The role of sediment supply in esker formation and ice tunnel evolution

lable at ScienceDirect

Quaternary Science Reviews 115 (2015) 50e77

Contents lists avai

Quaternary Science Reviews

journal homepage wwwelsevier comlocatequascirev

The role of sediment supply in esker formation and ice tunnelevolution

Matthew J Burke a Tracy A Brennand a Darren B Sjogren b

a Department of Geography Simon Fraser University 8888 University Drive Burnaby BC V5A 1S6 Canadab Department of Geography University of Calgary 2500 University Drive Calgary AB T2N 1N4 Canada

a r t i c l e i n f o

Article historyReceived 18 August 2014Received in revised form23 February 2015Accepted 25 February 2015Available online

KeywordsEskerIce tunnelGlacial hydrologyGround-penetrating radar (GPR)Electrical resistivity tomography (ERT)GeomorphologySedimentary architecture

Corresponding author Tel thorn1 778 782 3617 faxE-mail address tabrennasfuca (TA Brennand)

a b s t r a c t

Meltwater is an important part of the glacier system as it can directly influence ice sheet dynamicsAlthough it is important that ice sheet models incorporate accurate information about subglacialmeltwater processes the relative inaccessibility of contemporary ice sheet beds makes direct investi-gation challenging Former ice sheet beds contain a wealth of meltwater landforms such as eskers that ifaccurately interpreted can provide detailed insight into the hydrology of former ice sheets Eskers are thecasts of ice-walled channels and are a common landform within the footprint of the last Laurentide andCordilleran Ice Sheets In south-western Alberta esker distribution suggests that both water and sedi-ment supply may have been important controls the longest esker ridge segments are located withinmeltwater valleys partially filled by glaciofluvial sediments whereas the shortest esker ridge segmentsare located in areas dominated by clast-poor till Through detailed esker ridge planform and crest-typemapping and near surface geophysics we reveal morpho-sedimentary relationships that suggest eskersedimentation was dynamic but that esker distribution and architecture were primarily governed bysediment supply Through comparison of these data with data from eskers elsewhere we suggest threeformative scenarios 1) where sediment supply and flow powers were high coarse sediment loads resultin rapid deposition and rates of thermo-mechanical ice tunnel growth is exceeded by the rate of icetunnel closure due to sediment infilling High sedimentation rates reduce ice tunnel cross-sectional areacause an increase in meltwater flow velocity and force ice tunnel growth Thus ice tunnel growth isfastest where sedimentation rate is highest this positive feedback results in a non-uniform ice tunnelgeometry and favours macroform development and non-uniform ridge geometry 2) Where sedimentsupply is limited but flow power high the rate of sedimentation is less than the rate of thermo-mechanical ice tunnel growth Here the ice tunnel enlarges faster than it fills with sediment and itsevolution is independent of sedimentation resulting in more uniform ice tunnel geometry In these casesesker architecture is dominated by extensive vertical accretion of tabular units and ridge geometry ismore uniform 3) Where sediment is truly supply-limited the sedimentation rate is negligible regardlessof water supply and like scenario 2 ice tunnel growth is independent of sediment deposition forming arelatively uniform ice tunnel (or eroding the bed) Because meltwater flows transport few gravel claststhe ice tunnel is not completely filled with gravel and instead saturated and pressurized diamicton orbedrock (if deformable) from beneath the surrounding ice is ldquosqueezedrdquo into the relatively low pressureice tunnel during waning flow (or after ice tunnel shutdown) resulting in deformation of limited gravelsdeposited within the ice tunnel and a landform cored with diamicton or deformed bedrock and with arelatively uniform ridge geometry Our data demonstrate that an esker map is a minimum map of ice-walled channel location and that continued detailed investigation of morpho-sedimentary relation-ships is essential to gaining a complete picture of esker forming processes Validating the morpho-sedimentary relationships identified in south-western Alberta (and other areas) with a larger data setmay allow improved remote predictive esker mapping over larger areas and inferences to be made aboutspatial and temporal variations in esker depositional environments and ice tunnel evolution

thorn1 778 782 5841

MJ Burke et al Quaternary Science Reviews 115 (2015) 50e77 51

1 Background and rationale

Given that evolution of the subglacial hydrologic system candirectly influence ice dynamics it is important that ice sheetmodels incorporate accurate information about the timescales andthe processes responsible for ice tunnel evolution However therelative inaccessibility of contemporary ice sheet beds makes directinvestigation of subglacial hydrologic systems challenging (Frickerand Scambos 2009) Former ice sheet beds contain a wealth ofglaciofluvial landforms such as eskers that if correctly interpretedcan provide detailed insight into the hydrology of former andcontemporary ice sheets Eskers typically defined as ridges ofstratified sand and gravel are the main depositional signature ofchannelized meltwater flow on within or beneath ice masses(Banerjee and McDonald 1975) and record the location of ice-walled channels on former ice sheet beds Consequently eskershave been used to infer former ice sheet dynamics (eg Shilts1984 Aylsworth and Shilts 1989 Clark and Walder 1994Brennand 2000 Storrar et al 2014) and to test numericalmodels (eg Boulton et al 2009) However much debate remainsas to the processes responsible for esker formation (Cummingset al 2011) and although recent work has shown that the glacialhydrologic system is never in steady state (eg Gray 2005Wingham et al 2006 Fricker et al 2007 Bell 2008 Stearnset al 2008 Bartholomaus et al 2011) and that eskers may actu-ally record relatively short lived and dynamic events (egBrennand 1994 Burke et al 2008 2010 2012a Cummings et al2011) numerical models have typically assumed steady stateesker formation (eg Hooke and Fastook 2007 Boulton et al2009) Detailed work on glacial lake outburst flood (GLOF) eskersat contemporary glaciers has demonstrated that the balance be-tween water supply and sediment supply drive esker formation(Burke et al 2008 2010) These observations are echoed in thelandform record and suggest that dynamic esker deposition isfundamentally controlled by variations in ice tunnel geometry(accommodation space) and sedimentation rate (Brennand 1994Brennand and Shaw 1996 Burke et al 2012a) Over the shorttimescales of transient events often recorded in eskers(Shulmeister 1989 Brennand 1994 2000 Brennand and Shaw1996 Cummings et al 2011) ice tunnel geometry is a balancebetween growth by frictional melting and mechanical excavationand ice tunnel closure through sedimentation rather than ice creep(Burke et al 2012a) It is imperative that numerical models areconsistent with the landform record thus they should incorporateboth the transient conditions as well as seasonal drainage patterns(Burke et al 2012b) However most investigations of eskers havefocused on regions containing the largest examples and the largestexamples within regions (Shilts 1984 Brennand 1994 Burke et al2008 2010 2012a Cummings et al 2011 Storrar et al 2014)where there was both a ready supply of meltwater (Brennand2000 Storrar et al 2014) and sediment (Brennand 2000) Thelack of large eskers on the Prairies has resulted in only limitedinvestigation of these landforms (Stalker 1960) yet understandingwhat controlled their architecture and apparent regionallyderanged distribution (Brennand 2000) could provide furtherinsight into the fundamental controls on esker formation Theassertion that esker distribution in North America was primarilysubstrate controlled (Clark andWalder 1994 Boulton et al 1996) isclearly insufficient to explain the presence of 506 esker ridge seg-ments in south-western Alberta (Fig 1) The suggestion thatmeltwater supply controlled the presence of the longest eskers onthe Canadian shield (Brennand 2000 Storrar et al 2014) fails toaddress what controlled the distribution of the eskers in thewestern Canadian sedimentary basin It is only through investiga-tion of these smaller esker systems coupled with our knowledge of

their larger counterparts and observations from contemporary icemasses that a unifying theory of the controls on esker formationcan be achieved

In this paper we combine detailed geomorphic and geophysicaldata from four eskers in south-western Alberta in order to assessthe controls on their formation and distribution Because theseeskers have distinct geomorphic characteristics and depositionalsettings we compare their morpho-sedimentary characteristics toidentify the controls on their variability Through comparison ofthese data with detailed morpho-sedimentary investigations ofeskers elsewhere we suggest models for esker formation thatprovide insight into the major controls on the evolution of thesubglacial hydrologic system at the ice tunnel scale and ultimatelytake a step closer to a unified understanding of the controls onesker formation

11 Regional setting

The plains of southern Alberta are situated on south-west dip-ping Tertiary and Cretaceous aged sandstones and shales that thineastward from the foothills of the RockyMountains This bedrock isoverlain by a carapace of glacial material which due to the friablenature of local bedrock typically has a fine-grained textureparticularly in areas overlaying Cretaceous shales (Pawluk andBayrock 1969) These sediments record the coalescence of iceflowing eastward from the Cordilleran Ice Sheet and ice flowingwestward from the Laurentide Ice Sheet (Jackson et al 2011) Theseflows converged and deflected toward the south-southeast acrossthe plains of southern Alberta (Klassen 1989 Evans et al 2008)reaching a maximum extent about 20e21 14C ka BP (Dyke et al2002) This southerly ice flow and its associated meltwater havegenerated a wealth of glacial landforms including mega-flutesmega-lineations tracts of smoothed terrain moraines meltwaterchannels eskers and buried valleys (eg Sjogren and Rains 1995Munro and Shaw 1997 Sjogren et al 2002 Evans et al 20082012) Although till is often thin and fine grained it thickens atlarge moraine complexes where multiple units are stacked (Evanset al 2008) Together these sediments and landforms have beenassigned to landsystems associated with rapid ice flow (Evans2000 Evans et al 2008 OCofaigh et al 2009) or inferred to re-cord meltwater erosion (Rains et al 1993 Shaw et al 1996) Eskersin south-western Alberta range in length but the two longest es-kers (Strathmore and Clear Lake eskers) are located within broadmeltwater valleys (Fig 1) that are orientated northwest-southeastand probably occur downflow of ice-dammed lakes (St-Onge1972) Despite the plethora of eskers in south-western Alberta(Shetsen 1987 1990 Evans et al 2008) they have received littledetailed attention probably due to their relatively small size(making identification in aerial photographs and on digital eleva-tion models (DEMs) difficult) and lack of major sedimentexposures

2 Methods

21 Landform mapping and quantification

We mapped esker ridge segment crestlines esker system geo-morphology meltwater valley walls and the geomorphology ofmeltwater valley fill in a GIS Esker crestlines and meltwater valleywalls were mapped as polylines using LiDAR-derived DEMs (15 mpostings vertical and horizontal accuracies of 03 m and 05 mrespectively AltaLIS 2014) with multiple hillshade illuminationsWe used sub-regional patterns to differentiate eskers from mo-raines and used elevated linear breaks in slope to delineate melt-water valley walls Meltwater valleys typically contain underfit

Fig 1 Topography of south-western Alberta showing locations of eskers mapped from LiDAR-derived digital elevation models (DEMs 15 m postings vertical and horizontalaccuracies of 03 m and 05 m respectively AltaLIS (2014)) The location of the study area is located on the inset political map of central and western Canada northwest USA Thelabelled boxes indicate the locations of panels aec in Fig 2

MJ Burke et al Quaternary Science Reviews 115 (2015) 50e7752

streams no streams andor eskers When mapping esker systemand meltwater valley fill geomorphology (as polygons in a GIS)stereographic aerial photographs ground observations and waterwell records (where available) were used alongside the DEMs Wequalify the lsquoeskerrsquo as being composed of multiple longitudinally-

aligned lsquoesker ridge segmentsrsquo along with the short gaps thatseparate them We use lsquoesker systemrsquo to refer to the esker andassociated fans An lsquoesker networkrsquo includes a lsquotrunkrsquo esker andmultiple lsquotributaryrsquo eskers andor multiple longitudinally-alignedesker systems that are separated by gaps gt5 km We qualify esker

morphology based upon its topographic long profile cross profilesand planform (Figs 2 and 3) following the criteria outlined byBurke et al (2012a) 1) round-crested ridge segments approximatehalf cylinders (Fig 3a b) 2) flat-topped ridge segments approxi-mate half hexagonal prisms 3) sharp-crested ridge segmentsapproximate triangular prisms (Fig 3a) and 4) ridge enlargementsare segments where the esker ridge abruptly increases in width(and likely height) (Fig 3a c) No flat-topped ridge segments wereidentified and ridge segments are either singular or anabranchedAlthough round-crested ridge segments typically have uniformlong-profiles (relatively flat with only minor metre-scale

Fig 2 Geomorphic maps of a) Strathmore esker system b) Clear Lake esker system and c)model (AltaLIS 2014) Polygons of glaciolacustrine sediment are from Shetsen (1987) TheHolocene material

undulations) and cross-sections in some examples they may benon-uniform (include localized and abrupt increases in ridgeheight ridge heightenings) Lobate fan-shaped mounds of sedi-ment in close proximity or terminal to esker ridges are mapped asfans Meltwater valley fill is identified as valley-wide glaciofluvialdeposits forming hummocky kettled plains containing circularlakes elevated above inset (Holocene) alluvium We quantifiedesker ridge height and width by extracting elevation profiles(within the boundaries of esker ridge polygons) from the DEM atregular intervals along each esker (total of 214 profiles from alleskers) Esker height was established for each profile by subtracting

Didsbury esker network and Torrington esker ridge overlain on a hill-shaded terrainareas not covered by glaciolacustrine sediment are undifferentiated Quaternary and

Fig 3 a) Representative topographic cross profiles of round-crested and sharp-crested esker ridge segments as well as ridge enlargements Photographs of a typical b) round-crested esker ridge segment and c) esker ridge enlargement Truck (circled) for scale in b The white line in c highlights the esker long profile note the downflow drop inelevation from the ridge enlargement (right) to a connected round-crested ridge segment (left)

the lowest elevation from the highest Esker width was taken as theplanimetric polygon width Esker ridge continuity is calculated by

ethsum segment length=ethsum segment lengththorn gapsTHORNTHORN 100

22 Geophysical data collection processing and inversion

Because sediment sections are rare within eskers in south-western Alberta shallow geophysical techniques were employedto investigate the sedimentary architecture of the Strathmore(Fig 2a) Clear Lake (Fig 2b) Torrington (Fig 2c) and Didsburyeskers (Fig 2c) In total gt17 km of 100 MHz common offset (CO)ground-penetrating radar (GPR) lines were collected as nine grids(Fig 2 and Figs S1 and S2 in supporting information) along thelength of the eskers using a Sensorsamp Software Inc pulseEKKO PROsystem (Table 1) Grid location was primarily governed by site ac-cess and suitability to geophysical surveying but an attempt was

Table 1Summary of GPR data collection

Esker Grid no Ridge morphology

Strathmore 1 Ridge enlargement2 Round-crested3b Round-crested

Clear Lake 4 Round-crested5c Round-crested

Downglacier Didsbury 6 Ridge enlargement amp round-crestedUpglacier Didsbury 7d Round-crested

8e Round-crestedTorrington 9d Ridge enlargement amp round-crested

a Total line length is the sum of individual line lengths within the gridb GPR performance at grid 3 was very poor due to the low resistivity of surface material

manuscript because they duplicate other data (eg grid 2 Fig 7aec)c Geophysical data for grid 5 are presented in supporting information (Figs S26e37)d Only the GPR lines that intersect ERT profiles are presented because GPR performance

core of the esker The GPR lines not shown for grids 7 and 9 are similar to those presene GPR performance was poor because of the presence of significant volumes of low resi

poor site access a select set of representative GPR lines are presented (Fig 16)f Semblance velocity analysis of CMP data collected at the Torrington esker was unsucc

for all surveyed eskers in southern Alberta

made to incorporate the range of morphological elements presentDuring CO data collection the GPR antennas were co-polarised andperpendicular-broadside to the survey line to reduce reflectionsfrom offline sources (Arcone et al 1995) Antennas had a constantseparation of 1 m and data were collected in step mode (025 m)along the lines to improve ground coupling and trace stacking (32traces) Thirteen common mid-point (CMP) profiles collected at allesker grids provide estimated subsurface average velocities(Table 1) that were used to convert two-way travel time (TWT) intodepth and for data processing

Electrical resistivity tomography (ERT) lines were collected ateach grid except Didsbury esker grid 8 (due to site access) using aneight channel Advanced Geosciences Inc Super Sting system In-strument configuration is summarized in Table 2 All GPR and ERTlines were surveyed for topographic variation using a real-timekinematic differential global positioning system Where possiblesediment sections and surface materials were examined alongside

No of lines Total line length (m)a Mean GPR wave velocity (mns)

18 1708 011285 (plusmn 00094)16 22292 300

13 2343 012532 (plusmn 00016)23 4536514 16715 009513 (plusmn 001)16 147711 123612 1612 01149f

ERT profiles are presented in supporting information (Fig S17) rather than themain

because they duplicate the architectural style at grid 2was poor due to the presence of significant volumes of low resistivity material in theted in Fig 15stivity material in the esker core Because no ERT data were collected at grid 8 due to

essful Thus the velocity used for data processing and depth correction is an average

Table 2Summary of ERT configuration

Esker Grid no Line noa Length (m) Array type Max n value Electrode spacing (m) Data removal () RMS ()b L2c

Strathmore 1 Y5 95 Dipoleedipole 6 1 67 342 0412 Y1 1665 15 62 277 071

Y4 111 1 31 292 0923 X1 1665 15 48 327 086

Y1 112 1 29 293 086Clear Lake 4 Y8 1665 15 75 297 098

5 Y18 1665 15 71 497 069Didsbury 6 Y1 1665 15 26 347 13

7 Y7 111 1 47 291 091Torrington 9 Y2 111 1 89 294 096

Y5 111 1 16 282 086

a Line numbers correspond to those used for GPR data with the exception of grid 3 (Strathmore esker) where only poor quality GPR data were collected and so not includedin this paper

b Root mean squarec Squared and summed differences between apparent and actual resistivity values

material collected from near surface augering (lt2 m depth) andfrom water well records (where available) to facilitate geophysicaldata interpretation

GPR data processing was carried out in REFLEXW v60 andincluded static correction lsquodewowrsquo filtering bandpass filteringmigration background removal filtering application of a gainfunction and topographic correction ERT inversionwas carried outin AGI EarthImager 2D v240 using the finite element method as aforward model a smooth model inversion routine damped trans-form topographic correction removal of negative resistivity valuesa minimum voltage filter and minimum and maximum apparentresistivity filters Inversion resulted in a maximum of 16e75 dataremoval RMS (root mean squared) errors of 277e497 and l2

values (squared and summed differences between apparent andactual resistivity values) of 041e13 (Table 2)

23 Geophysical data interpretation

In GPR profiles high-amplitude continuous reflections thatdemarcate termination of reflections above and below are identi-fied as bounding surfaces that define radar elements Onlybounding surfaces that can be traced across multiple GPR profiles(where possible) are picked Radar elements are labelled inapproximate stratigraphic order This order was based on super-imposition (wherever possible) and correlation of radar elementsnot directly superimposed that have a similar depth geometry andinternal characteristics These elements are interpreted usingmethodology employed by Burke et al (2012a) whereby radar el-ements are individually classified by accretion type This classifi-cation is based upon the geometry of reflections within individualradar elements as well as ERT characteristics (where appropriate)Classification of accretion types is independent of landform growthLandform growth is inferred by radar element stacking (egupglacier stacking means upglacier growth) which may be inde-pendent of radar element accretion type (eg vertically accretingradar elements may be stacked in an upglacier direction indicatingupglacier landform growth) We identify seven stratigraphic andarchitectural elements (Fig 4)

1 Broad meltwater valley fill elements are composed of relativelycontinuous reflections that are typically sub-horizontal regard-less of GPR line orientation They form tabular radar elementsthat are stratigraphically below less uniform elements and mayextend beyond the esker flanks suggesting they are moreextensive than the esker In ERT profiles these elements haveresistivity values of ~250e5000 U-m and are similar to broadcanal fill deposits identified by Burke et al (2012a b)

2 Diamicton or bedrock forms a low-resistivity (~2e200 U-m) unitin ERT profiles across esker ridges and in GPR profiles its upperbounding surface forms a high-amplitude irregular (sometimesroughly convex-up) reflection below which the GPR signalrapidly attenuates In exposures this unit appears to becomposed of diapiric clast-poor diamicton with diapiric struc-tures orientated pseudo-vertically towards the esker crestDiamicton-cored eskers have been previously reported inAlberta (cf Stalker 1960) The ERT characteristics of these unitsare similar to diamicton identified in other geophysical surveys(cf Burke et al 2012a b) though the slightly lower resistivitiesin south-western Alberta are likely associated with a muchfiner-grained matrix texture (cf Pawluk and Bayrock 1969)Alternatively these lower resistivities could record soft sedi-mentary bedrock These units are primarily identified in the ERTbut are classified away from the ERT by tracing its correspondingupper bounding surface through the GPR grid Where the upperbounding surface rises into the esker ridge in cross-section weclassify it as deformed diamicton or bedrock

3 Vertical accretion is identified by sub-horizontal reflectionswithin the esker ridge that are typically conformable to oronlap the lower bounding surface of the radar element Thesereflections are consistent with sand and gravel sheets previouslyreported in GPR studies (Heinz and Aigner 2003 Wooldridgeand Hickin 2005 Burke et al 2008 2010 2012a) Vertically-accreting radar elements often have a broadly tabular geome-try though theymay have beenmodified by the lower boundingsurface of stratigraphically younger radar elements Resistivityof these units range from ~100 to 10700 U-m

4 Headward accretion is recorded within the esker ridge by re-flections that dip upflow at an angle (up to ~18 from horizontal)conformable to or greater than that of the lower boundingsurface Typically upglacier-dipping reflections form upglacier-dipping radar elements that are consistent with backset bedsreported in other GPR studies (Fiore et al 2002) and sedimen-tary exposures within eskers (eg Brennand and Shaw 1996)

5 Downflow accretion is recorded within the esker ridge by re-flections that dip in a downflow direction at an angle (up to ~15

from horizontal) conformable to or greater than that of thelower bounding surface Downflow accreting radar elementstypically have a geometry that dips in a down-glacier directionand reflection characteristics are similar to large cross beds re-ported from GPR data collected on eskers (Fiore et al 2002Burke et al 2008 2010 2012a) and sedimentary exposureswithin eskers (eg Brennand and Shaw 1996) These foresetbeds typically record downflow accretion of sand and gravel

Fig 4 Summary of radar elements (RE) and stratigraphy Each radar element is classified into accretion types following the criteria set out in the figure

sheets on the lee-side of older deposits (Wooldridge and Hickin2005) and have resistivity values of ~1500e2200 U-m

6 Lateral accretion is identified within the esker ridge in flowperpendicular lines by reflections that dip normal to flow at anangle (up to ~20 from horizontal) conformable to or greaterthan that of the lower bounding surface In flow parallel linesreflections appear sub-horizontal These beds are consistentwith lateral accretion of gravel sheets (Wooldridge and Hickin2005 Kostic and Aigner 2007) and have been reported in es-kers elsewhere (Burke et al 2012a)

7 Slumped material is recorded by asymmetrical trough-shapedradar elements on the esker flanks that increase in thicknesstoward the base of the esker flank The lower bounding surfaceof these elements truncates deeper reflections and the ele-ments are composed of discontinuous and slightly derangedreflections (reflection offsets are common) In ERT profiles theytypically have resistivities of ~400e5000 U-m Slump depositsare common within eskers (Banerjee and McDonald 1975) asthey are associated with removal of lateral ice support and thecharacteristics of those imaged in Alberta are similar to thosefrom eskers elsewhere (Burke et al 2012ab)

8 Lake sediments or fine-grained diamicton radar elements are thestratigraphically-highest elements on esker ridges and may becomposed of sub-horizontal reflections that drape the lowerbounding surface Given these elements are often thin (lt3 mthick) internal reflections may be poorly resolved (due to theclose proximity of the elements to the airwave and ground-wave) but they can be linked to very low resistivity (~300 U-msimilar to that recorded from equivalent material elsewhere inAlberta Smith and Sjogren 2006) units in ERT profiles and thepresence of fine sands andor silts observed at the landformsurface in sediment exposures and shallow augering Wheresediment exposures do exist a distinction between fine-graineddiamicton and lake sediments may be made If no ground truthinformation is available we rely on the surficial geology map ofShetsen (1987) to identify whether glaciolacustrine sediment ismapped close to the grid location If so these elements areclassified as lake sediments and if not then they are classified asfine-grained diamicton (likely ablation till left behind during icesheet decay) However we acknowledge that these in-terpretations are tentative given the resolution of mapping inShetsen (1987)

Table 3Summary of esker ridge dimensions based on the detailed mapping in Fig 2

Esker ridge (length)a Morphologyb M

Strathmore (~38 km) Round-crested (50)Ridge enlargement (16)All (66)

Clear Lake (~39 km) Round-crested (43)Ridge enlargement (14)All (57)

Down-glacier Didsbury (~12 km) Round-crested (29)Sharp-crested (5) 1Ridge enlargement (10) 1All (44)

Up-glacier Didsbury (~95 km) Round-crested (32)Torrington (~15 km) Round-crested (13)

Ridge enlargement (2)All (15)

a Including gapsb The bracketed numbers refer to the number of elevation cross profiles used to calcuc This refers to landform height above surrounding terrain which typically underestim

post-depositional burial and ldquoinsettingrdquo into the substrate in some instances)d This should be considered a maximum estimate or primary esker width given that p

3 Esker system types in south-western Alberta and theirmorphology

There are 506 esker ridge segments in the study area that occurin two environments (Fig 1) 1) within meltwater valleys (50 ofesker ridge segments with a combined length of ~148 km) and 2)within areas mapped as clast-poor till by Shetsen (1987) (50 ofesker ridge segments with a combined length of ~133 km) Mostesker systems are either entirely within meltwater valleys (egStrathmore and Clear Lake esker systems) or entirely within areasof clast-poor till (eg Torrington esker ridge) but in some caseseskers may straddle both environments (eg Didsbury eskernetwork) sitting upon clast-poor till for some of their length andwithin meltwater valleys for the remainder (Figs 1 and 2c) In thispaper we examine four eskers that cover this range of depositionalenvironments The geomorphology of these esker systems wasmapped in detail (Fig 2 Table 3) and in all cases variations in ridgesegment morphology do not correlate to changes in substratetopography (Fig 5) as proposed by Shreve (1985) (eg flat toppedridge segments are not associated with adverse slopes) Althoughmost of these esker ridges have undulating long-profiles at the km-scale at the hectometre (hm) scale (eg grid- or ridge segmentscale) they are typically uniform with occasional non-uniformsections Generalized descriptions of each esker systemnetworkare presented detailed measurements are presented in Table 3 Inall cases reference to esker system area excludes gaps whereassystem length includes gaps

31 Strathmore esker system

The Strathmore esker system (~32 km2) is mainly a single ridgesitting within the walls of a meltwater valley that includes exten-sive valley fill (~47 km2) (Fig 2a) Although the esker ridge has anundulating long profile at the km-scale (Fig 5) at the hm-scale it isrelatively uniform (non-undulating) The esker is semi-continuous(629 continuity) and flanked by kettle holes along its entirelength (Fig 2a) It is composed of thirty-five longitudinally-alignedridge segments that are ~02e45 km long and classified as round-crested (lt100 mwide 437 of system area) or ridge enlargements(gt100mwide 101 of system area) A fan (partly modified by laterfluvial activity) occurs at the esker terminus (462 of system area

ean height (m)c Mean width (m)d Mean widthheight

3 plusmn 1 69 plusmn 14 23 plusmn 126 plusmn 2 135 plusmn 23 25 plusmn 104 plusmn 2 85 plusmn 33 24 plusmn 116 plusmn 3 108 plusmn 22 21 plusmn 118 plusmn 2 205 plusmn 35 26 plusmn 77 plusmn 3 135 plusmn 51 23 plusmn 104 plusmn 2 52 plusmn 14 16 plusmn 80 plusmn 1 58 plusmn 10 6 plusmn 11 plusmn 5 124 plusmn 25 15 plusmn 96 plusmn 4 43 plusmn 46 15 plusmn 102 plusmn 1 46 plusmn 10 28 plusmn 164 plusmn 2 61 plusmn 13 19 plusmn 75 plusmn 2 126 plusmn 5 30 plusmn 144 plusmn 2 70 plusmn 26 21 plusmn 9

late the mean heights and widthsates the height of the sediment body comprising the esker (ie height is reduced by

ost-depositional slumping has the effect of increasing width in many cases

Fig 5 Topographic long profiles of esker crestlines (classified by morphologic ridge type) and the terrain (till surface meltwater valley floor) immediately adjacent to the esker Thelocations of GPR andor ERT grids and subaqueous fans are indicated

but only ~2 km of its length) but the lack of any further fansupglacier suggests the esker was formed synchronously (cfBrennand 2000)

32 Clear Lake esker system

The Clear Lake esker system (~92 km2) is a single ridge that sitswithin thewalls of a meltwater valley that includes extensive valleyfill (~496 km2) The km-scale undulating esker ridge long profile(Fig 5) is not reflected at the hm-scale where it is typically uniformThe esker is semi-continuous (515 continuity) often flanked bykettle holes along its entire length (Fig 2b) and is composed ofeighteen longitudinally-aligned ridge segments that are~02e38 km long These ridge segments are classified as round-crested (lt150 m wide 22 of system area) and ridge enlarge-ments (gt150 mwide 8 of system area) Four fans are identified ataround 225 km 265 km 33 km and 38 km (esker terminus) alongthe esker system (Figs 2b and 5) (70 of system area but only~11 km of the total system length) Although the presence of threefans upglacier from the esker system terminus may suggest time-

transgressive fan deposition (cf Brennand 2000) the lack of fansupglacier from fan 4 and superimposition of fan 3 onto the eskerfavours synchronous esker ridge formation prior to fan deposition(cf Brennand 2000) at least of its northern half

33 Didsbury esker network

The Didsbury esker network consists of upglacier and down-glacier esker systems separated by an ~6 km gap (Fig 2c) At thekm-scale the network has an undulating long profile with un-dulations being more pronounced in the downglacier system(Fig 5)

The upglacier esker system contains a set of branched ridgesThe trunk ridge is relatively uniform in long-profile at the hm-scaleis highly continuous (924 continuity) covers an area of 06 km2

and sits within an area mapped as lt1 m thick clast-poor till(Shetsen 1987) The upglacier esker system (trunk and tributaries)is composed of seventeen longitudinally-aligned ridge segments~015e12 km long that are classified solely as round-crested(lt100 m wide)

The downglacier esker system covers an area of 11 km2 and issingle thread highly continuous (91 continuity) and locatedwithin a broader meltwater valley that has extensive valley fill(~21 km2) It is typically flanked on both sides by kettle holes(Fig 2c) and is composed of twenty-seven longitudinally-alignedridge segments that are ~011e146 km long and classified asround crested (lt100 m wide 363 of the system area) sharpcrested (23 of the system area) and ridge enlargements (gt100 mwide 131 of the system area) Although round-crested ridgesegments are relatively uniform in long-profile at the hm-scalethey include occasional non-uniform sections (ridge heighteningby up to ~5 m within ~100 m) This system terminates in a fan(~053 km2 484 of the system area but only ~15 km of thesystem length) It has a broadly lobate planform and dips to thesoutheast parallel to the adjacent esker ridge crestline Theabsence of multiple fans along the downglacier system isconsistent with synchronous esker formation (cf Brennand2000) However the change in orientation between the upgla-cier and downglacier systems could suggest a time-transgressivecomponent to network formation with the absence of an obviousfan at the terminus of the upglacier esker possibly being linked tosediment supply constraints

34 Torrington esker

The Torrington esker covers an area of ~009 km2 and is a singlecontinuous ridge that sits within an area dominantly mapped aslt1 m thick clast-poor till (Shetsen 1987) and is aligned down- andup-glacier with meltwater valley reaches (lt250 m wide) that areincised into the till surface and do not include channel fill Theridge which is relatively uniform in long profile at the km-scalehas a slightly undulating long profile (ridge heightening by up to~5 m within ~100 m) at the hm-scale (Fig 5) and is composed of asingle ridge segment classified as a ridge enlargement (gt100 mwide 168 of the ridge area) at its head and as round-crested forthe remainder of its length (lt100 m wide 832 of the ridge area)There are no fans associated with this esker suggesting synchro-nous formation (cf Brennand 2000) The lack of a fan at the eskerterminus may be explained by either 1) the ice margin at the timeof esker formation was some distance down glacier suggesting theridgewas formed synchronous to the small meltwater channels up-and down-glacier from it or 2) there was insufficient sediment toform a fan

4 Landform architecture

GPR data are presented as fence diagrams or two-dimensionallyand overlain by the corresponding ERT profiles as available (Figs 67 9 11 13e16) All GPR and ERT profiles are presented without anyvertical exaggeration and an esker-normalized colour scale isapplied to ERT profiles All GPR data are presented two-dimensionally in the supporting information (Figs S3eS16S18e25 and S28e41) alongside animations (A) that show land-form growth at grids 1e2 and 4e6 (A 1e5) The geophysical andsedimentary data is described and interpreted (following thecriteria described in x23 and summarized in Fig 4) for each grid inturn but those grids located on ridge segments within meltwatervalleys and areas of clast-poor till are distinguished Offsets be-tween reflections occur throughout the data and are interpreted asfaults as their characteristics are consistent with those in other GPRsurveys (cf Fiore et al 2002 Burke et al 2008 2010 2012aWoodward et al 2008)

41 Esker ridge segments within meltwater valleys containingbroad valley fill

411 Strathmore esker4111 Grid 1 Grid 1 was collected on a ridge enlargement that issurrounded by meltwater valley fill (Figs 2a and S1) The inter-pretation below is based upon a dense grid of lines from whichonly a representative selection are shown in Fig 6 due to the closeline spacing All GPR lines and animations are presented inFigs S3eS10 A1 One ERT profile was collected here that imagesthree resistivity units (RU1 e 3) (Figs 6 and 7a) These unitscorrespond well to the GPR data though the deepest (RU1) is onlyidentified in the ERT profile as it is at a depth greater thanmaximum GPR penetration RU1 has a concave-up upper boundingsurface extends beyond the full width of the esker and is associ-ated with an order of magnitude change in resistivity from ~450 to10700 U-m above to ~20e280 U-m within RU1 (Fig 7a) AlthoughRU1 coincides with the water table at the start and end of the line(Fig 7) it dips below it at the line centre (beneath the esker crest)and so its low resistivity is unlikely to be associated solelywith highwater content Instead RU1 has a geophysical signature consistentwith that observed for diamicton beneath other eskers in Alberta(Figs 6 and 7a) RU2 fills the depression within RU1 and corre-sponds to the bulk of esker ridge sediments This unit has a re-sistivity of ~300e10700 U-m (Fig 7a) consistent with the coarsesand and pebble gravel identified by shallow augering (extendedinto the upper part of RU2) These coarse sediments correspond togood GPR performance throughout the grid (Fig 6) though GPR didnot penetrate the full thickness of the esker (at least at line Y5)RU3 which appears lt1 m thick within the ERT profile (Fig 7a) isnot identified within the GPR data because it is obscured by theairwave and groundwave RU3 forms a low-resistivity unit(~200e600 U-m) at the landform surface that corresponds to finesand and occasional pebbles identified by shallow augeringAlthough RU3 is consistent with either fine-grained diamicton orlake sediments its characteristics identified by augering suggest itis fine-grained diamicton

RU2 is composed of at least 51 radar elements (Figs 6 and 8 andFigs S3eS10) that have been classified based upon accretion types(Fig 4) RE-A to RE-H record earliest (as imaged by the GPR)deposition within the landform and form convex-up radar ele-ments up to ~80 m long and ~6 m thick (Fig 6) These elements arecomposed of reflections consistent with downflow (RE-A to RE-E)and vertical (RE-F to RE-H) accretion and are stacked in a down-glacier direction recording down-glacier landform growth (Stages1e8 Fig 8 A 1) Downflowe (RE-I) and verticallye (RE-J and RE-KFig 6) accreting elements are then stacked on top of these earliestdeposits (RE-A to RE-H) to record limited vertical landform growth(stages 9e11 Fig 8 A 1) prior to resumed down-glacier stacking ofdownflow e (RE-L and RE-M) and vertically e (RE-N and RE-O)accreting radar elements (stages 12e14 Fig 8 A 1) Reflectionswithin RE-O are truncated by the lower bounding surfaces of RE-Pand RE-Q (Figs 6 and S3) suggesting this element was moreextensive in an up-glacier direction Further down-glacier landformgrowth on the lee of RE-N is recorded by down-glacier stackedvertically e (RE-P) and down-glacier accreting (RE-Q) elements(Fig 6 stages 15e16 Fig 8 A 1) The final stages of esker depositionare recorded by headward (RE-R S and X) vertical (RE-T Y Z) anddown-flow (RE-U to RE-W) accretion of elements that are stackedin an up-glacier direction (Fig 6) indicating up-glacier landformgrowth (stages 17e24 Fig 8 animation 1) RE-AA to RE-AV areconsistent with slumps (Fig 4) that along with faults here recordcollapse of the esker flanks following removal of ice support

Supplementary video related to grid 1 can be found at httpdxdoiorg101016jquascirev201502017

Fig 6 Fence diagram of GPR and ERT profiles collected at grid 1 (Strathmore esker system) Given the close line spacing not all GPR lines are included (see Figs S3 S5 S7 S9 insupporting information for 2D representations of all GPR lines and radar elements) The data have been processed following the protocol detailed in x 22 Radar element boundingsurfaces are indicated by bold lines and radar elements have been labelled in approximate order of deposition (A e AV) Offsets in reflection (faults) are indicated by dashed greenlines (these are white where overlain on the ERT profile) and the black dashed line indicates a continuous horizontal reflection that does not correspond to primary beddingstructure (water table) (For interpretation of the references to colour in this figure legend the reader is referred to the web version of this article)

4112 Grid 2 Grid 2 was collected at a round-crested ridge~18 km down-glacier of a ridge enlargement (Figs 2a and S1)Although broad meltwater valley fill is mapped within this area(Fig 2a) and appears in water well records (eg Alberta Govern-ments wells 0218158 0218370) much of this appears to have beeneroded and removed from the area immediately surrounding the

esker (probably through a combination of cannibalization duringesker formation and subsequent Holocene fluvial activity) Two ERTprofiles (Y1 and Y4) were collected here that reveal three resistivityunits (RU1-3 Fig 7b c) The lowest RU1 has a resistivity of~20e300 U-m in both profiles which is an order of magnitudechange from ~300 to 6500U-m above Where this unit is within the

Fig 7 Two-dimensional ERT profiles overlain on the corresponding GPR lines collected at a) line Y5 of Strathmore esker grid 1 b) line Y1 of Strathmore esker grid 2 c) line Y4 ofStrathmore esker grid 2 d) line Y8 of Clear Lake esker grid 4 e) line Y1 of Didsbury esker grid 6 Data were processed following the protocol outlined in x 22 Resistivity units arelabelled RU1 to RU5 and delimited by bold dashed lines Radar bounding surfaces are indicated by thin solid lines and labelled in approximate order of deposition (all REs for eachgrid are shown in Figs S3eS16 S18eS41 in supporting information) Note that with the exception of RE-BV and RE-BW (d) RU3 and RU4 do not correspond to radar elementsbecause they are thin and located at the landform surface where they are obscured in the GPR data by the airwave and groundwave The water table elevations are indicated by bluearrows on the y-axes when derived from the elevation of nearby water bodies and by a white dashed line labelled WT where imaged in the GPR data The red arrows along the topof the profiles show the position (along profile) of shallow augering (For interpretation of the references to colour in this figure legend the reader is referred to the web version ofthis article)

penetration depth of the GPR it is associated with rapid signalattenuation and ringing (Fig 7b) Although this unit is at a similarelevation to the water table on line Y4 (Fig 7b) it does not directlycoincide with it on either ERT profile Rather the presence of higherresistivity material above RU1 but below the water table (eg at

~80e140 m Fig 7b) suggests water content alone cannot accountfor RU1 Instead the geophysical characteristics of RU1 are consis-tent with diamicton (Fig 4) confirmed in a shallow pit nearby OnY1 RU1 has a concave-up profile (Fig 7b) whereas on Y4 it isirregular and has characteristics (Fig 7c) consistent with slight

Fig 8 Fence diagram of interpreted GPR lines collected at grid 1 (Strathmore esker) Given the close line spacing not all GPR lines are included (see Figs S4 S6 S8 S10 insupporting information for 2D representations of all GPR lines) Radar elements have been colour-coded based upon the broad style of deposition within each radar elementfollowing the protocol detailed in x 23 and Fig 4 These colours correspond to those in the legend (bottom right) The radar elements have been numbered in approximate order ofdeposition (1e25) The white dashed lines mark faults (reflection offsets) and the black dashed line corresponds to the water table (For interpretation of the references to colour inthis figure legend the reader is referred to the web version of this article)

deformation toward the esker crestline RU2 which forms the bulkof the esker ridge has a resistivity of ~300e6500 U-m which isconsistent with coarse sand and gravel RU3 which appears lt15 mthick within the ERT profiles is not identified within the GPR databecause it is mostly obscured by the air- and ground-waves RU3 is

a low-resistivity unit (~100e600 U-m) that discontinuously capsthe landform (Fig 7b c) The characteristics of RU3 are consistentwith lake sediments or fine-grained diamicton (Fig 4) and becauseglaciolacustrine deposits are mapped here (Fig 2a) by Shetsen(1987) RU3 likely records lake sediments

Fig 9 Fence diagram of GPR and ERT profiles collected at grid 2 (Strathmore esker system) The data have been processed following the protocol detailed in x 22 Radar elementbounding surfaces are indicated by bold lines and radar elements have been labelled in approximate order of deposition (A e AY) Offsets in reflection (faults) are indicated by thedashed green lines (these are white where overlain on the ERT profiles) The black dashed line indicates the maximum depth of GPR penetration (For interpretation of the ref-erences to colour in this figure legend the reader is referred to the web version of this article)

GPR data reveal that the Strathmore esker (RU2) at grid 2 iscomposed of at least 52 radar elements (Figs 9 and S11e16)though GPR penetration does not appear to cover the full eskerthickness (~20 m as delimited by the high resistivity values(~300e6500 U-m) in the ERT profiles Fig 7b c) Earliest

deposition as imaged by the GPR is recorded by RE-A to RE-Dwhich form ~75e135 m long and ~30e50 m wide convex-up el-ements that are disconnected by zones of erosion (truncation ofinternal reflections suggest at least some erosion) or non-deposition (Fig 9) Elements RE-A to RE-D indicate vertical

Fig 10 Fence diagram of interpreted GPR lines collected at grid 2 (Strathmore esker system) Radar elements have been colour-coded based upon the broad style of depositionwithin each radar element following the protocol detailed in x 23 and Fig 4 These colours correspond to those in the legend (right) The radar elements have been numbered inapproximate order of deposition (1e18) The white dashed lines mark faults (reflection offsets) (For interpretation of the references to colour in this figure legend the reader isreferred to the web version of this article)

accretion of both vertically- and laterally-stacked elementsrecording initially disconnected vertical and lateral landformgrowth within depocentres (stages 1e2 Fig 10 A 2) This initialdeposition is followed by very low angle (lt5) up-glacier andlateral stacking of vertically-accreting radar elements (RE-E and

RE-F Fig 9) that record up-glacier and lateral landform growthonto the stoss side of initial depocentres (stage 3 Fig 10 A 2)Although RE-G to RE-L (Fig 9) are generally stacked in a down-glacier direction they are dominated by vertical accretion (RE-Kand RE-L are downflow accreting) and these elements are more

Fig 11 Fence diagram of GPR and ERT profiles collected at grid 4 (Clear Lake esker system) The data have been processed following the protocol detailed in x 22 Radar elementbounding surfaces are indicated by bold lines and radar elements have been labelled in approximate order of deposition (A e CA) Offsets in reflection (faults) are indicated by thedashed green lines (these are white where overlain on the ERT profile) (For interpretation of the references to colour in this figure legend the reader is referred to the web versionof this article)

Fig 12 Fence diagram of interpreted GPR lines collected at grid 4 (Clear Lake esker) Radar elements have been colour-coded based upon the broad style of deposition within eachradar element following the protocol detailed in x 23 and Fig 4 These colours correspond to those in the legend (top right) The radar elements have been numbered inapproximate order of deposition (1e38) The white dashed lines mark faults (reflection offsets) (For interpretation of the references to colour in this figure legend the reader isreferred to the web version of this article)

Fig 13 Fence diagram of GPR and ERT profiles collected at grid 6 (Didsbury esker system) The data have been processed following the protocol detailed in x 22 Radar elementbounding surfaces are indicated by bold lines and radar elements have been labelled in approximate order of deposition (A e BF) Offsets in reflection (faults) are indicated by thedashed green lines (these are white where overlain on the ERT profile) (For interpretation of the references to colour in this figure legend the reader is referred to the web versionof this article)

Fig 14 Fence diagram of interpreted GPR lines collected at grid 6 (Didsbury esker system) Radar elements have been colour-coded based upon the broad style of deposition withineach radar element following the protocol detailed in x 23 and Fig 4 These colours correspond to those in the legend (bottom right) The radar elements have been numbered inapproximate order of deposition (1e20) The dashed lines mark faults (reflection offsets) (For interpretation of the references to colour in this figure legend the reader is referred tothe web version of this article)

Fig 15 a) Photograph of an exposure through the upglacier Didsbury esker ridge The exposure which is perpendicular to ridge crestline is located ~250 m upglacier from grid 7Three stratigraphic units are identified 1) a core of deformed diamicton with diapiric structures rising toward the esker crest 2) waterlain sand and gravel with no obvious primarystructure 3) fine-grained diamictonwith occasional outsized clasts Two dimensional ERT profiles collected at b) Didsbury esker ridge grid 7 (note the close correspondence to unitsin exposure (a)) and c) line Y5 and d) line Y2 of the Torrington esker ridge (grid 9) Resistivity units are labelled RU1 e RU3 and are delimited by the dashed lines The red arrowalong top of panel b shows the position (along profile) of shallow augering The water table elevations (derived from the elevation of nearby water bodies) are indicated on b-d byblue and white arrows along the y-axes The ERT profiles from the Torrington esker ridge use a uniform resistivity scale Close up photographs of e) unit 3dfine-grained diamictonf) unit 2dsand and gravel and g) unit 1ddeformed diamicton Knife handle is 10 cm long Pick is 70 cm long (For interpretation of the references to colour in this figure legend thereader is referred to the web version of this article)

suggestive of vertical landform growth (given their near horizontaland tabular element geometry) atop and down-glacier from earlierdeposits (stages 4e7 Fig 10 A 2) Deposition of RE-G to RE-Lconnects previously disconnected depocentres and is followed bythe final stages of esker deposition recorded by RE-M to RE-Z(Fig 9) RE-M to RE-T are tabular radar elements ~70e390 mlong and ~15e50 m wide that indicate disconnected verticalstacking of vertically accreting radar elements (stages 8e12Fig 10 A 2) RE-U to RE-Z are dominated by vertical accretion andalthough stacked in a down-glacier direction this stacking is verylow angle (Fig 9) suggesting general vertical landform growthdown-glacier of earlier deposits (stages 13e17 Fig 10 A 2)

Radar elements AA to AY (Fig 9) are consistent with slumps(Fig 4) that record collapse of the esker flanks following removalof ice support (stage 18 Fig 10 A 2) Faults are abundantregardless of line orientation suggesting some faulting wasgenerated by processes additional to those associated with icesupport removal

412 Clear Lake eskerTwo grids of data were collected at the Clear Lake esker both of

which were located at round-crested ridge segments (Figs 2b and

Fig 16 Two-dimensional GPR profiles of selected data from grid 8 (upglacier Didsbury esker ridge) The data have been processed following the protocol detailed in x 22 Radarelement bounding surfaces are indicated by bold lines and the dashed line corresponds to a major attenuator interpreted as a diamicton surface Green dashed lines correspond toreflection offsets (faults)

S1) These ridge segments are located within a broad meltwatervalley that includes valley fill Grid 4 is ~2 km down-glacier of aridge enlargement and grid 5 is ~5 km down-glacier of grid 4(Figs 2b and S1) Because grid 5 has a signature that is very similarto Strathmore esker grid 2 (widespread vertical landform growthand a high frequency of faulting see Figs S26eS37 A 4) only grid 4data is described and interpreted here (Figs 7d 11 and 12)

The one ERT profile collected at grid 4 can be divided into threeresistivity units (RU1e3 Fig 7d) RU1 has a resistivity of~60e500 U-m and represents an order of magnitude change from~500 to 8900 U-m above (Fig 7d) RU1 is associated with a high-amplitude and semi-continuous (throughout the grid) reflectionin the GPR data below which the GPR signal rapidly attenuates(Figs 7d and 11) Although this reflection is in close proximity to thewater table RU1 rises above the water table in places and so isunlikely to reflect high water content alone but instead is moreconsistent with slightly deformed diamicton or bedrock (Fig 4)near the esker crest (eg at ~85 m on line Y8 Fig 7d) RU2 has ahigh resistivity (~500e8900 U-m) with the exception of ananomalously low resistivity (~150e208 U-m) at ~105e120 m(Fig 7d) and intermediate resistivities (~200e500 U-m) at30e45 m 55e70 m and 75e90 m The low-resistivity area corre-sponds to an area of significant data removal so is likely due to poorinversion rather than the presence of low resistivity materialInstead RU2 has resistivity values largely consistent with sand andgravel though the intermediate resistivity zones may record in-jection of fines during a subsequent episode of substrate defor-mation RU3 is a low resistivity unit (~100e600 U-m) up to ~8 m

thick on the landform flanks that is either fine-grained diamicton orlake sediments

RU2 and RU3 which rest directly atop the substrate (RU1) arecomposed of 79 radar elements (Figs 11 and S18eS25) Thedeepest radar elements within RU2 are RE-A to RE-C (Fig 7d) Theyform broad and tabular elements which typically fill depressions(eg at ~200e450 m on line X1 Fig 11) within the irregular sub-strate andor extend beyond the esker flanks (Fig 11) Thesecharacteristics are consistent with a broad meltwater valley fill(Figs 4 and 12 (stage 1)) that we map around the esker (Figs 2band S1) Because these units have the same resistivity as thosemaking up the esker they are not identified as a separate resistivityunit (Fig 7d)

The remaining radar elements within RU2 (RE-D to RE-BM)make up the esker These elements although dominated by verti-cal accretion show alternating episodes of up-glacier and down-glacier landform growth (Figs 11 and 12) Earliest deposition isrecorded by RE-D to RE-I which form broad tabular radar elementsthat indicate vertical accretion (Fig 11) of vertically-stacked radarelements (some minor downflow growth) separated by zones ofnon-deposition or erosion (stages 2e3 Fig 12 A 3) This initialdeposition is followed by up-glacier stacking of RE-J to RE-S thatmainly accreted vertically (RE-J RE-K RE-M to RE-R) with limitedheadward accretion (RE-L and RE-S) Together radar elements RE-Jto RE-S record up-glacier landform growth at the up-glacier end ofthe grid (stages 4e11 Fig 12 A 3) RE-T to RE-AD are vertically-accreting elements that record widespread vertical landformgrowth atop earlier deposits (stages 12e14 Fig 12 A 3) This is

followed at the up-glacier end of the grid by down-glacier stackingof tabular and sub-horizontal elements composed of reflectionsconsistent with down-flow (RE-AE) and vertical (RE-AF and RE-AG)accretion (stages 15e17 Fig 12) Subsequent up-glacier landformgrowth begins at the down-glacier end and progresses up-glacier(stages 18e24 Fig 12 animation 3) and although dominated byup-glacier stacking with some headward accretion (RE-AN and RE-AS) most accretion is vertical (RE-AH to RE-AM and RE-AO to RE-AR) and these vertically-accreting elements do not dip in anupglacier direction (Fig 11) Subsequent vertical landform growthof vertically-accreting elements at the up-glacier end of the grid(RE-AT to RE-AZ stages 25e28 Fig12) is followed by down-glacierstacking of downflow e (RE-BA) vertically e (RE-BB and RE-BD toRE-BE) and laterally e (RE-BC) accreting elements (stages 29e32Fig 12 A 3) The final stages of esker deposition are focussed at thedown-glacier end of the grid where they are recorded by RE-BF toRE-BM (Fig 11) These elements which include vertical (RE-BF toRE-BJ and RE-BM) down-flow (RE-BL) and lateral accretion (RE-BK) record vertical landform growth (stages 33e37 Fig 12 A 3)Faults are abundant regardless of line orientation and becauseslumps are absent this suggests faulting was not associated withice support removal

No slumping was identified within the grid (Figs 11 and 12)which is likely due to the landform being relatively broad and low-lying here with gentle flank slopes (~3e6) Instead the strati-graphically youngest radar elements are RE-BN to RE-CA (Fig 11)These elements are consistent with RU3 on Y8 (Fig 7d) Becauseglaciolacustrine sediment is mapped adjacent to the grid (Shetsen1987) and GPR could not penetrate a significant thickness of fine-grained diamicton RE-BN to RE-CA and RU3 likely record sandyglaciolacustrine sedimentation following esker deposition Thepresence of lake sediments could support time-transgressivedeposition of fans atop a synchronously-formed esker duringsubsequent ice retreat (cf Brennand and Shaw 1996 Brennand2000)

Supplementary video related to the Clear Lake Esker can befound at httpdxdoiorg101016jquascirev201502017

413 Downglacier Didsbury esker (Grid 6)Grid 9 is at a ~14 km long ridge segment located within a

meltwater valley It is composed of alternating round-crestedridge segments and ridge enlargements and is surrounded bybroad meltwater valley fill (Figs 2c and S2) Geophysical surveyscovered both ridge morphologies and a gravel pit was locatedimmediately upflow of the grid (Fig S2) One ERT profilecollected on the up-glacier ridge enlargement records four re-sistivity units (RU1 e 4 in Fig 7e) that correspond well to the GPRdata in which 57 radar elements are identified (labelled AeZ andAAeBF in Fig 13) RU1 is the stratigraphically oldest unit and iscomposed of low resistivity (~11e200 U-m) material AlthoughRU1 is largely below water table its upper boundary rises ~5 mabove it and so it is unlikely to reflect a high water content alone(Fig 7e) Because the upper boundary of RU1 corresponds to anorder of magnitude change in resistivity (from ~120 to 6400 U-mabove) and forms a high-amplitude reflection in the GPR databelow which the GPR signal rapidly attenuates it is consistentwith diamicton or bedrock (Fig 4) RU2 comprises the bulk of theesker ridge and has a resistivity of ~120e6400 U-m (Fig 7e)consistent with the sand to cobble gravel (~015 m diameter) andrare boulders (b-axis up to 09 m) observed within a gravel pitexcavated into the esker ~10 m from the up-glacier end of the grid(Fig S2) RU2 is capped by RU3 and RU4 which form thin units atthe ground surface that are typically obscured by the airwave andgroundwave and so are not recorded in the GPR lines RU3 forms alow resistivity unit (~250e600 U-m) up to 3 m thick that caps the

landform and surrounding ground surface Its characteristics areconsistent with fine-grained diamicton or lake sediments (Fig 4)Given lake sediments are mapped to the west and east by Shetsen(1987) we suggest RU3 is most likely lake sediment depositedafter esker deposition RU4 is the stratigraphically youngest unitand has a resistivity of ~30e60 U-m (Fig 7e) Its lower boundingsurface is concave-up though the unit seems to extend beyondthe ERT profile The location of this unit corresponds to that of akettle hole flanking the esker ridge (Fig S2) therefore RU4 isinterpreted as recording a kettle hole infill (possibly wind-blownsediments)

Radar elements A to O are the stratigraphically lowest radarelements within RU2 and form broad tabular elements that filllows in the underlying topography (RU1) These extend beyondthe esker flanks and beneath the full width of the esker (eg lineY2 Fig 13) unless RU1 rises above the maximum elevation(~932 m asl) of these elements (eg line Y1 Fig 13) RE-A to RE-Olikely correspond to the meltwater valley fill that is mapped here(Figs 2c and S2) and their high-resistivity values (even whenbelow water table Fig 7e) suggest the presence of coarse sand andgravel which is consistent with the up to 12 m of sand and gravel(within the meltwater valley) shown in a water well recordslt4 km from the grid (Alberta Government well 0298270 10654540154518)

Esker sedimentation is recorded by RE-P to RE-AN RE-P to RE-Srecord initial deposition of sand and gravel within the esker-forming ice tunnel and are ridge-wide elements up to ~270 mlong They record vertical accretion of gravel sheets (Fig 14) and arestacked in a down-glacier direction indicating down-glacier land-form growth (stages 2e4 Fig14 A 5) Subsequent vertical landformgrowth (stages 5e6 Fig 14) is recorded by RE-T to RE-X (Fig 13)which are ~40e90 m long and indicate vertical accretion AlthoughRE-T to RE-X are ridge-wide in places the lower bounding surfacesof shallower elements truncate their internal reflections (eg RE-Uand RE-V on line Y1 Fig 13) suggesting they were more extensiveprior to subsequent esker deposition RE-Y to RE-AA exhibit up-glacier stacking (Fig 13) and are composed of steeply-dipping (upto 15 in an up-glacier direction) reflections that indicate headwardaccretion and discontinuous up-glacier landform growth (stages7e9 Fig 14 A 5) of the down-glacier ridge enlargement (Figs 2cand S2) The gap between RE-Y and RE-AA is filled with the down-flow dipping (downflow accretion) and sub-horizontal (verticalaccretion) reflections of RE-AB to RE-AC and RE-D respectivelyThese radar elements record down-glacier landform growth on thelee-side of RE-Y (stages 10e11 Fig 14) Subsequent esker deposi-tion recorded by RE-AE to RE-AN is more uniform and these ele-ments are dominated by vertical accretion (except RE-AM and RE-AN which record downflow accretion) of extensive (~30e160 mlong) vertically-stacked radar elements (stages 12e19 Fig 14 ani-mation 5)

Radar elements AO to BF have resistivity values similar to thosewithin the esker ridge (the low resistivity component that forms acapping unit (RU3) is not imaged in the GPR because of the airwaveand groundwave) and so were likely sourced from material withinthe esker These elements are consistent with slumps (Fig 4) thatrecord collapse of the esker flanks following removal of ice support(stage 20 Fig 14 A 5) Such slumping and collapse probably alsoaccounts of the sharp-crested ridge segment downflow of grid 6which is bordered by a kettle hole on one flank and incised by acontemporary river on the other Faults are abundant regardless ofline orientation suggesting that some faults were generated byprocesses other than those associated with removal of lateral icesupport

42 Esker ridge segments in areas dominated by clast-poor till

421 Upglacier Didsbury esker4211 Grid 7 Grid 7 was collected on a round-crested ridgesegment (Figs 2c and S2) that is located in an area dominated byclast-poor till (Shetsen 1987) GPR and ERT profiles were located~250e300 m down-glacier of four closely-spaced (within ~50 m)exposures cut transverse to ridge crestline (Fig 15a) at a trench anda road cut One ERT profile was collected at grid 7 (Fig 15b) thatrecords three resistivity units (RU1 e 3) that correspond well tounits (lithofacies) 1e3 observed in all exposures (Fig 15a b) and tothe GPR data GPR performance was poor throughout the grid andso only one representative GPR line (Y7) is presented (Fig 15b) GPRpenetration was ~1e45 m with maximum penetration along theesker flanks and minimum penetration along the esker crest(Fig 15b) A high-amplitude convex-up reflection that rises to thelandform surface at the esker crest is identified throughout the gridbelow which the GPR signal rapidly attenuates (independent ofsystem electronics) This reflection corresponds to the upperboundary of RU1 which is defined by an order of magnitude changein resistivity values from ~2 to 60 U-m within RU1 to ~60e1660 U-m above it (Fig 15b) In the exposures RU1 corresponds to dia-micton (unit 1) that includes vertical foliation and diapiric struc-tures orientated towards the esker crest (Fig 15a b g) RU1 isinterpreted as a diamicton that has characteristics similar toregional tills and was deformed toward the esker crest RU2 is~1e3 m thick has resistivity values of ~200e1660 U-m that areconsistent with the sand and pebbles observed by shallow auger-ing RU2 correlates with unit 2 in the exposures (Fig 15a b) whichis composed of coarse sand to pebble gravel with no obvious pri-mary structure (Fig 15a f) and was deposited (waterlain) withinthe esker-forming ice tunnel prior to Unit 1 deformation RU3 has aresistivity of ~60e200 U-m and corresponds to the massive finesand with occasional cobbles (fine-grained diamicton) that makesup unit 3 (Fig 15a b e)

4212 Grid 8 Grid 8 is located at a round-crested ridge segment inan area mapped as clast-poor till (Shetsen 1987) Although theridge is within a broader meltwater valley no valley fill is mappedin the area immediately around the esker ridge and the landformcharacteristics detailed below (Figs 2c and S2) suggest a geomor-phic setting more similar to grids 7 and 9 than grids 1e6 Due tosite accessibility no ERT data were collected here but GPR datashow a similar signature to grid 7 and so only representative GPRlines are presented because of the generally poor GPR performance(Fig 16) Like at grid 7 a high-amplitude continuous convex-upreflection which is identified throughout the grid (upper bound-ing surface of RE-A) has characteristics consistent with diamictonor bedrock (Fig 4) Given RE-A is convex-up under the esker crest(Fig 16 line Y7) we interpret it as deformed probably subsequentto deposition of RE-B to RE-J RE-A is capped by nine ~1e5 m thickradar elements (B to J in Fig16) that are composed of discontinuoussub-horizontal reflections A high reflection discontinuity likelyreflects disturbed bedding within a capping unit of waterlain sandand gravel (equivalent to unit 2 at grid 7) and further supportssubsequent substrate deformation Faults are abundant regardlessof line orientation suggesting some faults were generated by pro-cesses other than those associated with removal of lateral icesupport

422 Torrington esker (Grid 9)The Torrington esker is composed of one ridge segment (Figs 2c

and S2) within an area mapped as clast-poor till (Shetsen 1987)The esker is composed of a ridge enlargement and a round-crestedridge section Two ERT profiles were collected atop the ridge

enlargement (Y2) and round-crested section (Y5) and record threeresistivity units (RU1 e 3) that correspond well to the GPR data(Figs15c d) GPR performancewas poor throughout the grid and soonly GPR lines collected along the ERT profiles (Y2 and Y5) arepresented (Fig 15c d) On line Y2 GPR penetration (lt7 m) wasgreatest on the south-eastern flank of the ridge with minimumpenetration along the esker crest and onto the north-western flank(Fig 15d) On line Y5 GPR penetration was lower than at Y2 butgreatest (lt5m) off the esker crest and onto the western flank of thelandform with minimum penetration along the top of the easternflank (Fig 15c) A high-amplitude but discontinuous and irregularreflection can be identified throughout the grid below which theGPR signal rapidly attenuates and significant noise (ringing) ispresent This reflection roughly corresponds to the upper boundaryof RU1 which is defined by an order of magnitude change in re-sistivity from ~70 to 3685 U-m above to ~11e250 U-m within RU1(Fig 15c) RU1 has an irregular upper boundary and different re-sistivity zones that dip away from the esker crest (Y5) it appears tobe injected into RU2 at ~32e55 m on Y5 (Fig 15c) These charac-teristics give the impression of structures orientated toward thecrest and its characteristics are consistent with the presence ofdeformed diamicton or bedrock (Fig 4) similar to grid 7 (Fig 15ag) RU2 is ~25e5 m thick and has a resistivity of ~400e3685 U-mon Y2 (Fig 15d) and ~701450 U-m on Y5 (Fig 15c) suggesting it iscomposed of sand and gravel but is more sandy within the round-crested section (downflow fining) In the GPR data RU2 includesdiscontinuous reflections that drape the lower bounding surface(with GPR penetration greatest where RU2 is thickest) and so RU2likely records waterlain sediment deposited prior to deformationas suggested by the diamicton or bedrock surrounding higher re-sistivity material (eg at ~32e55m on Y2 and ~50m on Y5 Fig 15cd) The esker is capped by RU3 which is a low resistivity unit(~3e200 U-m) with the characteristics of fine-grained diamicton orlake sediments (Fig 4) Given that no lake sediments are mapped inthis area (Shetsen 1987) we suggest that RU3 is likely fine-graineddiamicton similar in composition to that at the Didsbury esker (x421 Fig 15a e)

5 Sediment landform associations

Geophysical data from eskers in south-western Alberta showmajor morphological and architectural variation between themThe Strathmore Clear Lake and downglacier Didsbury esker sys-tems which sit within meltwater valleys containing broad valleyfills are composed of coarse sand and gravel that morphologicallyalternate between long round-crested ridge sections and rela-tively short ridge enlargements (Fig 2aec) or heightenings (non-uniform round-crested sections) In the Strathmore and ClearLake esker systems uniform round-crested ridge segments areassociated with extensive tabular units that are dominated byvertical accretion (grids 2 4 and 5 Fig 17a) and typically indicaterelatively uniform vertical landform growth over several kilo-metres which is not consistent with dynamic deposition Incontrast ridge enlargements are associated with alternating unitsof headward and downflow accretion (eg grid 1 Fig 17a) thatindicate episodes of down- and up-flow landform growth Non-uniform round-crested ridge segments in the downglacier Dids-bury esker system alternate between vertical accretion andlandform growth where the ridge height is lower (eg~100e450 m and ~575e650 m Figs 13 and 14) and headward ordownflow accretion and landform growth at heightenings (eg~450e575 m Figs 13 and 14 grid 6 Fig 17a) The dominance ofheadward and downflow accretion (and stacking of elements)within and spatially associated with ridge enlargements andwithin heightenings in round-crested sections is consistent with

Fig 17 Histograms showing the occurrence () of accretion types (individual radarelements were interpreted following the criteria outlined in Fig 4 and detailed in x 23)for a) each grid grouped by esker and b) a comparison (all GPR grids) for each eskerin Alberta with the Chasm esker British Columbia (Burke et al 2012a) and GLOF eskersat Skeiethararjeurookull Iceland and Bering Glacier Alaska (Burke et al 2008 2010)

the presence of macroforms These suggest a dynamic depositionthat is often associated with high-magnitude flow due to inter-action between ice tunnel geometry flow dynamics and sedi-mentation (Brennand 1994 2000 Brennand and Shaw 1996Burke et al 2008 2010 2012a) In contrast the Torrington andupglacier Didsbury esker ridges sit within areas dominated byclast-poor till These esker ridges which are dominantly round-crested and uniform in width and height are composed ofdeformed diamicton or bedrock that is capped by a thin layer ofwaterlain fine sand and pebble gravel (grids 7 8 9 Fig 17a)Although the head of the Torrington esker (grid 9) is classified asa ridge enlargement there is no evidence of a ridge-scale mac-roform here or significant landform growth prior to the phase ofdiamicton or bedrock deformation Instead the ridge enlargementprobably reflects a local supply of sand and gravel (perhaps fromup-flow meltwater valley erosion Fig 2) that was quicklyexhausted within a locally enlarged ice tunnel This starveddown-glacier flows of significant sediment as supported by theapparent down-ridge lowering of resistivity values

Faulting is abundant in all eskers and when this is restricted toflow-perpendicular orientations (grid 1) it can be largely attributedto removal of lateral ice support However at most grids faults arefrequent regardless of line orientation (grids 2 4e8) suggesting anadditional mechanism Although this is often thought to be asso-ciatedwith englacial deposition (eg Price1966) the association ofthe eskers with both subglacially-deformed diamicton andor ameltwater valley as well as limited distortion of primary reflectiongeometry is not consistent with englacial deposition Rather thefaults are consistent with syn-depositional faulting of esker sedi-ment and disturbance during episodes of substrate deformationThis suggests the accommodation space generated by ice tunnelgrowth that allowed subsequent landform growth was not filledrapidly enough to prevent collapse of earlier ice tunnel infill

6 Comparison with other eskers

In south-western Alberta there appears to be a link betweenmorphology landform architecture and substrate The associationof ridge enlargements with ridge-scale macroforms has been re-ported for many eskers (eg Brennand 1994 2000 Fiore et al2002 Burke et al 2008 2010 2012a) Observations from bothcontemporary glaciers (Burke et al 2008 2010) and former icesheet beds (eg Brennand 1994 Fiore et al 2002) suggest thatridge-scale macroforms are the product of dynamic depositionalenvironments typically associated with GLOFs During such eventsthere appears to be a direct link between conduit growth anddeposition (cf Brennand 1994 2000 Burke et al 2008 20102012a) with rapid changes in accommodation space being re-flected in esker lithofacies and sedimentary architecture Increasesin accommodation space along an ice tunnel result in sedimenta-tion and the development of a depocentre (bar-like obstacle) Thissedimentation decreases accommodation space enhancing localflow velocity and resulting in ice tunnel and landform growtharound the obstacle (Brennand 1994 Brennand and Shaw 1996Burke et al 2012a) Although similar dynamism is recordedwithin ridge enlargements and heightenings of the largest eskers insouth-western Alberta the ldquoclassicrdquo backset-foreset coupling ofcomposite macroforms (Brennand 1994 Brennand and Shaw1996) is less well developed than for GLOF eskers elsewhere(Burke et al 2008 2010) Furthermore uniform round-crestedridge segments which are typically connected to ridge enlarge-ments (no gap between them) and thus must have formed syn-chronously are lacking such dynamism In southern Ontario thepresence of pseudo-anticlinal macroforms was suggested byBrennand (1994) to be the product of a uniform ice tunnel geom-etry and high flow power However round-crested ridge segmentsin south-western Alberta are either composed of extensivevertically-accreted and flat lying radar elements that are testamentto uniform vertical landform (and ice tunnel) growth or are filledwith deformed diamicton or bedrock Neither is consistent withrapid deposition that ultimately forced ice tunnel evolution (cfBurke et al 2012a) The lack of pseudo-anticlinal macroformssuggests the conditions for their formation were not present likelyreflecting a low sediment supply rather than low flow power Suchdifferences in esker architecture are made clear by comparingoverall occurrence of accretion types between south-westernAlbertan eskers the Chasm esker in British Columbia (a GLOFesker) and GLOF eskers at Skeiethararjeurookull Iceland and BeringGlacier Alaska (Fig 17b) The lack of well-developed compositemacroforms more uniform architectures within uniform round-crested ridge segments and the presence of deformed diamictonor bedrock in south-western Albertan eskers is likely to be aproduct of different flow conditions conduit geometries andorsediment supply Subtle differences in composite macroform

architecture between the GLOF eskers at Skeiethararjeurookull and theBering Glacier were attributed to differences in sediment supplyand flow conditions between the esker forming events (Burke et al2010) and this was also invoked for formation of the Chasm esker inBritish Columbia (Burke et al 2012a) The most complex eskersedimentary architectures are associated with high flood powersand a large supply of sand and gravel The eskers at SkeiethararjeurookullBering Glacier and within meltwater corridors on the FraserPlateau (British Columbia) were all deposited in locations with aready supply of sand and gravel In south-western Alberta tills aretypically clast-poor (Pawluk and Bayrock 1969) meaning the sup-ply of sand and gravel required for esker building is low

7 The role of sediment supply in ice tunnel evolution andesker architecture

We suggest that variations in esker sedimentary architecture insouth-western Alberta and likely elsewhere is a direct conse-quence of changes in sediment supply because 1) there is a limitedsupply of esker building material in south-western Alberta whereeskers are relatively rare and smaller than in other regions 2) insouth-western Alberta the most well-developed landforms arelocated directly atop or in association with broad sand and graveldeposits within meltwater valleys (Strathmore Clear Lake and thedownglacier Didsbury esker systems) whereas the smaller eskerridge segments are in areas dominated by clast-poor till (upglacierDidsbury and Torrington esker ridges) and 3) architecturallycomplex eskers are typically found in regions with a significant

Fig 18 Evolutionary model for esker and ice tunnel development under three scenarios a) hand c) truly sediment supply-limited The blue arrows indicate tunnel growth which includetunnel For full explanation of this figure refer to x 7 (For interpretation of the references t

sediment supply (eg Brennand 1994 2000 Brennand and Shaw1996 Burke et al 2008 2010 2012a) The lack of significant eskerbuilding material in areas away from pre-existing sand and graveldeposits indicate most sediment was not sourced from the over-lying ice and that this icemust have been relatively clean Followingthis hypothesis we suggest three main depositional scenarios foreskers not only in south-western Alberta (Fig 18) but also eskerselsewhere under similar depositional environments

71 Scenario I e abundant sediment supply and flow power(Fig 18a)

Accommodation space is controlled by the balance between icetunnel growth and closure A high flow power tends to enlarge theice tunnel (increasing accommodation space) due to frictionalmelting and mechanical excavation of the ice and erosion ofearlier-deposits whereas a high sediment supply tends to close anice tunnel (reducing accommodation space) because the ice tunnelbecomes choked with sediment (Burke et al 2012a) During initialesker sedimentation any irregularities in ice tunnel geometry(height andor width) result in variations in accommodation spaceLarger accommodation spaces (ice-tunnel enlargements andheightenings) are rapidly filled with sediment if the sedimentationrate exceeds that of ice tunnel enlargement (stage 1 Fig 18a) Thiscreates an obstacle to flow and forces flow dynamic changes thatenhance frictional melting of the ice tunnel walls (and erosion ofearlier deposits) on the stoss side and atop of the obstacle Thislocally enhanced ice tunnel growth forces enlargement up-glacier

igh sediment supply and flow power b) limited sediment supply and high flow powers both thermo-mechanical excavation of the ice and erosion of sediment within the iceo colour in this figure legend the reader is referred to the web version of this article)

from the obstacle due to the turbulent vortices increasing ac-commodation space once more to develop a flow expansion on thestoss side of the obstacle (stage 2 Fig 18a) Backset beds (headwardaccretion) are depositedwithin the flow expansion due to high flowpower reducing the accommodation space on the stoss side of thedeveloping depocentre Where flow power is high enough thermo-mechanical expansion of the ice tunnel (ice roof meltmechanicalexcavation and sediment erosion) atop of the obstaclewhere flow isconstricted would force down-glacier ice tunnel growth near thebrink point of the obstacle generating a flow separation (stage 3Fig 18a) Subsequent deposition is then focussed on the lee side ofthe depocentre allowing progradation as well as continued head-ward accretion on the stoss side of the depocentre as the flowexpansion migrates upglacier due to ice tunnel enlargement (stage4 Fig 18a) Continued cycles of increasing and decreasing accom-modation space results in a complex esker architecture and thedevelopment of ridge scale macroforms (depocentres)

72 Scenario II e limited sediment supply and high flow power(Fig 18b)

In south-western Alberta uniform round-crested ridge seg-ments were generated by deposition that was less dynamic thanthat which generated ridge enlargementsheightenings As eskerbuilding material was sourced from broad meltwater valley fill

Fig 19 Conceptual plot showing the impact of sediment supply (amount and appropriatesubglacial ice tunnel and esker geometry and sedimentary architecture White areas delimitan esker cannot form (grey polygons) Where both water and sediment are present in sufficirelative supplies of water (flow power) and sediment which dictate the sedimentation ratometry are associated with high flow power and abundant sediment supply (scenario I redunder more moderate sediment supply and flow power However where sediment supply isubstrate is deformable it will be primarily composed of this deformed substrate (scenario Iexistence field is likely more tabular (ie has thickness) (For interpretation of the references

therewould only be a finite amount of material available to the flowunless the ice tunnel enlargedmigrated laterally to exploit un-tapped deposits Therefore development of depocentres (ridge-scale macroforms) along an ice tunnel would significantly reducesediment load downflow and it would not be replenished untilfurther sediment was made available to the flow due to excavationof underlying broad valley fill (if available) or reworking of upflowesker deposits In south-western Alberta such a low sedimentsupply particularly downflow of depocentres resulted in the rateof ice tunnel growth exceeding that of infilling and the ice tunnelwas able to evolve independent of sedimentation (stage 1 Fig 18b)This resulted in the development of a relatively uniform ice tunnel(for at least several hundred metres) that was filled by flat-lyingand largely vertically-accreting beds that became faulted syn- aswell as post-depositionally (stages 2e4 Fig 18b)

73 Scenario III e truly sediment supply-limited (Fig 18c)

The Torrington esker and upglacier part of the Didsbury eskerwere deposited within very low sediment supply environmentsRegardless of the flow conditions this resulted in the rate of icetunnel enlargement significantly exceeding the rate of sedimentinfilling and like for scenario II the ice tunnel was able to evolveindependent of sedimentation (stage 1 Fig 18c) Any limited sandand gravel within the flow may be deposited as discharge falls but

grain size for esker building) flow power (water supply) and sedimentation rate onunlikely conditions Where a supply of sediment andor water is absent or insufficientent quantities then esker morphology and sedimentary architecture is dependent upone The most architecturally complex eskers and those with the most non-uniform ge-zone) Eskers with uniform geometry and simple architecture (scenario II yellow) forms low or absent then regardless of flow power an esker either does not form or if theII green zone) Although depicted as a simple plane in this conceptual graph the eskerto colour in this figure legend the reader is referred to the web version of this article)

this does not fill the ice tunnel (stage 2 Fig 18c) and during thewaning stage(s) and immediately following ice tunnel shut down itwould be at a relatively low pressure compared to that directlybeneath the surrounding ice Given the ice here was resting atop afine-grained and probably saturated diamicton (Evans et al 2008)or bedrock this pressurized diamicton would have been squeezedinto the low pressure ice tunnel deforming any pre-existing sandand gravel and resulting in an esker with a deformed diamictoncore roughly conformable to the ridge (stage 3 Fig 18c) (Stalker1960 Tylmann et al 2013) If there is no deformable sedimentbeneath an ice tunnel then in areas of very low sediment supplythere would be no geomorphic signature of the ice tunnel unlessflows eroded into the substrate forming a tunnel channel or canal(cf Stalker 1960)

8 Conclusions and implications

Our data when compared with that from other eskers sug-gest that not only esker formation but landform morphology andarchitecture is primarily governed by the balance between watersupply sediment supply and the impact these have on sedi-mentation rate (Fig 19) Like for many eskers (eg Brennand1994 2000 Brennand and Shaw 1996 Fiore et al 2002 Burkeet al 2008 2010 Cummings et al 2011) the architecture ofthose in south-western Alberta suggests a highly dynamicdepositional environment Although many of the criteria sug-gested as diagnostic of high magnitude deposition within eskersare present (cf Burke et al 2010) their relatively fine-grainedcomposition (dominantly sand and pebbles with only limitedcobbles and boulders) suggests final esker architecture wasconstrained by sediment supply This constraint is reflected inlandform patterns because the largest eskers are located withinbroad meltwater valleys filled with pre-existing sand and gravelwhereas the shortest eskers that are dominantly composed ofdeformed diamicton or bedrock rather than waterlain sand andgravel are located in areas dominated by clast-poor till Wheresediment and water supply are high there is a dynamic linkbetween ice tunnel growth and sedimentation rate with icetunnel growth forcing sedimentation and sedimentation forcingice tunnel growth to develop macroforms (depocentres) andresulting in the most complex esker architecture and non-uniform esker geometry (scenario I Fig 19a) Where sedimentsupply is limited but flow power (water supply) remains highoften down glacier from depocentres (sediment exhaustion) ordue to an inadequate substrate composition (fine sediment willflush rather than deposit cf Cummings et al (2011)) sedimen-tation rate is low and the ice tunnel grows independent ofsedimentation resulting in more uniform ridge geometry and asedimentary architecture dominated by vertical landform growth(scenario II Fig 19b) Where sediment is truly supply-limited anesker sensu stricto (ridge of stratified sand and gravel Banerjeeand McDonald (1975)) cannot form because sedimentationrates are very low and the ice tunnel does not fill with waterlainsediment Instead where the ice tunnel is atop a deformablesubstrate the esker will be composed of deformed diamicton orbedrock due to squeezing of pressurized material into a lowpressure ice tunnel resulting in uniform esker geometry andconformable sedimentary architecture (scenario III Fig 19c)Given such landforms like lava-filled ice tunnels in BritishColumbia (cf Mathews 1958) record the infilling of an ice-walled channel and thus ice tunnel location we suggest theterm esker should apply to all ice tunnel infills regardless ofcomposition because the presence of stratified sand and gravel isirrelevant to locating former ice tunnels It should also be notedthat the absence of both a deformable substrate and insufficient

sediment for esker building means there will likely be nodepositional record of the ice-walled channel (scenario IIIFig 19c) Consequently although eskers can be used to identifyice-walled channel locations an esker map is inherently a min-imum map of ice-walled channel location Furthermore at leastin many examples (eg Brennand 1994 Cummings et al 2011Burke et al 2012a) esker sediments appear to record the lastmajor event(s) through an ice tunnel rather than the long termdrainage record These observations suggest care must be takenwhen using maps of esker crestline alone to test numerical icesheet models (eg Boulton et al 2009) Whilst esker crestlinemaps provide a minimum map of ice-walled channel locationthey cannot be used to infer sedimentation processes Instead adetailed assessment of esker ridge morphology coupled withknowledge of subsurface architecture and geomorphic context isneeded Our data suggest morpho-sedimentary relationshipsthat when likely variations in local sediment supply (iegeomorphic context) are taken into account could be used toelucidate esker sedimentation processes (eg ridge enlarge-ments are depocentresmacroforms) giving detailed insight intothe likely depositional environment and processes of ice tunnelevolution Validating these morpho-sedimentary relationshipswith larger data sets would improve the potential of remotepredictive mapping of eskers ultimately allowing inferences tobe made about spatial (and temporal) variations in esker archi-tecture and hence aggregate potential over much larger areas

Acknowledgements

This work was supported by a scholarship to MJB from TheLeverhulme Trust an NSERC Discovery grant and equipment grantsto TAB We are grateful to local landowners for granting access totheir land and thank Melissa Dinsdale and Aaron Dixon for assis-tance in the field We would like to thank Tim Fisher and NeilGlasser for their invaluable comments that have significantlyimproved the manuscript

Appendix A Supplementary data

Supplementary data related to this article (fence diagrams ofgrids 3 amp 5 data and interpretation (not presented in the manu-script) two-dimensional presentation of all GPR data and in-terpretations as well as animation 1-5) can be found at httpdxdoiorg101016jquascirev201502017

References

AltaLIS 2014 httpwwwaltaliscomproductsterrainlidar15_demhtml (accessed270714)

Arcone SA Lawson DE Delaney AJ 1995 Short-pulse radar wavelet recoveryand resolution of dielectric contrasts within englacial and basal ice of Mata-nuska Glacier Alaska USA J Glaciol 41 68e86

Aylsworth JM Shilts WW 1989 Glacial Features Around the Keewatin Ice DivideDistricts of Mackenzie and Keewatin Geological Survey of Canada p 21

Banerjee I McDonald BC 1975 Nature of esker sedimentation In Jopling AVMcDonald BC (Eds) Glaciofluvial and Glaciolacustrine Sedimentation Societyof Economic Palaeonologists and Mineralogists pp 304e320

Bartholomaus TC Anderson RS Anderson SP 2011 Growth and collapse of thedistributed subglacial hydrologic system of Kennicott Glacier Alaska USA andits effects on basal motion J Glaciol 57 985e1002

Bell RE 2008 The role of subglacial water in ice-sheet mass balance Nat Geosci 1297e304

Boulton GS Caban PE Van Gijssel K Leijnse A Punkari M Van Weert FHA1996 The impact of glaciation on the groundwater regime of NorthwestEurope Glob Planet Change 12 397e413

Boulton GS Hagdorn M Maillot PB Zatsepin S 2009 Drainage beneath icesheets groundwaterechannel coupling and the origin of esker systems fromformer ice sheets Quat Sci Rev 28 621e638

Brennand TA 1994 Macroforms large bedforms and rhythmic sedimentary se-quences in subglacial eskers south-central Ontario implications for eskergenesis and meltwater regime Sediment Geol 91 9e55

Brennand TA 2000 Deglacial meltwater drainage and glaciodynamics inferencesfrom Laurentide eskers Canada Geomorphology 32 263e293

Brennand TA Shaw J 1996 The Harricana glaciofluvial complex Abitibi regionQuebec its genesis and implications for meltwater regime and ice-sheet dy-namics Sediment Geol 102 221e262

Burke MJ Brennand TA Perkins AJ 2012a Transient subglacial hydrology of athin ice sheet insights from the Chasm esker British Columbia Canada QuatSci Rev 58 30e55

Burke MJ Brennand TA Perkins AJ 2012b Evolution of the subglacial hydro-logic system beneath the rapidly decaying Cordilleran Ice Sheet caused by ice-dammed lake drainage implications for meltwater-induced ice accelerationQuat Sci Rev 50 125e140

Burke MJ Woodward J Russell AJ Fleisher PJ Bailey PK 2008 Controls onthe sedimentary architecture of a single event englacial esker SkeiethararjeurookullIceland Quat Sci Rev 27 1829e1847

Burke MJ Woodward J Russell AJ Fleisher PJ Bailey PK 2010 The sedi-mentary architecture of outburst flood eskers a comparison of ground-penetrating radar data from Bering Glacier Alaska and Skeithararjokull Ice-land Geol Soc Am Bull 122 1637e1645

Clark PU Walder JS 1994 Subglacial drainage eskers and deforming beds beneaththe Laurentide and Eurasian ice sheets Geol Soc Am Bull 106 304e314

Cummings DI Gorrell G Guilbault JP Hunter JA Logan C Ponomarenko DAndre JMP Pullan SE Russell HAJ Sharpe DR 2011 Sequence stratig-raphy of a glaciated basin fill with a focus on esker sedimentation Geol SocAm Bull 123 1478e1496

Dyke AS Andrews JT Clark PU England JH Miller GH Shaw J Veillette JJ2002 The Laurentide and Innuitian ice sheets during the Last Glacial MaximumQuat Sci Rev 21 9e31

Evans DJA 2000 Quaternary geology and geomorphology of the Dinosaur Pro-vincial Park area and surrounding plains Alberta Canada the identification offormer glacial lobes drainage diversions and meltwater flood tracks Quat SciRev 19 931e958

Evans DJA Clark CD Rea BR 2008 Landform and sediment imprints of fastglacier flow in the southwest Laurentide Ice Sheet J Quat Sci 23 249e272

Evans DJA Hiemstra JF Boston CM Leighton I Cofaigh CO Rea BR 2012Till stratigraphy and sedimentology at the margins of terrestrially terminatingice streams case study of the western Canadian prairies and high plains QuatSci Rev 46 80e125

Fiore J Pugin A Beres N 2002 Etudes sedimentologique et georadar des depotssous-glaciaires de la vallee de Joux (vaud Suisse) Accretion en backsets dansun esker suivie dun jeurookulhlaup erosif Sedimentological and GPR studies ofsubglacial deposits in the Joux Valley (Vaud Switzerland) backset accretion inan esker followed by an erosive jeurookulhlaup vol 56 pp 19e32

Fricker HA Scambos T 2009 Connected subglacial lake activity on lower Mercerand Whillans Ice Streams West Antarctica 2003e2008 J Glaciol 55 303e315

Fricker HA Scambos T Bindschadler R Padman L 2007 An active subglacialwater system in West Antarctica mapped from space Science 315 1544e1548

Gray L 2005 Evidence for subglacial water transport in the West Antarctic IceSheet through three-dimensional satellite radar interferometry Geophys ResLett 32

Heinz J Aigner T 2003 Three-dimensional GPR analysis of various Quaternarygravel-bed braided river deposits (southwestern Germany) In Bristow CDJol HM (Eds) Ground Penetrating Radar in Sediments Geological Society ofLondon pp 99e110

Hooke RL Fastook J 2007 Thermal conditions at the bed of the Laurentide icesheet in Maine during deglaciation implications for esker formation J Glaciol53 646e658

Jackson LE Andriashek LD Phillips FM 2011 Limits of successive middle andlate pleistocene continental ice sheets interior plains of southern and centralalberta and adjacent areas Dev Quat Sci 15 575e589

Klassen RW 1989 Quaternary geology of the southern Canadian Interior PlainsIn Fulton RJ (Ed) The Geology of North America Quaternary Geology ofCanada and Greenland Geological Survey of Canada Geological Society ofAmerica Boulder CO pp 138e174

Kostic B Aigner T 2007 Sedimentary architecture and 3D ground-penetratingradar analysis of gravelly meandering river deposits (Neckar Valley SW Ger-many) Sedimentology 54 789e808

Mathews WH 1958 Geology of the Mount Garibaldi map-area southwesternBritish Columbia Canada part II geomorphology and quaternary volcanicrocks Geol Soc Am Bull 69 179e198

Munro M Shaw J 1997 Erosional origin of hummocky terrain in south-centralAlberta Canada Geology 25 1027

OCofaigh C Evans DJA Smith IR 2009 Large-scale reorganization and sedi-mentation of terrestrial ice streams during late Wisconsinan Laurentide IceSheet deglaciation Geol Soc Am Bull 122 743e756

Pawluk S Bayrock L 1969 Some characteristics and physical properties of Albertatills Res Counc Alta 26 78

Price RJ 1966 Eskers near the Casement glacier Alaska Geogr Ann Ser A 48111e125

Rains B Shaw J Skoye R Sjogren D Kvill D 1993 Late Wisconsin subglacialmegaflood paths in Alberta Geology 21 323

Shaw J Rains B Eyton R Welssling L 1996 Laurentide subglacial outburstfloods landform evidence from digital elevation models Can J Earth Sci 331154e1168

Shetsen I 1987 Quaternary Geology Southern Alberta Alberta Research CouncilMap 207

Shetsen I 1990 Quaternary Geology Central Alberta Alberta Research CouncilMap 213

Shilts WW 1984 Esker sedimentation models Deep Rose Lake map area districtof Keewatin Geol Surv Can Curr Res 84 217e222

Shreve RL 1985 Esker characteristics in terms of glacier physics Katahdin eskersystem Maine Geol Soc Am Bull 96 639e646

Shulmeister J 1989 Flood deposits in the Tweed Esker (southern Ontario Canada)Sediment Geol 65 153e163

Sjogren DB Rains RB 1995 Glaciofluvial erosional morphology and sediments ofthe Coronation e Spondin Scabland east-central Alberta Can J Earth Sci 32565e578

Sjogren DB Fisher TG Taylor LD Jol HM Munro-Stasiuk MJ 2002 Incipienttunnel channels Quat Int 90 41e56

Smith RC Sjogren DB 2006 An evaluation of electrical resistivity imaging (ERI)in Quaternary sediments southern Alberta Canada Geosphere 2 287e298

St-Onge DA 1972 Sequence of glacial lakes in north-central Alberta Geol SurvCan Bull 213

Stalker AM 1960 Ice-pressed drift forms and associated deposits in Alberta GeolSurv Can Bull 57 38

Stearns LA Smith BE Hamilton GS 2008 Increased flow speed on a large EastAntarctic outlet glacier caused by subglacial floods Nat Geosci 1 827e831

Storrar RD Stokes CR Evans DJA 2014 Increased channelization of subglacialdrainage during deglaciation of the Laurentide Ice Sheet Geology

Tylmann K Piotrowski JA Wysota W 2013 The icebed interface mosaicdeforming spots intervening with stable areas under the fringe of the Scandi-navian Ice Sheet at Sampława Poland Boreas 42 428e441

Wingham DJ Siegert MJ Shepherd A Muir AS 2006 Rapid discharge connectsAntarctic subglacial lakes Nature 440 1033e1036

Woodward J Burke MJ Tinsley R Russell AJ 2008 Investigating reworking ofproglacial sediments using GPR Skeiethararjeurookull Iceland In 12th InternationalConference on Ground Penetrating Radar University of Birmingham

Wooldridge CL Hickin EJ 2005 Radar architecture and evolution of channel barsin Wandering Gravel-Bed Rivers Fraser and Squamish Rivers British ColumbiaCanada J Sediment Res 75 844e860

SUPPORTING INFORMATION The role of sediment supply in esker formation and ice tunnel evolution

Figure S1 Maps of grids 1-5 (Strathmore and Clear Lake esker systems) overlain on esker morphology and meltwater valley fill Grid locations relative to one another and the landforms are shown in Fig 2

Figure S2 Maps of grids 6-9 (Didsbury esker network and Torrington esker ridge) overlain on esker morphology and meltwater valley fill Grid locations relative to one another and the landforms are shown in Fig 2

Figure S3 Two-dimensional views of GPR profiles X1 and X3 collected at grid 1 (Strathmore esker system) True relative line orientations are

shown in Figure S1 The data have been processed following the protocol described in sect 22 of the manuscript and interpreted following the

method outlined in sect 23 Radar element bounding surfaces are indicated by bold black lines and radar elements are labelled in approximate

depositional order Reflection offsets (faults) are indicated by dashed green lines and water table (continuous horizontal reflection that does not

correspond to primary bedding structure) is indicated by a black dashed line

Figure S4 Two-dimensional interpreted GPR profiles X1 and X3 collected at grid 1 (Strathmore esker system) True relative line orientations

are shown in Figure S1 Radar elements have been colour-coded based upon the broad style of deposition using the criteria outlined in sect 23

These colours correspond to those in legend (left) The radar elements are numbered in approximate depositional order Water table is

indicated by a black dashed line and faults by white dashed lines

Figure S5 Two-dimensional views of GPR profiles X2a and X2b collected at grid 1 (Strathmore esker system) True relative line orientations

are shown in Figure S1 The data have been processed following the protocol described in sect 22 of the manuscript and interpreted following the

Figure S6 Two-dimensional interpreted GPR profiles X2a and X2b collected at grid 1 (Strathmore esker system) True relative line orientations

Figure S7 Two-dimensional views of GPR and ERT profiles Y1 ndash Y8 collected at grid 1 (Strathmore esker system) True relative line orientations are shown in Figure S1 The data have been processed following the protocol described in sect 22 of the manuscript and interpreted following the method outlined in sect 23 Radar element bounding surfaces are indicated by bold black lines and radar elements are labelled in approximate depositional order Reflection offsets (faults) are indicated by dashed green lines and water table (continuous horizontal reflection that does not correspond to primary bedding structure) is indicated by a black dashed line

Figure S8 Two-dimensional interpreted GPR profiles Y1 ndash Y8 collected at grid 1 (Strathmore esker system) True relative line orientations are shown in Figure S1 Radar elements have been colour-coded based upon the broad style of deposition using the criteria outlined in sect 23 These colours correspond to those in legend (left) The radar elements are numbered in approximate depositional order Water table is indicated by a black dashed line and faults by white dashed lines

Figure S9 Two-dimensional views of GPR profiles Y9 ndash Y14 collected at grid 1 (Strathmore

esker system) True relative line orientations are shown in Figure S1 The data have been

processed following the protocol described in sect 22 of the manuscript and interpreted

following the method outlined in sect 23 Radar element bounding surfaces are indicated by

bold black lines and radar elements are labelled in approximate depositional order

Reflection offsets (faults) are indicated by dashed green lines and water table (continuous

horizontal reflection that does not correspond to primary bedding structure) is indicated by

black dashed line

Figure S10 Two-dimensional interpreted GPR profiles Y9 ndash Y14 collected at grid 1

(Strathmore esker system) True relative line orientations are shown in Figure S1 Radar

elements have been colour-coded based upon the broad style of deposition using the

criteria outlined in sect 23 These colours correspond to those in legend (left) The radar

elements are numbered in order of deposition

Figure S11 Two-dimensional views of GPR profile X1 collected at grid 2 (Strathmore esker system) True relative line orientations are shown

in Figure S1 The data have been processed following the protocol described in sect 22 of the manuscript and interpreted following the method

outlined in sect 23 Radar element bounding surfaces are indicated by bold lines and radar elements are labelled in approximate depositional

order Reflection offsets (faults) are indicated by dashed green lines and black dashed line indicates the maximum depth of GPR penetration

Figure S12 Two-dimensional interpreted GPR profile X1 collected at grid 2 (Strathmore esker system) True relative line orientations are

shown in Figure S1 Radar elements have been colour-coded based upon the broad style of deposition using the criteria outlined in sect 23

These colours correspond to those in legend (bottom right) The radar elements are numbered in approximate depositional order The black

dashed line indicates the maximum depth of GPR penetration and faults are indicated by white dashed lines

Figure S13 [PAGE 13] Two-dimensional views of GPR and ERT profiles Y1 ndash Y10

collected at grid 2 (Strathmore esker system) True relative line orientations are shown in

Figure S1 The data have been processed following the protocol described in sect 22 of the

manuscript and interpreted following the method outlined in sect 23 Radar element bounding

surfaces are indicated by bold black lines and radar elements are labelled in approximate

depositional order Reflection offsets (faults) are indicated by dashed green lines (white over

ERT profiles) and black dashed line indicates the maximum depth of GPR penetration

Figure S14 [PAGE 14] Two-dimensional interpreted GPR profiles Y1 ndash Y10 collected at

grid 2 (Strathmore esker system) True relative line orientations are shown in Figure S1

Radar elements have been colour-coded based upon the broad style of deposition using the

criteria outlined in sect 23 These colours correspond to those in legend (middle right) The

radar elements are numbered in approximate depositional order The black dashed line

indicates the maximum depth of GPR penetration and faults are indicated by white dashed

Figure S15 Two-dimensional views of GPR profiles Y11 ndash Y14 collected at grid 2 (Strathmore esker system) True relative line orientations are

depositional order Reflection offsets (faults) are indicated by dashed green lines and black dashed line indicates the maximum depth of GPR

penetration

Figure S16 Two-dimensional interpreted GPR profiles Y11 ndash Y14 collected at grid 2 (Strathmore esker system) True relative line orientations

These colours correspond to those in legend (bottom) The radar elements are numbered in approximate depositional order The black dashed

line indicates the maximum depth of GPR penetration and faults are indicated by white dashed lines

Figure S17 Two-dimensional view of ERT profiles X1 and Y1 collected at grid 3 (Strathmore esker system) The intersection points of the lines

are indicated by the labelled vertical lines The water table elevation (derived from that of nearby water bodies) is indicated by the blue arrow on

the y-axes The red arrows on the x-axes indicate the position of shallow augering GPR performance was poor here despite the presence of

up to 9 m of high-resistivity material within the landform This is likely due to the presence of an ~1 m thick surface unit of low-resistivity material

that had become saturated by several days of rainfall immediately prior to data collection

Figure S18 Two-dimensional views of GPR profile X1a collected at grid 4 (Clear Lake esker system) True relative line orientations are shown

outlined in sect 23 Radar element bounding surfaces are indicated by bold black lines and radar elements are labelled in approximate

depositional order Reflection offsets (faults) are indicated by dashed green lines

Figure S19 Two-dimensional interpreted GPR profile X1a collected at grid 4 (Clear Lake esker system) True relative line orientations are

These colours correspond to those in legend (bottom) The radar elements are numbered in approximate depositional order Faults are

indicated by white dashed lines

Figure S20 [PAGE 21] Two-dimensional views of GPR profile X1b collected at grid 4 (Clear Lake esker system) True relative line orientations are shown in Figure S1 The data have been processed following the protocol described in sect 22 of the manuscript and interpreted following the method outlined in sect 23 Radar element bounding surfaces are indicated by bold black lines and radar elements are labelled in approximate depositional order Reflection offsets (faults) are indicated by dashed green lines Figure S21 [PAGE 22] Two-dimensional interpreted GPR profile X1b collected at grid 4 (Clear Lake esker) True relative line orientations are shown in Figure S1 Radar elements have been colour-coded based upon the broad style of deposition using the criteria outlined in sect 23 These colours correspond to those in legend (bottom right) The radar elements are numbered in approximate depositional order Faults are indicated by white dashed lines Figure S22 [PAGE 23] Two-dimensional views of GPR profiles Y1 ndash Y7 collected at grid 4 (Clear Lake esker system) True relative line orientations are shown in Figure S1 The data have been processed following the protocol described in sect 22 of the manuscript and interpreted following the method outlined in sect 23 Radar element bounding surfaces are indicated by bold black lines (approximate position is dashed) and radar elements are labelled in approximate depositional order Reflection offsets (faults) are indicated by dashed green lines Figure S23 [PAGE 24] Two-dimensional interpreted GPR profiles Y1 ndash Y7 collected at grid 4 (Clear Lake esker system) True relative line orientations are shown in Figure S1 Radar elements have been colour-coded based upon the broad style of deposition using the criteria outlined in sect 23 These colours correspond to those in legend (bottom right) The radar elements are numbered in approximate depositional order Faults are indicated by white dashed lines Figure S24 [PAGE 25] Two-dimensional views of GPR profiles Y8 ndash Y11 collected at grid 4 (Clear Lake esker system) True relative line orientations are shown in Figure S1 The data have been processed following the protocol described in sect 22 of the manuscript and interpreted following the method outlined in sect 23 Radar element bounding surfaces are indicated by bold black lines and radar elements are labelled in approximate depositional order Reflection offsets (faults) are indicated by dashed green lines Figure S25 [PAGE 26] Two-dimensional interpreted GPR profiles Y8 ndash Y11 collected at grid 4 (Clear Lake esker system) True relative line orientations are shown in Figure S1 Radar elements have been colour-coded based upon the broad style of deposition using the criteria outlined in sect 23 These colours correspond to those in legend (bottom) The radar elements are numbered in approximate depositional order Faults are indicated by white dashed lines

Figure S26 [PAGE 28] Fence diagram of GPR and ERT profiles collected at grid 5 (Clear Lake esker system) The data have been processed following the protocol detailed in sect 22 Radar element bounding surfaces are indicated by bold lines and radar elements have been labelled in approximate order of deposition (A - CD) Offsets in reflection (faults) are indicated by dashed green lines (these are white where overlain on the ERT profile) Figure S27 [PAGE 29] Fence diagram of interpreted GPR lines collected at grid 5 (Clear Lake esker system) Radar elements have been colour-coded based upon the broad style of deposition within each radar element following the protocol detailed in sect 23 These colours correspond to those in the legend (bottom right) The radar elements have been numbered in approximate order of deposition (1-23) The white dashed lines mark faults (reflection offsets) Figure S28 [PAGE 30] Two-dimensional views of GPR profile X1 (0 ndash 1000 m) collected at grid 5 (Clear Lake esker system) True relative line orientations are shown in Figure S1 The data have been processed following the protocol described in sect 22 of the manuscript and interpreted following the method outlined in sect 23 Radar element bounding surfaces are indicated by bold black lines and radar elements are labelled in approximate depositional order Reflection offsets (faults) are indicated by dashed green lines Figure S29 [PAGE 31] Two-dimensional interpreted GPR profile X1 (0 ndash 1000 m) collected at grid 5 (Clear Lake esker system) True relative line orientations are shown in Figure S1 Radar elements have been colour-coded based upon the broad style of deposition using the criteria outlined in sect 23 These colours correspond to those in legend (bottom) The radar elements are numbered in approximate depositional order Faults are indicated by white dashed lines Figure S30 [PAGE 32] Two-dimensional views of GPR profile X1 (1000 ndash 2000 m) collected at grid 5 (Clear Lake esker system) True relative line orientations are shown in Figure S1 The data have been processed following the protocol described in sect 22 of the manuscript and interpreted following the method outlined in sect 23 Radar element bounding surfaces are indicated by bold black lines and radar elements are labelled in approximate depositional order Reflection offsets (faults) are indicated by dashed green lines Figure S31 [PAGE 33] Two-dimensional interpreted GPR profile X1 (1000 ndash 2000 m) collected at grid 5 (Clear Lake esker system) True relative line orientations are shown in Figure S1 Radar elements have been colour-coded based upon the broad style of deposition using the criteria outlined in sect 23 These colours correspond to those in legend (bottom) The radar elements are numbered in approximate depositional order Faults are indicated by white dashed lines

Figure S32 Two-dimensional views of GPR profile X1 (2000 ndash 2250 m) collected at grid 5 (Clear Lake esker system) True relative line

orientations are shown in Figure S1 The data have been processed following the protocol described in sect 22 of the manuscript and interpreted

following the method outlined in sect 23 Radar element bounding surfaces are indicated by bold black lines and radar elements are labelled in

approximate depositional order Reflection offsets (faults) are indicated by dashed green lines

Figure S33 Two-dimensional interpreted GPR profile X1 (2000 ndash 2250 m) collected at grid 5 (Clear Lake esker system) True relative line

orientations are shown in Figure S1 Radar elements have been colour-coded based upon the broad style of deposition using the criteria

outlined in sect 23 These colours correspond to those in legend (bottom right) The radar elements are numbered in approximate depositional

order Faults are indicated by white dashed lines

Figure S34 [PAGE 37] Two-dimensional views of GPR profiles Y1 ndash Y13 collected at grid 5

(Clear Lake esker system) True relative line orientations are shown in Figure S1 The data

have been processed following the protocol described in sect 22 of the manuscript and

interpreted following the method outlined in sect 23 Radar element bounding surfaces are

indicated by bold black lines and radar elements are labelled in approximate depositional

order Reflection offsets (faults) are indicated by dashed green lines

grid 5 (Clear Lake esker system) True relative line orientations are shown in Figure S1

criteria outlined in sect 23 These colours correspond to those in legend (bottom right) The

radar elements are numbered in approximate depositional order Faults are indicated by

white dashed lines

collected at grid 5 (Clear Lake esker system) True relative line orientations are shown in

the ERT profile)

white dashed lines

Figure S38 Two-dimensional views of GPR profile X1 collected at grid 9 (Didsbury esker network) True relative line orientations are shown in Figure S2 The data have been processed following the protocol described in sect 22 of the manuscript and interpreted following the method outlined in sect 23 Radar element bounding surfaces are indicated by bold black lines and radar elements are labelled in approximate depositional order Reflection offsets (faults) are indicated by dashed green lines

Figure S39 Two-dimensional interpreted GPR profile X1 collected at grid 9 (Didsbury esker network) True relative line orientations are shown in Figure S2 Radar elements have been colour-coded based upon the broad style of deposition using the criteria outlined in sect 23 These colours correspond to those in legend (bottom) The radar elements are numbered in approximate depositional order Faults are indicated by white dashed lines

(Didsbury esker network) True relative line orientations are shown in Figure S2 The data

order Reflection offsets (faults) are indicated by dashed green lines (white over the ERT

profile)

Figure S41 [PAGE 45] Two-dimensional interpreted GPR profiles Y1 ndash Y8 collected at grid

9 (Didsbury esker) True relative line orientations are shown in Figure S2 Radar elements

have been colour-coded based upon the broad style of deposition using the criteria outlined

in sect 23 These colours correspond to those in legend (bottom) The radar elements are

numbered in approximate depositional order Faults are indicated by white dashed lines

The role of sediment supply in esker formation and ice tunnel evolution


11 Regional setting

2 Methods




3 Esker system types in south-western Alberta and their morphology




34 Torrington esker


41 Esker ridge segments within meltwater valleys containing broad valley fill

411 Strathmore esker

4111 Grid 1

4112 Grid 2

412 Clear Lake esker

413 Downglacier Didsbury esker (Grid 6)


421 Upglacier Didsbury esker

4211 Grid 7

4212 Grid 8

422 Torrington esker (Grid 9)



7 The role of sediment supply in ice tunnel evolution and esker architecture

71 Scenario I ndash abundant sediment supply and flow power (Fig 18a)

72 Scenario II ndash limited sediment supply and high flow power (Fig 18b)

73 Scenario III ndash truly sediment supply-limited (Fig 18c)


Acknowledgements


References

Given that evolution of the subglacial hydrologic system candirectly influence ice dynamics it is important that ice sheetmodels incorporate accurate information about the timescales andthe processes responsible for ice tunnel evolution However therelative inaccessibility of contemporary ice sheet beds makes directinvestigation of subglacial hydrologic systems challenging (Frickerand Scambos 2009) Former ice sheet beds contain a wealth ofglaciofluvial landforms such as eskers that if correctly interpretedcan provide detailed insight into the hydrology of former andcontemporary ice sheets Eskers typically defined as ridges ofstratified sand and gravel are the main depositional signature ofchannelized meltwater flow on within or beneath ice masses(Banerjee and McDonald 1975) and record the location of ice-walled channels on former ice sheet beds Consequently eskershave been used to infer former ice sheet dynamics (eg Shilts1984 Aylsworth and Shilts 1989 Clark and Walder 1994Brennand 2000 Storrar et al 2014) and to test numericalmodels (eg Boulton et al 2009) However much debate remainsas to the processes responsible for esker formation (Cummingset al 2011) and although recent work has shown that the glacialhydrologic system is never in steady state (eg Gray 2005Wingham et al 2006 Fricker et al 2007 Bell 2008 Stearnset al 2008 Bartholomaus et al 2011) and that eskers may actu-ally record relatively short lived and dynamic events (egBrennand 1994 Burke et al 2008 2010 2012a Cummings et al2011) numerical models have typically assumed steady stateesker formation (eg Hooke and Fastook 2007 Boulton et al2009) Detailed work on glacial lake outburst flood (GLOF) eskersat contemporary glaciers has demonstrated that the balance be-tween water supply and sediment supply drive esker formation(Burke et al 2008 2010) These observations are echoed in thelandform record and suggest that dynamic esker deposition isfundamentally controlled by variations in ice tunnel geometry(accommodation space) and sedimentation rate (Brennand 1994Brennand and Shaw 1996 Burke et al 2012a) Over the shorttimescales of transient events often recorded in eskers(Shulmeister 1989 Brennand 1994 2000 Brennand and Shaw1996 Cummings et al 2011) ice tunnel geometry is a balancebetween growth by frictional melting and mechanical excavationand ice tunnel closure through sedimentation rather than ice creep(Burke et al 2012a) It is imperative that numerical models areconsistent with the landform record thus they should incorporateboth the transient conditions as well as seasonal drainage patterns(Burke et al 2012b) However most investigations of eskers havefocused on regions containing the largest examples and the largestexamples within regions (Shilts 1984 Brennand 1994 Burke et al2008 2010 2012a Cummings et al 2011 Storrar et al 2014)where there was both a ready supply of meltwater (Brennand2000 Storrar et al 2014) and sediment (Brennand 2000) Thelack of large eskers on the Prairies has resulted in only limitedinvestigation of these landforms (Stalker 1960) yet understandingwhat controlled their architecture and apparent regionallyderanged distribution (Brennand 2000) could provide furtherinsight into the fundamental controls on esker formation Theassertion that esker distribution in North America was primarilysubstrate controlled (Clark andWalder 1994 Boulton et al 1996) isclearly insufficient to explain the presence of 506 esker ridge seg-ments in south-western Alberta (Fig 1) The suggestion thatmeltwater supply controlled the presence of the longest eskers onthe Canadian shield (Brennand 2000 Storrar et al 2014) fails toaddress what controlled the distribution of the eskers in thewestern Canadian sedimentary basin It is only through investiga-tion of these smaller esker systems coupled with our knowledge of

their larger counterparts and observations from contemporary icemasses that a unifying theory of the controls on esker formationcan be achieved

In this paper we combine detailed geomorphic and geophysicaldata from four eskers in south-western Alberta in order to assessthe controls on their formation and distribution Because theseeskers have distinct geomorphic characteristics and depositionalsettings we compare their morpho-sedimentary characteristics toidentify the controls on their variability Through comparison ofthese data with detailed morpho-sedimentary investigations ofeskers elsewhere we suggest models for esker formation thatprovide insight into the major controls on the evolution of thesubglacial hydrologic system at the ice tunnel scale and ultimatelytake a step closer to a unified understanding of the controls onesker formation

11 Regional setting

The plains of southern Alberta are situated on south-west dip-ping Tertiary and Cretaceous aged sandstones and shales that thineastward from the foothills of the RockyMountains This bedrock isoverlain by a carapace of glacial material which due to the friablenature of local bedrock typically has a fine-grained textureparticularly in areas overlaying Cretaceous shales (Pawluk andBayrock 1969) These sediments record the coalescence of iceflowing eastward from the Cordilleran Ice Sheet and ice flowingwestward from the Laurentide Ice Sheet (Jackson et al 2011) Theseflows converged and deflected toward the south-southeast acrossthe plains of southern Alberta (Klassen 1989 Evans et al 2008)reaching a maximum extent about 20e21 14C ka BP (Dyke et al2002) This southerly ice flow and its associated meltwater havegenerated a wealth of glacial landforms including mega-flutesmega-lineations tracts of smoothed terrain moraines meltwaterchannels eskers and buried valleys (eg Sjogren and Rains 1995Munro and Shaw 1997 Sjogren et al 2002 Evans et al 20082012) Although till is often thin and fine grained it thickens atlarge moraine complexes where multiple units are stacked (Evanset al 2008) Together these sediments and landforms have beenassigned to landsystems associated with rapid ice flow (Evans2000 Evans et al 2008 OCofaigh et al 2009) or inferred to re-cord meltwater erosion (Rains et al 1993 Shaw et al 1996) Eskersin south-western Alberta range in length but the two longest es-kers (Strathmore and Clear Lake eskers) are located within broadmeltwater valleys (Fig 1) that are orientated northwest-southeastand probably occur downflow of ice-dammed lakes (St-Onge1972) Despite the plethora of eskers in south-western Alberta(Shetsen 1987 1990 Evans et al 2008) they have received littledetailed attention probably due to their relatively small size(making identification in aerial photographs and on digital eleva-tion models (DEMs) difficult) and lack of major sedimentexposures

2 Methods

We mapped esker ridge segment crestlines esker system geo-morphology meltwater valley walls and the geomorphology ofmeltwater valley fill in a GIS Esker crestlines and meltwater valleywalls were mapped as polylines using LiDAR-derived DEMs (15 mpostings vertical and horizontal accuracies of 03 m and 05 mrespectively AltaLIS 2014) with multiple hillshade illuminationsWe used sub-regional patterns to differentiate eskers from mo-raines and used elevated linear breaks in slope to delineate melt-water valley walls Meltwater valleys typically contain underfit

18 1708 011285 (plusmn 00094)16 22292 300

13 2343 012532 (plusmn 00016)23 4536514 16715 009513 (plusmn 001)16 147711 123612 1612 01149f

Y4 111 1 31 292 0923 X1 1665 15 48 327 086

Y1 112 1 29 293 086Clear Lake 4 Y8 1665 15 75 297 098

5 Y18 1665 15 71 497 069Didsbury 6 Y1 1665 15 26 347 13

7 Y7 111 1 47 291 091Torrington 9 Y2 111 1 89 294 096

Y5 111 1 16 282 086

34 Torrington esker

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

18 1708 011285 (plusmn 00094)16 22292 300

13 2343 012532 (plusmn 00016)23 4536514 16715 009513 (plusmn 001)16 147711 123612 1612 01149f

Y4 111 1 31 292 0923 X1 1665 15 48 327 086

Y1 112 1 29 293 086Clear Lake 4 Y8 1665 15 75 297 098

5 Y18 1665 15 71 497 069Didsbury 6 Y1 1665 15 26 347 13

7 Y7 111 1 47 291 091Torrington 9 Y2 111 1 89 294 096

Y5 111 1 16 282 086

34 Torrington esker

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

18 1708 011285 (plusmn 00094)16 22292 300

13 2343 012532 (plusmn 00016)23 4536514 16715 009513 (plusmn 001)16 147711 123612 1612 01149f

Y4 111 1 31 292 0923 X1 1665 15 48 327 086

Y1 112 1 29 293 086Clear Lake 4 Y8 1665 15 75 297 098

5 Y18 1665 15 71 497 069Didsbury 6 Y1 1665 15 26 347 13

7 Y7 111 1 47 291 091Torrington 9 Y2 111 1 89 294 096

Y5 111 1 16 282 086

34 Torrington esker

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

18 1708 011285 (plusmn 00094)16 22292 300

13 2343 012532 (plusmn 00016)23 4536514 16715 009513 (plusmn 001)16 147711 123612 1612 01149f

Y4 111 1 31 292 0923 X1 1665 15 48 327 086

Y1 112 1 29 293 086Clear Lake 4 Y8 1665 15 75 297 098

5 Y18 1665 15 71 497 069Didsbury 6 Y1 1665 15 26 347 13

7 Y7 111 1 47 291 091Torrington 9 Y2 111 1 89 294 096

Y5 111 1 16 282 086

34 Torrington esker

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Y4 111 1 31 292 0923 X1 1665 15 48 327 086

Y1 112 1 29 293 086Clear Lake 4 Y8 1665 15 75 297 098

5 Y18 1665 15 71 497 069Didsbury 6 Y1 1665 15 26 347 13

7 Y7 111 1 47 291 091Torrington 9 Y2 111 1 89 294 096

Y5 111 1 16 282 086

34 Torrington esker

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

34 Torrington esker

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

34 Torrington esker

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

34 Torrington esker

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

34 Torrington esker

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Acknowledgements

References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

black dashed line

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

penetration

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

white dashed lines

the ERT profile)

white dashed lines

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

profile)


11 Regional setting

2 Methods








34 Torrington esker




4111 Grid 1

4112 Grid 2





4211 Grid 7

4212 Grid 8









Acknowledgements


References

Documents

The role of sediment supply in esker formation and ice tunnel evolution