correspondingto that of the irregularunderwaterprofile of the Glacier Tongue.The stepsaverage17 m in

thickness andtypicallydisplaydiscontinuities of 0.1øCin temperature, 0.04%oin salinityand3.5 x 10-4 g cm-3 in density.The observations arecompared with theoryandlaboratoryexperiments of celldevelopment and lateral flow near ice melting into vertically stratifiedsalt water. At this location, subsurface seawateris inferred to remain above the in situ freezingpoint year-round,and containssufficientheat to accountfor much of the Glacier Tongue thinning by basal melting. An adequatevolume of meltwater would resultto producethe measuredsalinity steps.We discussrelated observationsand someimplications of this processfor oceancirculationand biologicalproductivityin the Antarctic.


times the impact on density (o0 of a 0.1øC temperature

In mid-February 1979 theU.S.Coast Guard icebreaker change. Temperature isnearly isothermal below 250m,and Glacier occupied several stations in thesouthern partof more than 0.2øC above theinsitu freezing point [Fujino etal., McMurdo Sound (Figure 1),including three near theendof 1974] atalldepths. Several ofthestation 217steps canbe

identified on nearby station 218, but the relative importance of spatialversustemporal changescould not be determined, as these stationswere taken on an opportunity basis. The 2b). Stations217-219 werelocated250-500 m apart with 218 CTD recordsare relatively noisy, possiblydue to systemreat the glacierterminus.At each positionthe ship'sscrew-prosponse time and slowsensordescentrate. The differingdescent pellerswere slowlyrotatedwhile its bow was held againstthe and ascent profiles,and turbulencenear the large local gradipack ice. ents also suggestactively evolving,not remnant features. Temperature,conductivity,and pressurewere measuredbethe Erebus Glacier Tongue (EGT, Figure 2a). At that time, only the tip of the EGT protrudedfrom the sea ice (Figure

tween the sea surface and bottom

with a Neil


On and near the Antarctic


ductivity/temperature/depth (CTD) instrument. The CTD lowering rate of approximately40 m min-l correspondedto

onetemperature and salinityvaluefor each2 cm of depth:


shelf we have fre-

quently recordedthermohalinestaircasesof varying dimensionsat intermediate depths. Some were of the variety generafly ascribed to double-diffusiveprocesses,e.g., cold fresh over warm salty water beneath the Antarctic Surface Water temperatureminimum, or warm salty over cold fresh water below the Circumpolar Deep Water that intrudes onto the continental shelf. At other times, steps were barely discernible,or not much greaterthan the resolutionof the instrumentation then being used. Station 217 is unlike any of the above, but graphically displaysa staticallystable configuration of relatively warm fresh over cold salty water. This is analogousto the homogeneoussummer surface layer that overliesthe temperature minimum (stepsalso occur at that

The recordswere transferredfrom audio to digital tape and salinity computedfrom conductivity.Temperatureand salinity bias correctionswere applied from reversingthermometer and water sampledata taken with a General OceanicsRosette during ascentof the CTD. Severalof the stationsrevealedstepsin temperature,salinity, and density that are more than an order of magnitude larger than resolutionof the CTD instrument. The best-defined and most regular stepsoccurredon station217 (Figure 3), and average17 m in thickness.Within a temperaturerange of-1.2 to -1.9øC and a salinity range.of 34.2 to 34.8%o, the transition),but here the stairsare more uniform and descend to greater depths. major stepsdisplaydiscontinuitiesof about 0.1o C in temperThe step structureobservedhere differsin an important reature,0.04%0in salinity,and 3.5 x 10-4 g cm -3 in density.At spect from that reported,e.g., by Neal et al. [1969], under a these low temperatures,a salinity change of 0.04%ohas 10 drifting ice island in the Arctic. In that casethe stepswere Copyright¸ 1981 by the American GeophysicalUnion. formed by the one-dimensional process of cooling from Paper number 1C0546. 0148-0227/81/001C-0546501.00













Fig.1. Oceanographic stations (CTD211-219) takenduring February 1979fromtheU.S.icebreaker Glacier in McMurdo Sound.Stations217-219 werenear the end of ErebusGlacier Tongue,which at that time projectednearly 13 km fromRossIslandintothe Sound.Topographic featuresarefromtheUSGS mapRossIsland,Antarctica,ST 57-60/6.

above,whichproducesa staticallyunstabletemperaturepro- One set of investigations[Neshyba,1977;Josberger,1979] asfile. In our situationthe coolingis from the sideand produces sumeslittle or no salinity gradient in the surroundingwater

stepseventhoughbothtemperature andsalinityprofilesare and focusesupon the boundary layer near the ice. The other work [Huppertand Turner, 1978, 1980;Huppertand Josberger,


1980]concentrates on the influencesof verticalsalinity(den-

sity) stratification. Neshyba[1977] conjecturedthat adjacentto a melting ice Radio echorecordswere obtainedin January 1978along an wall there will be a turbulent boundary layer which entrains approximate centerlineof theEGT [Drewryetal., 1980].The antennasystem of fourcolineardipolesmountedat 1/4 wave- water from the environment.Consideringa two-layer model lengthbeneath thestarboard wingof a C-130(Hercules) air- of the oceanand using the criterion that entrainmentis just craftgivesa radiationpatternwithbeamangleof 80-90øfore- sufficientto ensurethat the density in the boundary layer at

and-aftandup to 60ø in the transverse direction.At a flight heightof 300m; returnscanbe observed fromthe uppersurface,the bottomof surfacecrevasses, the ice/waterinterface containingcrevasse-like structures at bottom,and the ice tongue"teeth"on eitheredgeandassociated crevasses which penetrate towardthe centerof the icetongue.Extensive surface crevassingnear the landward end gave rise to clutter

thesurfaceequalsthedensityof theupperlayer,Neshybacal-

echoeswhich obscured returns from the base of the tongue

the ice.

there.Interpretationof the basalprofileis thussomewhatuncertain,and our ice thicknessmay be in error by +15 m. The resolution is better than this but is a function of bottom

roughness and lateral continuity. The ice thicknessdata indicate a two-stepthinning beginning at the groundingline. The main thinningoccursat

culates an entrainment

ratio for the Weddell

Sea of 60 to 1

(that is, for eachunit of meltwaterin the boundarylayer there is entrained 60 units of surroundingwater). Subsequently, Josbergerand Neshyba[1980] placed concentratedrhodamine dye at 14-18 m depthsalong the slopingwall of an Arctic icebergand foundthat it roseat =7 cm s-• to the surfacenext to Greisman[1979] consideredthe thermodynamics involved in the meltingprocessand calculatedthe amountof heat requiredfrom the turbulent,entrainingboundarylayerto melt the ice. This lead to a minimum entrainmentratio e givenby



depthsbelow about 75 m; the outer third of the Glacier Tonguethinsby 25 m at most.The radarbasalprofileshowsa roughness underneath theEGT thatis aboutthesamemagnitude as the thermohalinestepson station217. Due to prob-

where L is the latent heat of fusion of ice, p• the density of

grounding lineis at =275m), salinityanddensitystepsin the


freshwater,Cpthe specificheatof seawater, AT the temperature excessof the far-field water over that at the ice wall, and

poothe densityof the far field, which is assumed to be independent of depth. By considering the mixture of freshwater lems of resolutionand interpretationof the echo sounding data notedabove,we can only speculateat presentthat anom- and far-field water in the boundarylayer, it followedthat the aliesfrom an equilibriumbasalprofileare real andrelatedto densitydifferenceAt>betweenthe far field and the boundary the thermohalinesteps.Belowthe influenceof the EGT (the layer is given by seawaterare considerably lessdistinctand temperatureis essentiallyisothermal(Figure 3). THEORY

Recentlythere has been a seriesof investigations of the fluid dynamicsof ice walls meltinginto surroundingwater.

If the far field is vertically stratified,Greisman statedthat the water in the boundarylayer flowsupwardonly by an amount





---.e;'• '7,;,"

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

- ?•

.,. ..


Fig. 2a. Aerial obliquephotographof ErebusGlacier Tongueat a time (= 2100Z, February 18, 1965)whenlittle solid pack ice remainedin ErebusBay. The Ice Tonguehad increasedabout2 km in lengthby the time of the CTD stations reportedhere.The seasurfacepatternsand drifting ice suggestflow towardthe Glacier Tonguefrom the south.

where dp/dz is the vertical densitygradient before it losesits cinity of the ice is altogether different [Huppert and Turner, relative buoyancyand turnsinward to form a seriesof layers. 1980]. Next to the ice there is a thin upward-flowing layer of Greisman doesnot suggesta mechanismby which the layers meltwater, beyond which there is a thicker, turbulent, downcould be formed. ward-flowing layer. This outer turbulent layer entrains meltJosbergerconducteda seriesof carefully consideredand water from the inner boundary layer. Becauseof the increase analyzedexperimentsusingverticalice walls 1.2m high melt- in salinity, and thereby density, with depth, the water in this ing into water at temperaturesfrom -1 øC to +20øC at a uni- cooled outer layer reachesa level of neutral buoyancy and form salinity of around 33%0.Above a very short laminar flows away from the ice in a seriesof nearly horizontal layers. boundary layer with an upward-flowing inner layer and a The laboratory experiments of Huppert and Turner [1978, downward-flowingouter layer, Josbergerobserveda turbulent 1980]and Huppertand Josberger[1980]and their related analboundary layer of solely upward motion. The width of this ysessuggestthat the layer scaleis well representedby

layerincreased aszl/n,wherez is the verticallengthmeasured from the bottom of the turbulent layer. This is in agreement with a similaritysolutionobtainedby Josbergerfrom the ap-


h--0.65 [p(T l,,½oo) - p(Too, ,½oo)] dz


proximateNavier-Stokesequationsof motion. In the presenceof a verticalsalinitygradient,flow in the vi- whereto(T,S) is the densityas a functionof temperatureand




Fig. 2b. A northward viewonFebruary15,1979fromtheicebreaker Glacierat a positionslightlysouthof station217. At thistime onlythewesternterminusof the ErebusGlacierTongueprotrudedat rightcenterfrom the seaice.Tent Island is in the left centerbackground, and the lowerslopeof Mt. Erebusappearsin the right distance.




34.2 34.3 34.4 34.5 34.6 34.7 34.8 34.9 I I I I I I I 27.4








,It 12.5 10.0 7.5






I00 ß

200'Tf 250




40 0





z::::::::::::::... ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: :::..:::::2:::::.. •:•:•:•:•:•:L ;.•:3: ...... :•:.. •:•


Fig. 3. (Left) TemperatureT, salinityS, and density{It profilesfrom Glacierstation217, locateda few hundredmeters

southof thewesternterminusof ErebusGlacierTongue(EGT). The in situfreezingpoint(Tf) is afterFujinoet al. [1974]. (Upper fight) A verticalsectionalong the EGT centerline. The upper surfaceof the floatingglacieris definedat I km intervals(_100 m) to _+2m in the vertical.The equilibriumbasalprofile(dash-dotline) is calculatedfor free flotationof a

slabof meandensity0.87Mg m-3. TheJanuary1978radarbasalprofile(dottedline)hasa verticalresolution of _+15m as shownby the errorbars.North and southedgesof the EGT are approximately40%thinnerthan the centerline profile. (Lower right) Dashedline connectsline soundingsof the seafloor north and southof the EGT. Circled dotsare bottom depthson CTD stations218,and219.The EGT movesoff RossIslandinto waterthat is 225-275m deep.The irregularsea floorbeneaththe EGT continuesfor at leastanother5 km westof its terminus(DMA chart29322,Cape Roydsto Hut Point).



TABLE 1. Computedand ObservedVertical Step Dimensions


$, %

0 50 100 150 200 250 300 350 400

34.117 34.412 34.523 34.572 34.637 34.710 34.734 34.768 34.795

T, øC -

1.53 1.23 1.35 1.52 1.73 1.90 1.90 1.90 1.89

Tf, øC - 1.87 - 1.92 - 1.97 -2.01 -2.05 -2.09 -2.13 -2.17 -2.21

AT,øC 0.34 0.69 0.62 0.49 0.32 0.19 0.23 0.27 0.32



ApX 107


1.0273505 1.0278203 1.0281551 1.0284421 1.0287431 1.0290492 1.0293103 1.0295791 1.0298415

1.0273584 1.0278396 1.0281732 1.0284568 1.0287530 1.0290552 1.0293180 1.0295886 1.0298534

79 193 181 147 98 60 77 95 119

6 14 13 11 7 4 6 7 9

Obs. 5 17 27 29 31 34 28 15 35

Salinityandtemperature fromstation217,in situfreezing temperature Tf fromFujinoet al. [1974]. Density(p andpf) after GebhartandMolendorf[1977]. Layerthickness h is from equation(4) and from the layerat each50 m intervalon station 217.

salinity,Tf is the freezingpoint of waterat the meanfar-field within larger ones,particularlybelow 250 m, but in general

salinity•oo,anddps/dzis theverticaldensitygradientdueto below the Glacier snout the primary stepsare thicker than salt. From the Huppert and Turner experimentsit appears predictedby a factor of 2 or more. Assumingthat our formuthat the factor 0.65 holdsover a largerangeof scales,or Gra- lation is essentiallycorrectand that we shouldfocusupon the shof numbers, and this is consistent with the numerical inmajor steps,what can account for the discrepancybetween tegrationsof Chenet al. [1971]and Chen[1974].The experi- predictedand observedlayer thickness? mentsindicatedthat the layer thicknesses were uniform with The equilibriumbottom profile of the EGT more nearly redepthand the mean far-field salinitywasneededin (4), as in- semblesa sloping ceiling than a vertical wall. The vertical dicated,althoughthe theoryas so far developedsuggests that scaleof Figure 3 exaggeratesthe thicknessgradient,which is the layer thicknessh might respondto the local rather than at most about 1:40. We (H.E.H.) found that the flow due to the mean far-field salinity. an ice block sloping45ø in the laboratorytank was not essentially different from that producedby a vertical wall. This is STATION 217 becausethe scalesare determinedby the vertical displacement Appearanceof the best-developedstep structureon a sta- inducedby coolingfrom the side and not the angle at which tion closeto the EGT suggested that its formationmight result the coolingtakesplace.However, there may be a critical slope from meltingof ice into the salinitygradient.The validity of beyond which relatively lessboundary layer meltwater rises thishypothesis can be examinedquantitativelyby comparing and more is entrained into the layered structure. Another differencebetween the field and laboratory meathe observedlayer thicknesses with thosepredictedby (4). The vertical densitygradient due to salt can be evaluated surementsis that, with increasingdepth in the water column, by fitting a line of constantslopethroughthe salinityprofile the CTD recordsare further from the presumedsourceof the of station 217. Orienting the line to connecta salinity of steps.Sincethe steps(layers)are staticallystable,vertical mix34.523%0at 100 m and one of 34.637960at 200 m, we determine ing within a layer will be greaterthan mixing betweenlayers. thatd,•oo/dz -- 1.1x 10-3%o m-• andhencethatpo-• dps/dz-- Turbulent exchangeof water acrossthe discontinuitiesmight 8.8 x 10-? m-l, wherepois the meandensity(•103 kg m-3). resultin more energeticlayersgrowingat the expenseof adjaThe resultsof applying (4) at nine representativestation 217 cent, less-energetic ones.At each 50 m depth interval down to depths,salinitiesand temperaturesare shownin Table 1. The the groundingline, observedlayer thicknesson station 217 is freezingpoint temperatureas a functionof salinityand pressure was calculated from the relationship in Fujino et al. DISTANCE FROM ICE (km) [1974], after correctingthe sign of their pressureterm. The 2 4 6 8 I0 12 35 density as a function of temperatureand salinity was evaluated by the formulation of Gebhartand Molendorf [1977], 3O and the parametersof station217 were taken to representthe far-field


Applicationof (4) to station217 resultsin predictedlayer thicknessesof 4-14 m, with layers in the 50-150 m interval about twice as thick as those at the surface and below 150 m.

Aside from the surface and bottom boundaries,the layer thicknessin this caseis, more simply,proportionalto the driv-

ing temperature, AT-- T-

Ti (Figure4; seealso Greisman

[1979]and Neshybaand Josberger[1980]). The major stepsobservedon station217 rangefrom 5 to 35 m thick, averaging 17 m. The thicknessof the layer nearest each 50 m depth incrementis shownin Table 1. Agreement betweentheoryand observationis bestin the upperwater column as might be expected.The tank experimentswere done with verticalice walls, and the EGT terminusis more nearly vertical in the top 50-75 m (Figure 3). There are small steps

• z


ho-'-18.5+ I.•SD




hT--I.6 +18.5 (T-Tf) ,,.•.-•

I.d I0 5


-;•S• •..•.-.•+ B•_•.• S I 0.1

I 0.2

I 0.3

I 0.4

I 0.5

I 0.6

T-Tf (øC) FiB. 4. (Lowe0 Computedlayer thickness(equation(4)) related to differencebetweenin situ temperatureT and freezingtemperature Th (Upper)Observed layerthickness (station217)relatedto horizontal distancefrom ice (Figure 3). S and B refer to seasurfaceand bottom points.




directly proportional to horizontal distance from the EGT

column and would thus need to add a minimum 6 cm3 of fresh

(Figure 4). The line interceptis only 4 m greaterthan predictedby (4) and suggests a verticalstepgrowthof 1.3m km-•

waterper cm2 of basalsurfaceto producestepsof the sizeob-

from the ice.

This salinity changedefinesa rough area over which such melt-induced steps could be distributed each year. For example,if the basalmelt rate of the EGT were between0.3 and

Sincethe stepson station217 were not coherentover a wide area, we could infer that they coalescefairly •apidly into thicker layers. We have observed50 m and thicker homogeneouslayersabove and within the high-salinityshelf water in the RossSea. The thickestlayerson station217 were near the sea floor, where interactionscould be expectedbetween currems and bottom irregularities. There are shoals at several depthsnear the EGT (Figure 3), and internalwavesgenerated by tidal flow around thesefeatures[e.g., Farmer and Smith, 1980] could provide an energy sourcefor mixing within the layers. In the laboratory experimentsof Huppertand Turner [1978], ridgingof the ice block was causedby differentialmelting associatedwith the layered convection.If the lower surface of the Glacier Tongue alsohas a steppedstructure,it could result from the convectivecells or from internal wavesthat propagate at the oceandensitysteps.Waves are particularly effective in eroding ice at the sea surface [Martin et al., 1978]. Greater turbulencewhere the internal wavesimpinge on the ice might increasethe melting rate. There are severalreportsof layered structurenear melting glacial ice in the Arctic. Greisman[1979] refers to 10 m convective cells in d'Iberville Fiord, Northwest Territories, where

the glacier wall-seawater temperature difference (0.3øC) is comparableto our EGT case.However,we cannotreadfly apply his relationship,which containsthe unknownentrainment ratio e, to theseAntarcticdata. Josberger and Neshyba[1980]

served on station 217.

3.0 m yr-•, then stepsof the dimensionsobservedon station 217 couldbe generatedover an areaof 75-750 km•, or 5-50 timesthe basalarea of the EGT. The correspondingresidence time for the station 217 water column under the EGT


then be 0.2 to 0.02 yr. An area of 350 km• and residencetime of 0.04 yr would apply to the 1.4 m yr-• melt rate estimated above. Is sufficient heat available

below the sea surface to melt 1.4

m yr-• off the baseof the ErebusGlacierTongue?The water column between 75 and 250 m ranges from about 0.7ø to 0.2øC abovethe in situ freezingtemperature(Figure 3). From year-roundtemperaturemeasurementsto 550 m depths[Tresslet and Ommundsen,1962]we would expectthat mostof the subsurface water column

in McMurdo

Sound also remains

abovethe in situ freezingpoint during winter. Supposethe annual averagetemperature above freezing were only 0.1øC, aboutthe mean differencebetweenTf and a line connecting

Tf at the surfacewith the coldestportionof T at about250 m (Figure 3). Then a 200 m thick water columnexchanged25 timesyr-• (reciprocalof the 0.04 yr residence time) would require only 30% of that heat to melt the 1.4 m of ice. Thus there is sufficient heat in the water column to melt the ice and

adequatemeltwaterto producethe observedsalinitystepsin a reasonable residence time. The effective transfer of heat from

have'reported theformationof 5 m isothermal layersnearan

water to ice dependsupon mixing coefficientsand boundary layer dynamics,a discussion of which is beyondthe scopeof

iceberggroundedin 90 m of water, but in a densitygradient

these data.

several times that of our station 217.

Tidal currentscould move a seriesof stepsaroundthe EGT region,but little is known about the flow at this location.The sailing directionsfor Antarctica [U.S. HydrographicOffice, 1960]indicatea northwestwardsurfacecurrent as strongas 3 knots nearby, but also note that this Cape Armitage current





The ice depth data and stress/strainrelationshipsindicate that the EGT is floating throughout its length [Holdsworth, 1974]. Due to its continuouswestward movement, the EGT increasedin length by about 4750 m between 1947 and 1979 (Holdsworth[1974], updated by the satellite-navigatedship positionof the icebreakerGlacierduringoccupationof station 218, 77ø42.3'S,166ø23.9'E).Assumingonly minor calving at the terminusduring that time resultsin a rate of advance of

about 150m yr-•. Allowingfor integratedhorizontalstretch-

has been observed to set southward

under certain conditions.

One of us (G.H.) drilled soundingholesthrough the sea ice north and south of the EGT. Water (and fish) upwelled through severalof the holes on the south side, suggestinga flow from that direction in the near-surfacelayer. One could alsoinfer a current from the southby the seasurfaceand drift

icepalternsin Fig. 2b. Duringstation217 the windwasfrom 96 ø at 7-15 knots which would

have induced a surface flow

ing of the ice beginningat the groundingline, then ice at the toward the south and away from the pack ice edge. Subsurface current measurements made at various times presentlocation of the terminushas been suspendedin seawater that is above the in situ freezing point for about 85 and locationsin McMurdo Sound, usually for brief periods, years.The data showthe EGT to be thinning by =280 m be- were summarizedby Heath [1977]. Within a regime of high tween the groundingline and westernterminus.Allowing for variability and strongdiurnal tides, he discerneda pattern of creepthinning [Weertman,1957],approximately120 m would mean drift toward the south in the middle of the Sound and to be due to basal melting. This would correspondto a melt rate the north on the easternand western sides,with the greatest of about 1.4 m yr-•. velocitiesat mid-depths.The current velocitiesfor depthsbeIs melting at the baseof the EGT sufficientto generatethe low 50 m in Heath'sTable 4 averageabout10cm s-•. Moving thermohalinestaircaseshownin Figure 3? We assumethat a at thisspeed,a 15 kra• watermasswouldremainbeneaththe salt deficit can be defined by the area between the salinity/ EGT for only 3 hoursrather than the 2 weeksimplied by our depth profile and a line tangent to the right-hand side of the 0.04 yr residencetime. Once the tidal signal is filtered out salinity/depth profile in the 100-200 m interval. Taking however, residual drift will be considerablylessthan 10 cm 0.02%oas a typical salinity discontinuity,we find that each 20 s-•. In any case,only first-orderestimatescan be made from

m stepin a cm2 watercolumnrepresents a 0.02g saltdeficit, these limited observations. whichin turn corresponds to the additionof about0.6 cm3 of There is evidencefor some freezing at shallowerlevels on freshwater. The EGT is now melting into a 200 m thick water the ErebusGlacier Tongue. The radar basalprofile (Figure 3)










217 214






Gloc/•r Stotions 213, 214,217 Glacier Slotions 211-12,215-16, 218-19 500



• i


I •



- 2.0






Fig. 5. Compositesalinity/depthand temperature/depthprofilesfor Glacierstations211-219 in southernMcMurdo Sound,Antarctica. The stationlocationsare shownin Figure 1.

showsa roughly constantthicknessover the outer 4 km of the

ward the intermediate

station 217 characteristics.

The remain-

EGT, while surface elevations decrease toward the end. This

ing stationsclusteraround217 in the 100-400 m depth range, impliesan increasingaverageice columndensitytoward the and their spacingprovidesa rough estimateof the area influend, which could result from the presenceof higher-density encedby meltwater from the Erebus Glacier Tongue. An elsea ice. Summer is brief in the Antarctic, and above 75 m the lipse that includesall stationsbut 213 and 214 coversnearly EGT is well within the range of seasurfaceheat flux. Further, 1000km2, more than twice the area projectedabove.Howthe EGT internal temperaturegradient is likely to be steeper ever, we cannotreadily estimatethe contributionsthat might near the thinner end, which could facilitate winter freezingin be made by other glacial ice in the vicinity nor the residence that sector. From McMurdo

Sound winter observations,

Loewe [1961] has reported that ice blocks lowered to 47 m grew in size,and Dayton et al. [1969]found that ice platelets accumulateon bottom and on suspendedlines to depthsof 33 m. However, it doesnot seempossiblethat the thermohaline structure we measured resulted from freezing. Continued melting and thinning would be expected from the summer water column temperatureprofile. Any layered structurethat resulted from freezing of seawater deep beneath the ice shelveswould probably be slight due to the low freezing rate there and not persistfor long in the relatively active current regime [Jacobset al., 1979a]. REGIONAL


time of meltwater

in McMurdo


Salinities of the stationsgrouped around station 217 convergein the vicinity of 120m (Fig. 5). That depth corresponds to the temperaturemaximum on station 213 and probably marks the lower limit of recent sea surfaceprocessesin this area. The warmer water on stationslike 213 is ultimately derived from distantCircumpolarDeep Water, the primary heat source for glacial melting [Jacobs et al., 1979a]. Melting shouldbe greatestat intermediatedepthsand densityhorizons in common with the Circumpolar Deep Water where it intrudes onto the continental shelf. For example, the warmest water on station 222 at the Ross Sea continental margin (75ø18'S,175ø35'W)had a densitysimilarto that at 120m on station 217.

Salinitiesof all deeperstationsin Figure 5 convergeagain Subsurface meltingof glacialice influencesthe regionalwater columnwell beyondthe point where discretestepscan be near 500 m, possiblymarking the lower limit of glacial meltfollowed.Figure 5 is a compositeof the verticalCTD profiles ing in the Ross Sea. Below 500 m the water column may be for stations 211-219 in McMurdo Sound. The data fall into dominatedby the effectsof freezingat the baseof the RossIce three separateregimesin the depth interval correspondingto Shelf perimeter [Zotikov et al., 1980], on the walls of bottom in the floating glacial ice [Barrett, 1975; Clough et that of the EGT base.Station214 with the highestsalinityand crevasses lowesttemperatureis typical of the high-salinityshelf water al., 1975] and from deep convectionduring winter pack ice that predominatesin the southwestsectorof the RossSea [Ja- formation and in polynyasalong the Victoria Land coast. cobset al., 1970a]. Station 213 has the lowest salinity and DISCUSSION highesttemperature,resemblingseveralother stationson this These CTD measurements demonstrate the melt-driven part of the continentalshelfthat displaya warm core[Car and Codispoti, 1968,station9; Bourkland,1969,station5; Jacobset structuringof subsurfacecirculation into convectioncells of al., 1970b,station721]. The type 213 stationappearslesscom- limited vertical extent.This extendslaboratoryand Northern mon than the type 214, and mixing of the two would tend to- Hemisphereobservationsof layering near melting ice to Ant-



arctic field conditions where a weaker density stratification tions of increasedplant and animal I•re near the ice [e.g.,Ja-

prevails. Themeanlayerthickness isnear17m, aboutan or- cobset al., 1979b;,4inley and Jacobs,1981]. der of magnitude less than estimatedfor the Antarctic by Neshybaand Josberger[1980]. Relatively little of the glacial meltwater may reach the sea surface,thereby limiting effects on nutrient cycles and crystal formation in upwelling water [Neshyba,1977;Fotdvik and Kvinge, 1974]. Separationof the water column into discretedensitylevels should facilitate differential advectionor diffusion of properties within layersand retard vertical motionsacrossthe layers. This has implicationsfor the way in which important glacial

ice tracers,e.g.,the O•8/O'6 ratio [Weisset at., 1979],will be injectedinto the ocean.Individual stepsmay not persistunaltered very far from the source,but we have recordedsimilar, less distinct structures near the Ross Ice Shelf and elsewhere

,4cknowledgments.This work was supportedby NSF grantsDPP 77-22209and DPP 79-18674to ColumbiaUniversity;by the Ministry of Defence (ProcurementExecutive);and by U.K. Natural Environmerit ResearchCouncil grant GR3/2291 to ScottPolar ResearchInstitute.Logisticsupportwas providedby the National ScienceFoundation Division of Polar Programs.We have been ably assistedin variousaspectsof this studyby J. Szelag,A. Amos,D. Woodroffe,P. Woodroffe,A. Gordon,E. Jankowski,and H. Steed.The photographs were taken by the U.S. Navy (Figure 2a) and by S. Patla (Figure 2b). Severalcolleaguesmade helpful suggestions on the manuscript.Typing and drafting were done by A. Baker, K. O'Neill, and M. Turner. LDGO




Ainley, D. G., and S.S. Jacobs,Seabirdaffinitiesfor oceanand ice boundariesin the Antarctic, Deep Sea Res.,in press,1981. Barrett, P. J., Seawaternear the head of the Ross Ice Shelf, Nature,

in the Ross Sea. If melt-driven stratificationis widespread on the Antarctic continentalshelves,it could contributethrough256, 390-392, 1975. out the year to a circulationpatternmore alignedwith isopycBourkland,M., Oceanographiccruisesummary,RossSea,Antarctica, hals than heretoforemight have been realized. Feb. and Mar. 1969, Inform. Rep. 69-75, 11 pp., Naval Oceanogr. The existence of the long and narrow Erebus Glacier Office, Washington,D.C., 1969.

Tongueover severaldecadesillustratessomeaspectsof basal melting versusvertical sidewallmelting for the attachedAntarcticice shelvesand glaciers.Most recentanalysesof iceberg meltingin the seahaveconcentrated uponthe sidewallswhere melting per unit area may be more efficientthan at the base. Morgan and Budd [1978] computedcomparablemelt rates for the sidesand basesof icebergshowever,findingthe rate to be linearly proportionalto water temperatureand rangingfrom a few m yr-' in below-zerowaterto about 100m yr-• in 4-6øC water near the Polar Front (Antarctic Convergence).Ignoring horizontal spreadingand calving, net melting of the EGT

Car, M., and L. A. Codispoti, Oceanographiccruisesummary,Ross Sea, Antarctica, Feb. 1968, Inform. Rep. 68-64, 21 pp., Naval Oceanogr.Office,Washington,D.C., 1968. Chen, C. F., Onsetof cellularconvectionin a salinitygradientdue to a lateral temperaturegradient,J. Fluid Mech., 63, 563-576, 1974. Chen, C. F., D. G. Briggs,and W. A. Wirtz, Stabilityof thermal convection in a salinity gradient due to lateral heating, Int. J. Heat Mass Transfer,14, 57-65, 1971. Clough,J. W., K. C. Jezek,and J. D. Robertson,RISP drill site survey,,4ntarct.J. U.S., 10, 148-149, 1975. Dayton, P. K., G. A. Robilliard, and A. L. Devries,Anchorice formation in McMurdo Sound,Antarctica,and its biologicaleffects,Science, 163, 273-274, 1969.

sidewalls couldnot exceed2 m yr-• withoutalteringthe over- Drewry, D. J., P. T. Meldrum, and E. Jankowski,Radio echo soundall dimensionsof the Glacier. For comparison,a current ve-

locityof 10cm s-• alongthe 12.8km EGT length,substituted into equation (8) of Weeksand Campbell[1973], givesa melt

rate of 0.5 m yr-• for each0.1øCincrementof seawatertemperature above the freezing point. Jacobset at. [1979a] estimated somewhatlower valuesfor net melting betweenpoints under the Ross Ice Shelf.

Shallow melting will be curtailed in winter when the sea surfacefreezes.Deep melting will continuein winter [Robin, 1979]due to the effectof pressureon the meltingpoint (Ts in Figure 3) and to the lateral, subsurfaceinflow of warmer water as noted above.The Antarctic ice shelvesand glaciersexposeto seawatera basalarea more than 2 ordersof magnitude greaterthan the area of their vertical walls. In the Ross Sea and other continental margin areas we have investigated, there appearsto be a supply of heat from the Circumpolar Deep Water to the sub-iceregime.Thus basalmelting should have greater oceanographicand glaciologicalimpacts than wall melting evenif basalmassflux ratesare somewhatlower. In spiteof adequatenutrients in the Antarctic SurfaceWater, phytoplanktongrowth is limited there by the short season of sufficientsunlightand by a surfacelayer that can be deeply mixed by winds and seaice formation.In this nearly homogeneousenvironment,the melting of pack and glacial ice is no doubt biologicallyimportant. Enhancedstability of the shallower layers should limit the depth of mechanicalmixing, thereby retaining a greaterpercentageof viable phytoplankton within the euphoticzone. Finenko[1978]notesthe importance of this processin the Arctic and other regions.Particulates and biological forms could accumulateat the density discontinuities.Thesefactorsmay accountin part for observa-

ing of the Antarcticice sheet,1978-79,Polar Rec.,20, 43-51, 1980. Farmer, D. M., and J. D. Smith, Tidal interaction of stratified flow with a sill in Knight Inlet, Deep Sea Res., 27,4, 239-254, 1980. Finenko, Z. Z., Productionin plant populations,in Marine Ecology, vol. IV, chap. 2, Dynamics,edited by O. Kirme, pp. 13-87, John Wiley, New York, 1978. Foldvik, A., and T. Kvinge, Conditional instabilityof seawater at the freezingpoint, Deep Sea Res., 21, 169-174, 1974. Fujino, K., E. L. Lewis, and R. G. Perkin, The freezingpoint of seawater at pressuresup to 100 bars, J. Geophys.Res., 79, 1792-1797, 1974.

Gebhart,B., andJ. C. Molendorf,A newdensityrelationship for pure and salinewater, Deep Sea Res.,24(9), 831-848, 1977. Greisman,P., On upwellingdriven by the melt of ice shelvesand tidewater glaciers,Deep Sea Res., 26, 1051-1065, 1979. Heath, R., Circulationacrossthe ice sheffedgein McMurdo Sound, Antarctica,in Polar Oceans,editedby M. Dunbar,pp. 129-149, Arctic Instituteof North America,Calgary,Alberta, 1977. Holdsworth,G., Erebus Glacier Tongue, McMurdo Sound,Antarctica, J. Glaciol.,13(67), 27-35, 1974. Huppert,H. E., and E.G. Josberger, The meltingof icein cold stratified water, J. Phys. Oceanogr.,10, 953-960, 1980. Huppert,H. E., and J. S. Turner, On meltingicebergs,Nature, 271, 46-48, 1978.

Huppert,H. E., and J. S. Turner, Ice blocksmeltinginto a salinity gradient,J. Fluid Mech., 100, 367-384, 1980. Jacobs,S.S., A. F. Amos,and P.M. Bruchhausen, RossSeaoceanography and AntarcticBottomWater formation,DeepSea Res., 17, 935-962, 1970a.

Jacobs,S.S., P.M. Bruchhausen, and E. B. Bauer,EltaninReports, Cruises32-36, 1968,HydrographicStations,BottomPhotographs, Current Measurements,Tech. Rep. I-CM-1-70, 460 pp., LamontDoherty GeologicalObservatory,Palisades,New York, 1970b. Jacobs,S.S., A. L. Gordon, and J. L. Ardai, Circulation and melting beneath the Ross Ice Sheif, Science,203, 439-442, 1979a. Jacobs,S.S., A. L. Gordon, and A. F. Amos, Effect of glacial ice melting on the Antarctic Surface Water, Nature, 277, 469-471, 1979b.


Josberger,E.G., Laminar and turbulent boundarylayersadjacentto meltingverticalice wallsin salt water, Ph.D. thesis,Dep. of Oceanogr., Univ. of Washington,Seattle,1979. Josberger,E.G., and S. Neshyba,Icebergmelt-drivenconvectioninferred from field measurementsof temperature,Annals GlacioL,1, 113-117, 1981.

Loewe, F., On melting of fresh-waterice in seawater, J. GlacioL,3, 1051-1052, 1961.

Martin, S., E. Josberger, and P. Kaufman, Wave inducedheat transfer to an iceberg,in Iceberg Utilization,edited by A. Husseiny,pp. 260-264, Pergamon,New York, 1978. Morgan, V. I., and W. F. Budd,The distribution,movementand melt rate of Antarctic icebergs,in Iceberg Utilization,edited by A. Husseiny,pp. 220-228, Pergamon,New York, 1978. Neal, V. T., S. Neshyba,and W. Denner, Thermal stratificationin the


Tressler, W. L., and A.M. Ommundsen, Seasonal oceanographic studiesin McMurdo Sound, Antarctica, Rep. TR-125, U.S. Navy HydrographicOffice,Washington,D.C., 1962. U.S. HydrographicOffice, Sailing directionsfor Antarctica,Publ. 27, 433 pp., Washington,D.C., 1960. Weeks, W. F., and W. J. Campbell, Icebergsas a freshwatersource: An appraisal,J. Glaciol.,12, 207-233, 1973. Weertman, J., Deformation of floating ice shelves,J. Glaciol., 3, 3842, 1957. Weiss,R. F., H. G. Ostlund,and H. Craig, Geochemicalstudiesof the Weddell Sea, Deep Sea Res., 26, 1093-1120, 1979. Zotikov, I. A., V. S. Zagorodnov,and J. V. Raikovsky,Core drilling throughthe RossIce Shelf (Antarctica)confirmedbasal freezing, Science,207, 1463-1465, 1980.

Arctic Ocean, Science,166, 373-374, 1969.

Neshyba,S., Upwelling by icebergs,Nature, 267, 507-508. Neshyba,S., and E.G. Josberger,On the estimationof Antarcticiceberg melt rate, J. Phys.Oceanogr.,10, 1681-1685,1980. Robin, G. deQ., Formation, flow and disintegrationof ice shelves,J. Glaciol., 24, 259-271, 1979.


(ReceivedOctober20, 1980; revisedFebruary 23, 1981; acceptedApril 1, 1981.)

Suggest Documents