Journal of Glaciology , Vol. 49, No.166, 2003 Links between short-termvelocity variations and the subglacial hydrology ofa predominantly cold polythermal glacier Luke COPLAND,1 Martin J. SHARP,1 PeterW. NIENOW2 1Department of Earth and Atmospheric Sciences,University of Alberta,Edmonton, AlbertaT6G 2E3,Canada E-mail:[email protected] 2Department of Geography andTopographic Science,University of Glasgow,Glasgow G12 8QQ,Scotland ABSTRACT.Thesurface velocityof apredominantlycold polythermal glacier (John EvansGlacier ,Ellesmere Island,Canada) varies significantly on bothseasonal and short- ertime-scales. Seasonalvariations reflect the penetrationof supraglacial water to the glacierbed throughsignificant thicknesses ofcold ice. Shorter-term events areassociated withperiods of rapidly increasing water inputs tothe subglacialdrainage system. Early- seasonshort-term events immediatelyfollow the establishment ofadrainageconnection betweenglacier surface andglacier bed, and coincide with the onset ofsubglacial outflow atthe terminus. Amid-season short-term eventoccurred as surface meltingresumed fol- lowingcold weather ,andmay have been facilitated by partial closure ofsubglacial chan- nels duringthis coldperiod. There is acloseassociation between the timingand spatial distributionof horizontaland verticalvelocity anomalies, the temporalpattern ofsurface waterinput to the glacier,andthe formation,seasonal evolution and distribution of sub- glacialdrainage pathways. These factors presumablycontrol the occurrence ofhigh- water-pressure events andwater storage at the glacierbed. The observed coupling betweensurface waterinputs andglacier velocity may allow predominantly cold poly- thermal glaciersto respond rapidly to climate-induced changes in surface melting. 1.INTRODUCTION because:(a) water flow along intergranular vein networks is largelyabsent in cold ice, and (b) crevasses andmoulins, whichintroduce large-scale permeability to glacier ice, are Studies oftempera te andpredominantly warm polythermal rare onpredominantly cold glaciers due to low rates ofice de- glaciers(polythermal structure types bandc ofBlatter and formationand the refreezingof meltwater that drainsinto cre- Hutter,1991,fig.1)haveclearly demonstrated that there is an vasses. If this werethe case,the influenceof surface melting on intricate couplingbetween the subglacialhydrology and flow the flowof predominantly cold polythermal glaciers would be dynamicsof such glaciers(Iken, 1981;Bindschadler,1983;Iken limited. andothers, 1983;Iken and Bindschadler ,1986;Kam b,1987; Severalstudies, however,suggestthat the surfacevelocities Jansson,1 995,1996;Raymond and others, 1995;Harbor and ofsuch glaciersvary on a seasonalbasis (M u « ller andIken,1 973; others, 1997;Kavanaughand Clarke, 200 1;Mairand others, Iken,1974;Andreasen, 1985;Rabus and Echelm eyer,1997).This 2001,2002).Thiscoupling is especiallyimportant when surface is alsothe case onthe Greenlandice sheet, wheremelt -induced meltwater is ableto penetrate tothe glacierbed, and it is seasonalvariations in surface velocity have been observed in a fundamentalto such dynamicphenomena as glacier surges, regionwhere cold ice is 41200m thick(Zwally and others, seasonalvelocity variations and short-term high-velocity 2002).Thesestudies implythat surfacem eltwaters canand do events. Whether andhow this couplingaffects the dynamics penetrate tothe glacierbed through significant thicknesses of ofglaciers composed predominantly of ice atsub-freezing tem- ice atsu b-freezingtemperatu res, contraryto the suggestionof peratures is less clear.Thegoal of this paperis therefore toin- Hodgkins( 1997).Theyalso suggest that coupling between sur - vestigatethe relationshipsbetween surface melt, subglacial facemelting and the flowof predominantl ycold,but warm - hydrologyand the flowof a predominantlycold polythermal based,ice masses mayprovidea mechanismby which such ice glacierin the CanadianHigh Arctic. masses canrespond rapidly to changes in surface weather and Predominantlycold polythermal glaciers (polythermal climate (Zwallyand others, 2002).Givenmodel predictions structure types dande ofBlatter andH utter,1991,fig.1)are thatanthropogenicclimate warmingwill be m ost markedin characterizedby a thickmantle of cold ice overlyinga limited northernhigh latitudes (Manabeand others, 1991),andthe areaof temperate ice at,and immediately above, the glacier potentialcontri butionof Arctic glaciersto global sea level, bedin the ablationarea. Exam ples includeWhite Glacier,Axel understandingof this couplingis ahighscientific priority. HeibergIsland, Canada (Blatter ,1987),andMcCall Glacier , Alaska,U .S.A.(Rabus andEchelmeyer ,1997).Arecent review 1.1.Study site (Hodgkins,1 997)suggests thatpenetration of surface-derived meltwaters tothe beds ofsuch glaciersis limited. Thisis Thestudy was conducted at J ohnEvans Glacier ,a 165 km2 ¹ 337 Copland and others:Velocity variations and subglacial hydrology of apolythermal glacier ablationarea where the velocitymeasurements weremade (Coplandand Sharp, 200 1).Forthe period1 997^99,the mean annualair temperature at820 m a.s.l.was ^1 5.2³C. Highbed reflection powers in radio-echo sounding records indicatewarm ice atthe bedthroughout most of the lowerablation zone, except along the glaciermargins andwhere the ice isthin (Coplandand Sharp ,2001).Acon- tinuousinternal reflecting horizon over the centre ofthe lowerterminus suggests that the warmbasal ice reaches an averagethickness of20 m there. Inthe accumulationand upperablation areas, low bed reflection powers and 1 5m boreholetemperatures of^9.5 to ^1 5.1³C suggestthat the ice iscoldthroughout. Themelt seasontypically occurs betweenearly Juneand earlyAugust. At the start ofthe melt season,meltwater either pondson the glaciersurface orisrouteddirectly to the ice margins,and is unableto access the glacierinterior . Once englacialdrainage of supraglacially derived melt- waters is initiated,however ,approximately25% of the glaciersurface areadrains into moulins in a crevasse field atthe topof the terminus (Fig.2).Later inthe melt season, Fig.1.Bed topography (ma.s.l.)andlocation ofJohn Evans meltwater from upto 40% of the surface areaof the glacier Glacier. drainsto the glacierbed viathese moulinsand a secondcre- vasse field 6kmfurther upstream. Dye tracer experiments ¹ confirmthe linkbetween these waterinput locations and a predominantlycold polythermal valley glacier on the east majordrainage portal at the terminus (Binghamand coastof Ellesmere Island,N unavut,Canada ( 79³40 ’ N, others, 2003).Largeincreases inthe suspended-sediment 74³30’ W;Fig.1).Theglacier ranges in elevation from 1 00to content andelectrical conductivity(EC) ofthe wateras it 1500m a.s.l.,with the long-term equilibriumline at 750^ passes throughthe glaciersuggest that it is routedsub- ¹ 850m a.s.l.Ice depths reacha maximumof 400 m inthe glacially(Skidmore and Sharp, 1999). upperablation area, and average 1 00^250m inthe lower Inthe earlysummer ,the meltwater is trappedat the glacier Fig.2.Landsat 7image of John Evans Glacier (path049 ,row 002,10 July1999 )showing the location of the velocity stakes (black dots),geophones (whitedots ;LG lower geophone,MG middle geophone,UG upper geophone),surveying base ˆ ˆ ˆ stations (whitesquares) ,stream gauging stations (triangles) and artesian fountain observed in1998 ( ).Crevasse field indi- cates location where most supraglacial meltwater reaches the glacier bed. Lower weather station is located next to the middle geophone.Black box indicates horizontal extent of Figures 4and7. 338 Copland and others:Velocity variations and subglacial hydrologyofapolythermal glacier wasused toconvert the displacements tovelocities in cm d ^1 (24hours) .Velocitypatterns were interpolatedover the entire terminus regionfrom the pointmeasurements atthe stakes.Velocities wereset to0 aroundthe glacieredge, which isreasonablegiven that the glacierappears to be frozento its bedat the margins(Copland and Sharp, 200 1).Allinterpo- lationswere completed with the ``v4’’interpolationroutine inMatlab, which is basedon biharmonicspline interpola- tion(Sandwell, 1 987).Interpretations ofvelocity patterns aremade only for areas where stakes arepresent. Forthe purposes ofour discussion, the lowerterminus isdefinedas the areasouth of the uppergeophone (UG inFig .2),andthe upperterminus is definedas the areato the northand northwest ofthis geophone. 2.2.Surface velocity errors Fig.3.Artesian fountain observed on the lower terminus From the Geodimeter technicalspecifications, uncertainties between days 180 and186,1998; see Figure 2for location. indistance measurements amountto (2 mm 3 ppm). § ‡ Thisequates to 5mm overa typicalsurvey distance of § bedbehind the frozenterminus (Skidmoreand Sharp, 1 999). 1km.Angle measurements were madeto a resolutionof 2 ’’, Theinitial release ofsubglaciallyrouted meltwater from the whichequates toa maximumpotential error of 4.8 mm § glaciertypically occurs viaan artesian fountain on the overa distanceof 1 km.Instrument drift duringsurveys glaciersurface (Fig.3)and/ orupwellingthrough sediments wascorrected forby resurveying the positionof reference afewmetres infrontof the snout.Since observations began markers after everyseven or fewer stake measurements and in1 994,dates ofinitiationof subglacialoutflow have varied assumingthat drift waslinear between surveys. In the fol- between22 J uneand 1 1July(days 1 73^192).Asoutflowcon- lowingdiscussion, three types oferror areevaluated: tinues, the upwellingmigrates towardsthe snout,and is (i) Position