The formation of ice-marginal embankments into ice-dammed lakes in the eastern Puget Lowland,Washington, U.S.A., during the late Pleistocene

DEREK B. BOOTH

BORE \S 3i?,li;?illii;#XTtX'.'ft:*:ti3T::fi:;Tt'JTl""il':115li::';:*::.jl':t:',li?'i:1:Oslo. ISSN 0300-9483.

Large embankments, typically several kilometers in lateral extent and many tens of meters high, choke

the mouths of each alpine valley of the central Cascade Range, Washington Stale, U.S.A.. at or near their

junctions with the Puget Lowland. They comprise till and bedded gravel, sand, and silt, aggraded into ice-

dammed lakes. The embankments l ie within the late-Pleistocene Cordil leran ice-sheet l imit and so do not

mark the location ofthe ice-maximum terminus. Reconstruction ofthe subglacial hydraulic potential f ield

indicates that these ice-dammed lakes would have drained subglacially via spillways located near the junc-

tion of each alpine valley and the Lowland. Physical processes tended to stabilize the grounding line for

each ice tongue close to its respective spillway location. Because sedimentation rates are highest adjacent

to the grounding line, subaqueous sedimentation formed a growing embankment there. In some valleys,

subsequent subaerial lake drainage or decay of the active-ice dam resulted in late-stage deposition of del-

tas or valley trains. This analysis of ice-water behavior is based on physical principles that should be gener-

ally applicable to any environment where glaciers terminate against ice-dammed bodies of water.

Derek B. Booth, Il.S. Geological Survey, Quaternary Research Center, University of Washington AK-60,

Seattle, Washington 98195, U.S.A.; 6th February, 1985 (revised 7th October, 1985)

A striking characteristic of the topography alongthe western slopes of the Cascade Range ofWashington State, U.S.A., is the presence ofconstrictions in river valleys as they pass fromtheir bedrock-walled alpine reaches into the Pu-get Lowland (Figs. 1,2). These constrictive land-forms are not bedrock ridges but rather compriseglacial, fluvial, and lacustrine sedimentary mate-rial. They are here termed.'embankments' toavoid the common genetic implications of alter-native nomenclature such as 'morainal embank-ments' or'delta-moraines' that has been appliedto these features. All lie within the limits andnear the margin of the Puget lobe of the Cordille-ran ice sheet that occupied the Puget Lowlandmost recently about 15,000 years ago, during theVashon stade of the Fraser glaciation (Armstronget al. 1965\. Most of the embankment sedimentsare virtually unweathered, indicating transportand deposition during this late-glacial time(Booth 1986). A slightly earlier advance of al-pine glaciers that originated in the CascadeRange is inferred to have climaxed and wanedbefore the Vashon maximum (Cary & Carlston1937; Mackin 1941; Williams t97l;Porter 1976).The lower reaches of alpine valleys were thusfree of alpine-glacier ice during much or all of the

time that the Puget lobe occupied the Lowlandduring the Fraser glaciation.

Eleven valley-constricting embankments arerecognized in this area. Although a few earlyworkers believed that these embankments weredeposited by alpine ice flowing downvalley, moststudies of embankment morphology and clastprovenance have demonstrated that the sedimentwas derived from the Puget lobe (Cary & Carl-ston 1937; Mackin 1941; Knoll 1967; Williams1971; Hirsch 1975). These studies also docu-mented the presence of extensive ice-marginallakes, impounded by the Puget lobe, that filledthe lower reaches of the alpine valleys. A classicstudy by Mackin (19a1) of one of the major em-bankment systems characterized it as a set ofgreat glaciofluvial deltas that prograded into theadjacent alpine-valley lakes, veneered by till onthe ice-proximal faces. He believed that theyformed at the maximum stand of the Vashon-ageice sheet and were composed wholly of Vashon-age sediment. His work outlined the fundamentalrelationship between the ice sheet, alpine gla-ciers, and sedimentation associated with these icemasses that has guided all subsequent geologicstudy in this region.

This study addresses several characteristics of

248 Derek B. Booth

1 220 I 2 1030'

16 km0-

Fig. 1. Index map of geographical localities. Ernbankments aiemarked with stars. Vashon-age ice limit is shown with smalldots (Booth 1986), which approximates the boundary betweenthe Puget Lowland and the Cascade Ranse.

these embankments, particularly their commonvalley-mouth locations and the variety of sedi-ments composing them. The analysis is based inpart on available geologic data but stresses thephysical behavior of glacier ice near its marginand the attendant depositional processes that op-

BOREAS 15 (1986)

Fig. 2. Aerial view of the Wallace embankment (see Fig. 1 forlocation). View is to the east. The Wallace River has incisedthrough embankment sediments adjacent to the south bedrockwall of the valley in postglacial time.

erate wherever a glacier terminates in an ice-dammed lake. Proposed processes of aqueousdeposition adjacent to Pleistocene ice sheetshave received considerable recent attention a1several localities in eastern and central NorthAmerica (e.g. Dreimanis 1982; Shaw 1982; Eyles& Eyles 1983; Barnett 1985; Heiny 1985; Kulig1985). The examples from the Puget Lowlandshould prove quite relevant to these and othergeographically diverse localities. Despite rela-tively limited exposures of their geologic mate-rials, rendering detailed lithofacies schemes suchas proposed by Miall (1977) or Eyles & Eyles(1983) inappropriate, these features provide aunique opportunity to study ice-marginal pro-cesses in an area where lake geometry, drainageroutes, and configuration ofthe ice sheet are par-ticularly well constrained.

DescriptionThe eleven embankments in the study area (Fig.1) have several common characteristics. They oc-cupy alpine valleys, typically at or just above thejunction of the valley with the Lowland proper.Their tops are from 100 to several hundred me-ters above the valley floor and cover 1-1.0 km2.All appear to have filled their respective valleysfrom side to side (Fig. 2). Post-depositional rivererosion has incised them along the valley axis andtypically has removed much upvalley material.Less commonly, the downvalley faces have beenmodified as well (Fig. 3).

Embankment sediments comprise diamicfonand bedded gravel, sand, and silt in variable pro-

Bo

10 mi

,. .1Puget lowland

WASHINGTON

BOREAS 1s (1986)

F,g. J. North Fork Snoqualmie embankment (outl ined). Post-

deoositional erosion is evident both in the transverse incision by

th; North Fork Snoqualmie River (N 1/2 section 20) and by

modification of the ice-proximal face by recessional meltwater

flowing south down the valley of Deep Creek (Sections 18-19)'

Topography from USGS Mount Si 15' quadrangle. Ice-maxi-

mum limit shown by black dots; alt itude in feet.

portions. Fluvial sediment, deposited as eitherproglacial outwash or delta topset beds, typicallyforms much or all of their upper surfaces. Thecores of these embankments are more heterogen-eous and include significant amounts of subaque-ously or subglacially deposited sediment. These

Ice-marginalembankments 249

nonfluvial sediments are typically weakly strat-ified and lie in abrupt contact with the overlyingmaterial.

Three examples, treated in detail, exemplifyboth their similarities and differences. The SouthFork Tolt embankment is composed of thin, dis-continuous fluvial deposits covering a roundedridge of primarily glacially transported sedi-ments. The Middle-South Forks Snoqualmie em- -bankments, by contrast, are characterized by aplanar surface of fluvial sand and gravel that ex-tends both upvalley and laterally along the moun-tain front for many kilometers. The Pilchuck-Sul-tan embankment shares important characteristicsof both; it is a particularly useful example be-cause of good exposure and extensive subsurfacedata.

South Fork Tblt embankment

The South Fork Tolt embankment occupies anarea of about 2 km2, forming a broad north-risingridge with maximum relief of 150 m (500 ft) (Fig.4,A). Tiuncated sedimentary lavers on its steepeastern face indicate postglacial erosion of theembankment by the South Fork Tolt River.Through the southern part of the embankment,recent excavations have exposed a 100-m verticalsection of glacial, glaciolacustrine, and glacioflu-vial sediments.

This section is subdivided into three units (Fig.48). The lowest (unit 1) is an oxidized bouldery

Unit 3bedded sandand qravel

Unit 2crudely bedded

d iamicton

Unit Iox . d i ami cton

*A C-

2kmo#

1mi

Fig. 4. South Fork Tolt embankment. A: Topography from USGS Mount Si 15'quadrangle (embankment is outl ined). Ice-maxi-

mum limit shown by black dots; altitude in feet. B: Stratigraphy of embankment sediments exposed downstream of dam. Contacts

between units are not necessarily horizontal and so span a range of altitudes within the represented region. C: Pebbly silt diamicton

with crude horizontal bedding (location shown on Fig. 4.A; alt itude 520 m (1'710 ft)).

17 Boreas l5:3. 1986

)o3. .\ ?

,{-ixp/szoo.2'<;"-=-.../'

47042'30"=

250 Derek B. Booth

t2 , l

(

| ,-ro o u'

36: i

B',i::":sand andgravel

l4c

BOREAS ls (1986)

5OO m

350-4OO m300-

35O mUnit Ibeddedsand

Unit 2

and si l t OOm

Frg. 5. Middle-South ForksSnoqualmie embankment com-plex. A: Topography fromUSGS Mount Si and Banderal-5' quadrangles. Ice-maximumlimit shown by black dots; alt i-tude in feet (outl ined area in-cludes embankment and its SWcontinuation past CedarButte). B: Stratigraphy of em-bankment sediments in the vi-cinity of the Holocene channelof the South Fork SnoqualmieRiver.

unweath.t ' i l I

./--

, 2,km0ffi 12f 40'

lmi

diamicton containing lenses 5-30 cm thick ofcompact medium sand. Clast lithology and de-gree of weathering indicate probable depositionby an ice sheet of pre-Fraser age (Booth 1986).

A massive to crudely bedded diamicton (unit

2,Fig.48; Fig. 4C), 50 to 100 m thick, overliesthe lowest unit. This material constitutes the bulkof the embankment and defines the shape of itsupper surface. Gravel-sized clasts are round tosubangular, commonly striated, and of mixed

BOREAS ls (1986)

Iithology. The clasts have no apparent preferredorientation. The diamicton differs from typicalPuget-Lowland lodgment till in that it has a farlower degree of compaction and a slightly higherclay content in the matrix (c. l0%; cf. Crandell1963; Mulline aux 1970; Dethier er a/. 1981). Sub-horizontal pebble-poor zones, 5-20 cm thick thatextend laterally for several meters, are inter-spersed throughout the diamicton. Larger dis-continuous subhorizontal sand layers, extending1J:0 m, are clustered in, but not limited to, azone 2A40 m thick in the middle of the unit.Horizontal layering and planar contacts through-out the exposure show that no significant flow-ing, slumping, or glacitectonic deformation ofsediment occurred during or following deposi-tion. In one locality west of the south abutmentof the dam, silt and clay laminae drape over, andare depressed locally several centimeters be-neath, clasts (dropstones) as large as 15 cm indiameter.

A surface unit (unit 3, Fig. 48) of oxidized gla-ciofluvial recessional outwash, generally less than2 m thick, overlies the diamicton. It drapes mostof the embankment and thickens to form a con-structional surface only on the western (ice-prox-imal) slopes below about 500 m (1,600 ft) alti-tude, where the sediments merge with fluvial de-posits of the late ice recession (Booth 1986).

Middle and South Forks Snoqualmieembankments

A nearly continuous, flat-surfaced embankmentcomplex that chokes the valley mouths of theMiddle Fork and South Fork of the SnoqualmieRiver (Fig. 54) can be traced south for nearly 1-5km along the mountain front. It blocks the CedarRiver valley and terminates against ice-contactsediment at the northeast edge of the TaylorCreek valley (Frizzell et al. 1984). Its surface al-titude descends from about 510 m (L,670 ft) onthe proximal side of the Middle Fork embank-ment to 500 m (1,640 ft) at the South Fork em-bankment to 475 m (1,560 f0 northeast of ThylorCreek. Approximately one-third of this gradientis probably due to isostatic rebound since degla-ciation (Thorson 1981).

Most of the sediment exposed in roadcuts andstreambanks is gravel and sand, subrounded towell rounded, with better sorted sand layers andlenses throughout. Both median and maximumgrain sizes gradually decrease up the Middle andSouth Fork valleys, with pebble-poor lacustrine

Ice-marginalembankments 25I

silt deposits dominating beyond about 4 km fromthe ice-proximal face of the embankments. At theMiddle Fork embankment, this face is character-izedby ice-contact topography and moderately topoorly sorted gravel, sand, and silt and an ab-sence of recognizable lodgment till. A more com-plex record is revealed at the South Fork em-bankment (Fig. 58), perhaps owing to more ex-tensive exposure. Low on its ice-proximal face, -an area of fine-grained, horizontally bedded sandand silt is exposed (unit 1). Wood in this sedi-ment, in part converted to lignite, yielded a tnC

date of >50,000 years B.P. (Porter 1976; samplenumber UW-243). Drag folds and thrust faults, atan altitude of about 300 m (1,000 ft) near the topof the deposit, indicate shearing of beds to theeast, presumably by overriding ice. Unweathered(i.e. Vashon) till (unit 2, Fig. 58), which rests di-rectly on striated bedrock, is exposed only alongthe river topographically below the older sedi-mentary material. This till must have been depos-ited in a valley incised through pre-Vashon sedi-ment.

The uppermost sediment of this embankment(unit 3, Fig. 58) is part of an extensive reces-siohal outwash deposit over 100 m thick in thisarea. As Mackin (1941) first inferred, much ofthis fluvial sediment may have been deposited asdeltas into ice-marginal lakes. Deltaic prograda-tion, coupled with ice-margin retreat, resulted indeposition of deltaic and lacustrine sediments upthe Middle and South Fork valleys that mergewith a valley-train fluvial deposit southwest ofcedar Butte (Fig. 5A).

An analogous history is inferred for at leastone other embankment in this area. The morpho-logy of the Skykomish embankment (Fig. 6) iscontrolled by an extensive deposit of late-stagefluvial sediment that extends many kilometersupvalley. Poorly exposed beneath this material,however, is both a basal pebble-poor diamictonand a recessional moraine of Vashon age (Booth1e86).

P ilc huc k - S ultan emb ankment

Deposits constituting the embankment at thehead of the Pilchuck River (Figs. 7A and 7B) areexoosed in excavations associated with the con-struction and subsequent raising of adjacentCulmback Dam. They are augmented by exten-sive subsurface exploration (Converse Ward Da-vis Dixon 1979; Shaffer 1983). The basal em-bankment sediments are encountered in two drill

252 Derek B. Booth

holes near the downstream toe of the embank-ment. These sediments consist of more than 100m of fluvial sand and gravel that fill part of thebedrock valley of a pre-Vashon Sultan River be-neath the present embankment (unit 1, Fig. 78).The upper surface of the sand and gravel, at analtitude of about 410 m (1,350 ft), is 20-30 mabove the present altitude of the bedrock lipthrough which the Sultan River now drains intothe Sultan gorge. This upper surface may repre-sent a pre-Vashon-age floodplain of the SultanRiver, an interpretation that also implies drain-age through the Sultan gorge during pre-Vashontime. Alternatively, the surface may representthe level of proglacial fluvial deposition duringthe Vashon advance only (Shaffer 1983).

Bedded silt and pebbly silty diamicton (unit 2,Fig. 78) overlie the fluvial sediment. On thesouth dam abutment the unit is only 5-10 m thickand varies from rather typical massive and stonyVashon till (clast content about 25'/") to a poorlysorted, pebble-poor deposit containing discontin-uous silt layers 1-2 cm thick and a few gravel-richlenses. On the north abutment it thickens tonearly 100 m. In this area, clasts constitute lessthan 5"h of this unit (Fig. 7C); those present aresubangular to well rounded, 0.5-4 cm in dia-meter, and commonly striated. The matrix pri-marily consists of fine silt. Faint color changesand slightly coarser texture define subhorizontallayers in this silt up to l0 cm thick and severalmeters in lateral extent. Lenses of fine sand asmuch as I m thick are also present. Poor expo-sures farther east suggest that erosional remnants

BOREAS 15 (1986)

Fig. 6. Skykomish embankment (outlined).Pebble-poor diamicton is exposed in section 24 justsouth of the highway; the ridge in section 18 (alti-tude 610 m (2,001 ft)) is a bouldery moraine. To-pography from USGS Index 15' quadrangle. Ice-maximum limit shown by black dots; altitude infeet.

of this unit extend at least 3 km upvalley of themain embankment crest.

This unit is covered by bedded gravel and sand(unit 3, Fig. 78) that is at least 100 m thick on thenorthern part of the embankment. This fluvialsediment forms a nearly continuous kame terracethat begins at the 700-m (2,300-ft) altitude at abedrock notch on the north wall of the Pilchuckvalley, continues upvalley as a feature only a fewtens of meters wide, and broadens abruptly at theembankment itself. The gradient of the construc-tional surface is about 12 nlkm (65 ftlmi), whichprojects south to the altitude of a spillway at Ol-ney Pass (Fig. 7A).

Reconstruction of the ice-mareinalenvironmentIce-dammed lakes

The occupation of the Puget Lowland by the Cor-dilleran ice sheet required major readjustmentsin the prevailing westward and northwestwarddrainage out of the Cascade Range. Rivers werefirst diverted south along the advancing ice mar-gin. Eventually, the ice reached the eastern edgeof the Lowland, covering all ice-free routes. Gla-cial lakes then formed in the lower reaches of thewest-draining alpine valleys of the Cascades(Cary & Carlston 1937; Mackin 1941; Williams1971.; Porter 1976; Booth 1985, 1986).

Ice-marginal lakes of similar size and geometryare common in modern glacial environments

BOREAS 15 (1986) Ice-marginal embankments 253

48( '00'

12 40'2km

oF+lml

Unit 3bedded sandand grave' l70-

9Om

lO m

Uflit 2si l ty dianicton,

bedded sandand si l t

Unit Ibedded sandand qravel

Frg. Z. Pilchuck-sultan embankment. A: Topography from USGS Index and Silverton 15' quadrangles (embankment is outlined).Ice-maximum limit shown by black dots; altitude in feet. B: Stratigraphy of embankment sediments north of Culmback Darn(adapted from Shaffer 1983). C: Pebble-poor silty diamicton; zones of horizontally bedded silt are present above and below shovel(location shown on Fig. 7A; altitude 440 m (1,2140 ft)).

254 Derek B. Booth

(Stone 1963; Post & Mayo 1971). These lakestypically drain either through subaerial spillwaysupvalley of the ice or beneath the damming ice it-self (reports of long-term lake drainage over theice surface are notably rare).

In valleys adjacent to the Puget Lowland, bed-rock divides upvalley of the maximum ice marginstand well above the altitude of the damming ice.Thus, drainage of each lake could only be over,under, or around the ice dam (see below). Evenat lower ice stands, only the Sultan River valleyhad exposed a spillway over the bedrock divide(at Olney Pass; see Fig. 7A) that permitted sub-aerial lake drainage while ice still blocked thevalley mouths. All other valleys drained via theice dam for as long as active glacier ice remainedat their mouths.

Sedimentological processes in ice-marginalenvironments

The origin of deposits similar to the stratified,pebbly silt and diamict units in both the SouthFork Tolt and Pilchuck-Sultan embankments(unit 2, Figs. 4B and 78) has been discussed bymany authors (e.g. Dreimanis 1979; Evenson elal.1977;May 1977; Gibbard 1980;Eyles & Eyles1983; McCabe et al. 1984; Barnett 1985; Heiny1985). Common to all hypotheses is an inferredenvironment of subaqueous deposition, indicated in particular by laterally continuous, non-deformed stratification that presumably pre-cludes deposition (lodgment) by ice sliding overits bed.

All of the characteristics of the diamict units inthese Puget-Lowland embankments are compa-tible with, though not demanding of, sedimenta-tion processes characterized by fallout of debrisfrom the base of melting ice combined with sedi-ment influx from subglacial streams and the set-tling of suspended lacustrine sediments. The vari-able sediment load of ice and subglacial wateryields irregularly distributed pebble-poor layersand lenses within a more uniform lithology offine-grained suspended material. Tiaction cur-rents associated with water flow into or out of thelake locally produce stratified and sorted coarse-grained layers. The absence of sediment-flowfeatures rules out significant deposition by sub-aqueous flowtill (Evenson et al. 1977). The lackof overconsolidation or glaciodynamic structures(Dreimanis I976) near the top of most of thesedeposits implies that the ice may have onlygrounded rarely against these sediments (Gib-

BOREAS 15 (1986)

bard 1980; Eyles et a|.1982; Shaw 1982; Eyles &Eyles 1984).

Interaction of ice and water at the icemargin

Although the sedimentological evidence supportsthe existence of lacustrine environments adjacentto and beneath the ice sheet, these data alonecannot conclusively delineate this environment(cf. Karrow et al. 1984; Christe-Blick 1985). Fur-thermore, they offer little insight into the con-figuration of the ice sheet, the controls on lake-surface altitude, or the determination of whereand under what conditions sedimentation will oc-cur. These deficiencies can be corrected by ana-lyzing the mechanics of ice and water along a gla-cier margin. This analysis not only independentlyconfirms the presence of this inferred environ-ment, but also shows why embankments formedalmost exclusively at valley mouths, why they areubiquitous in this region, and whether they rep-resent climatically induced stillstands of the icemargin.

Ice limits. - Evidence for maximum extent of iceinto alpine valleys that enter the eastern PugetLowland is not well expressed on the steep valleysidewalls. The distribution of glacial erosion anderratics on the adjacent interfluves (Booth1986), however, indicates the vertical extent ofthese ice tongues. These data show that ice, atmaximum, stood from L00 to over 6(X) m abovethe altitude of the present embankment surfacesand in places nearly 700 m above the inferredVashon-age valley floor (Table 1). Embankmentaltitudes bear no systematic relationship to adja-cent ice-maximum altitudes, and thus embank-ments represent neither the position of maximumice advance (Knoll 1967) (because 600-m-thickice would have continued advancing beyondthem) nor a synchronous halt of the whole icefront during retreat.

Marginal lakes and subglacial hydrology. - Oncea marginal lake is impounded by ice, the lake'smaximum surface altitude equals the maximumhydraulic potential that its drainage route mustcross (Nye 1976). For subaerial drainage, this po-tential is simply the altitude of the spillway. Forsubglacial drainage paths, the hydraulic potentialhas two components: the bed altitude of thechannel and the pressure of the overlying ice(Shreve 1972). These are analogous to the eleva-

-j

' BOREAS 15 (1986) Ice-marginalembankments 255

Table l.Embankment altitudes, inferred glacial-age valley-floor altitudes, and reconstructed ice-surface altitudes in the Puget Low-

land.

Embankmentaltitude (m)r

Valley-floor altitude (m)'z Embankment Ice Iceheight3 altitudea thicknesss(m) (n) (m)

Pilchuck-Sultan 680/480Olney 510Wallace 800Skykomish 5401160Proctor 7ffiNorth Fork Tolt 550south Fork Tolt 610/600North Fork Snoqualmie 600/600Calligan Lake 7lOLake Hancock 700Middle-South Forks

Snoqualmie 490

10601050980860910880860850840830

780

3m 410 390370 510 490590 610 590720 130 130630 650 630430 460 450500 540 s00no 4% 390550 680 s90550 660 610

230 360 n0

290t902.4

2to410t30130100110/100zrcnf i12090

220

380/580ffi18032I/7n150330250t2ffi250t250130130

290

1 First number is the present embankment altitude. Second number is available data on the altitude of the highest non-fluvial sedi-

ment deposited subglacially and/or subaqueously.t Euid"nci for valley-floor altitude includes the top of bedrock or pre-Fraser material (minimum altitude) and the base of reces-

sional or postglaciil deposits (maximum altitude). The best estimate assumes a concave-upvalley profile and is probably accurate

in all cases to within 50 m.I Equals the difference between the embankment altitude and the best valley altitude'a Booth (1986)s Equals the difference between ice and embankment altitudes.

tion head and pressure head, respectively, com-mon in groundwater terminologY.

The hydraulic potential (or total head) at anypoint on the glacier bed can be expressed as analtitude, ignoring any dynamic component due tosliding of the ice (Booth 1984a):

H : zr + @,lq-)h, (1)

where 11 : the hydraulic potential, zr: bed alti-tude, p, : ice density, q, : water density, and ft: thickness of ice.

In these terms, .FI equals the altitude to whichwater would rise in a borehole that penetrates theglacier to its bed. The hydraulic potential in-creases with increasing ice thickness and de-creases with decreasing bed altitude. The gra-dient of H specifies the direction of water flow(i.e. from regions of high 11 to low 11). Note fromequation 1, however, that any change in 11 is pri-marily determined by changes in the altitude ofthe ice surface; bed altitude exerts an influenceonly one-eleventh as great (Shreve 1972,1985;Nye 1976). The hydraulic potential at the base ofthe ice can therefore be reconstructed (Fig. 8)from the reconstructed topography of the ice sur-f'ace and the glacier bed (approximately the pres-ent ground topographY).

The hydraulic potential beneath ice tonguesthat extended into alpine valleys in the study areathus increased upglacier (to the west and north-west, towards the center of the ice sheet). Thiswas the direction of greater ice-surface altitudes,more than compensating (via eqn. 1) for the des-cent of the valley floors in this (upglacier) direc-tion. Any potential route of subglacial drainagedown these valleys thus faced steadily rising hy'draulic potentials. At the edge of the Lowland,however, potential subglacial drainage routeswere not constrained by valley walls and so couldturn southward, in a direction of decreasing ice-surface altitudes, to follow paths where the hy-draulic potential decreased monotonically (cf.Fig. 8). A maximum value was therefore definedfor the potential along each complete drainageroute, between their'confined' west-flowing and'unconfined' south-flowing legs, which must havebeen exceeded before the water impounded in avalley could drain. The location of these maxi-mum potentials, near to where each valley en-tered the Lowland proper, defined the positionof an effective spillway for each lake.

Where subglacial lake drainage flowed oversuch a spillway upglacier from the snout of eachlocal ice tongue, some ice between the spillwayand the glacier margin probably began to float as

256 Derek B. Booth

Fig. 8. Contour map of subglacial hydraulic equipotentials cal-culated using equation 1, contour interval 15 m. Ground sur-face digit ized from 7.5' and 15' topographic maps; ice-surfacereconstruction from Booth (1984a: fig. 3.8). Embankment lo-cations are marked with open stars, subglacial spillways byheavy dots.

the altitude of the impounded lake approachedthe value of the spillway's hydraulic potential.Such lacustrine ice shelves have been observed insome areas (Nye 1976; Stone 1963; Post & Mayo1971, lakes no. 6 and 27; Sturm & Benson1985) though are absent or ill-defined in others(e.g. Holdsworth 1973; Bindschadler & Rasmus-sen 1983). An ice shelf will always tend to ad-vance into standing water because its unsup-ported portion above water level exerts a longi-tudinal stress greater than the water pressureacting on the submerged ice face (Weertman1957; Thomas 1973). The shelf thickness neces-sary for appreciable strain rates, observed in Ant-arctica to be 150-300 m (Robin 1979; Thomas1979), was exceeded during the Vashon stade inmost of the major alpine valleys entering the Pu-get Lowland (Table 1). Therefore, shelf ice adja-cent to the Puget lobe probably advanced beyondthe grounding line into these ice-marginal lakes.

BOREAS ls (1986)

Water flow into a marginal lake will eventuallyraise the water-surface altitude to equal the spill-way potential. If the spillway, and thus the drain-age, is subaerial over a bedrock divide, lake levelis stable at this point. If drainage is subglacial, amore complex sequence of events modeled byNye (1976) typically results in total or near-totaldrainage of the lake (Clarke 1982). Subglacialwater flow melts a tunnel, whose rate of expan-sion due to melting initially exceeds its closurerate due to ice deformation. The hydraulic poten-tial along the tunnel thus loses its componentfrom the ice overburden and drops precipitouslyto equal the bed altitude. Drainage should nowproceed catastrophically as a jcikulhlaup untillake level falls to the altitude of the highest bedspillway along the drainage route (2, in eqn. 1) oruntil ice finally reseals the tunnel during waningflow. Refilling of the lake begins the processanew (e.g. Stone 1963; Post & Mayo 1971; Mot-tershead & Collins 1976; Sturm & Benson1985). The lakes along the eastern margin of thePuget lobe probably refilled in a few tens of years

NORTH 4t" tsFORK TOLT

SOUTH FORK TOLT

CALLIGAN

H ANCOCK

ot*-----------ikt

MIDDLE-SOUTH FORKS. SNooUALMIE

Frg. 9. Boulder gravel deposited by episodic drainage out ofglacial Lake Skykomish (see Fig. 1 for location). Note person

a

Ice-marginal embankments 257BOREAS 15 (1986)

(Booth 1984a), and so this process repeated itself

numerous times during the Vashon stade' lne

oi"ai.t.O path of these outburst floods is etched

in the land'scape of the eastern Puget Lowland as

a 4O-km-long trough up to 1 km wide and over

ioo-r a".p [uootti 198'4b)' a portion-of .which is

depicted aiong the west edge of Fig' I I (now oc-

.rrpi"O Uy th-e North Fork of the Snoqualmie

niu.t;. Attnougtr largely swept clear."f t: i l-

ment, the trough is choked in one area by a 4uu-

r-,ni.t sectioi of fluvially transported bou-lder

gravel with clasts up to 3 m in diameter (Fig' 9)'" B""uur" no subierial drainage routes out of

the upper Cascade valleys existed- during tl9 V*

shon maximum, each ice-dammed lake in this re-

gion would have experienced subglacial drainage

E.rtlng to-" (usuatty extensive) period of .itsexiste"nce. Lake drainage over or around each ice

dam is precluded on both theoretical and em-

pirical grounds. Because water is more dense

ihun ice", its hydraulic potential will exceed that

of ice at the bed before the water surface ever

rises aboue the ice level (an obvious prerequisite

to supraglacial flow). Subaerial flow around an

ice dam -confined by steep valley walls (Mackin

1941) is equally implausible, requiring the.water

to flow .ottttuiy to the prevailing hydraulic gra-

dient (see below). These physical constralnts

are boine out by the absence of reported su-

praglacial lake-drainage systems on modern gla-

i i.r i. Ceotogic evidenie in the Puget Lowland is

equally in accord with these physical expecta-

tions; incised channels adjacent to each valley at

or near the maximum ice-margin position are no-

tably absent, implying a lack of significant sub-

aerial marginal drainage along the steep lnter-

fluves. Even at progressively lower ice stands'

o"fy n the Sultan River valley would subae.rial

patls have been uncovered (the farthest upvalley

teing Olney Pass; see Fig' 7A)' The rematrung

valleys drained subglacially for as long.as actrve

glacier ice remained in their lower reaches'

Sedimentation into ice-dammedmarginal lakesProcesses

The glaciolacustrine environment permits a vari-

ety o"f potential depositional processes' general-

#O h'fig. 10. These include basal meltout of

,"di*".tt 6eneath a floating shelf or berg' sub-

;;;;"t flowtill off the glacier snout' subglacial

GROUNDED ICE

Fie. 10. Possible ice-termini configurations and processes of till

deiosition in glaciolacustrine environments (adapted fromVor;

rei et ol. t983). More detailed discrimination of depositional

processes (e.g. Eyles & Eyles 1983: fig. 4) in the Plget-lobe

embankmenti is inhibited by insuffrcient exposure of the sedi-

ments themselves. With a stable gounding line beneath the ice

shelf depicted in the upper diagram, waterlain till will accumu-

late over time to form a ridge regardless of the topography of

the cross-hatched substrate.

ICE SHELF

grounding basall iner, - - t i l l

basal , t l t , 'Eneftout ' , i , ' t r i

ICEBERG

basal \ffiill]fr "tr ffi rli' i ! i,iilf,ii

258 Derek B. Booth

or supraglacial transport of sediment by runningwater into the lake, and settlement of suspendedlacustrine sediment (variously discussed singly orin combination by Rust & Romanelli 1975; May1977; Evenson et al. 1977; Fecht & Thllman 1978;Dreimanis 1979, 1982; Gibbard 1980; Hicock eral. 1978; Hillaire-Marcel et al. 1981; Orheim &Elverhoi 1981; Powell 1981; Eyles & Eyles 1983;Vorren et al. 1983; McCabe et al. 1984). Contin-ued aggradation, lowering of lake level, and/orexposure of a subaerial spillway will permit sub-sequent deltaic or subaerial fluvial sedimentationas well. These processes are not mutually exclu-sive; most studies have recognized their close as-sociation in even single exposures. The rates anddominant sites of sedimentation, however, maydiffer for each process.

Sedimentation by most subaqueous processesis concentrated at the grounding line (Orheim &Elverhoi 1981), because the glacier is the primarysource of sediment. Except in the presence of lo-calized fast currents near the ice margin, all butthe finest suspended sediment should be deposi-ted here (Edwards 1978; Eyles & Eyles 1983).The resulting deposit commonly forms shoals orridges beneath the ice at the grounding line.These linear ridges have been repeatedly ob-served or inferred beneath modern glaciers andice shelves (Holdsworth 1973; Barnett & Holds-worth 1974; Post 1975; Rust & Romanelli 1975;Edwards 1978; Hillaire-Marcel et al. 1981.; Powell1981; Sturm & Benson 1985).

Alternative processes that distribute subaque-ous sediment over a wider area ate likely to be in-effective in a confined ice-marginal lake. Becausethe debris content of ice tends to be highest at itsbase (Pessl & Frederick 1981), basal melting willyield the greatest volume of sediment near thegrounding line regardless of how far ice floats orextends into the basin (cf. Anderson et al. 1980).Based on study of modern Alaskan glaciers, thevolume of supraglacial debris available for moredistal transport is negligible by comparison to thebasal component (Evenson et al. 1985). Further-more, rafting by icebergs is restricted to thosemonths when the lake surface is unfrozen (Holds-worth 1973).

Sudden lowering of lake level during a periodof episodic drainage tends to destabilize an iceshelf and hastens its breakup into bergs (e.g. Post& Mayo 1971: lake no. 5 photo). Slower refillingof the lake would have a similar, though less pro-nounced effect (Holdsworth 1.973; Barnett &Holdsworth 1974). Such changes in ice configur-

BOREAS 15 (19861

ation upvalley of the grounding line, however,should have little effect on the primary pattern ofsedimentation, namely a pronounced decrease insedimentation away from the grounding line.These Puget-Lowland embankments thus reflectanticipated depositional processes focused nearthe grounding line of a glacier adjacent to a mar-ginal lake.

Location and stability of the grounding line

As the site of maximum subaqueous sedimenta-tion, the position of the grounding line controlsthe pattern of deposition. Formation of a discreteembankment requires that the glacier margin ter-minate in or against standing water and maintaina relatively stable grounding line for a consider-able time (Rust & Romanelli i975). A stablegrounding line requires, either singly or in combi-nation: (1) a long-term climatically controlledstillstand of the ice margin, coupled with either afixed water level or one that never rises highenough to float a portion of the ice; (2) presenceof a pre-existing topographic rise or shoal; (3) re-sistance to ice advance by confining topography;or (4) a process that incorporates the interde-pendence of the thickness of an ice dam with thewater depth in an impounded lake.

(l) Climatically controlled stillstand. - In the Pu-get Lowland, evidence does not seem to supportthis alternative (Thorson 1980; Booth 1986).Limiting radiocarbon dates on the Vashon stadefrom the central Lowland (Rigg & Gould 1957;Mullineaux et al. 1965) require rapid average ad-vance and retreat rates of the terminus of about100 m/a throughout the glaciation. Commensur-ate motion of the lateral ice margins would be ex-pected to have followed suit. A single stillstand ator near ice maximum (Mackin 1941) cannot ex-plain the formation of these embankments (Thble1) because the variation in ice thickness abovethem implies great variability in the length of theice-maximum tongue beyond them. No geologicevidence in the Lowland suggests the likelihoodof eleven such halts, needed to climatically sta-bilize each of the ice tongues at the present sitesof their respective embankments.

(2) Pre-existing shoals. - From the record of mul-tipfe glaciations in the Lowland (Armstrong et al.1965) and the ubiquitous presence of Vashon-ageembankments, pre-existing valley blockadeswere probably present when the Vashon ice sheet

BOREAS 1s (1986)

arrived. Older glacial sediments are indeed ob-served at the core of several of these embank-ments (e.g. Porter 1976; Figs. 48, 5C). This ex-planation, however, merely displaces back intime the question of why deposition is localized.It is also inapplicable for those embankmentswere deep exposures or drill cores reveal no suchmaterial.

The effect of shoals may nevertheless becomeimportant during the formation of a new em-bankment. Once sediment begins accumulating,the growth of an incipient embankment will tendto stabilize the grounding line (Post t975) andfurther localize deposition there.

(3) Confining topographic effects. - The positionof most of the embankments at valley mouthsempirically shows the importance of the valley-wall configuration. Its conceptually most simpleconsequence is to induce resistance to ice-frontadvance, thereby halting any advance of thegrounding line as well. As an ice tongue advancesinto a valley, it must thicken and steepen at thevalley entrance to compensate for the added lat-eral drag (Nye 1965). The convergent upvalleygeometry typical of most valleys amplifies this ef-fect (Mercer 1961; Funder 1972).It the valley isperched above the Lowland, such as those shownin Fig. 11, this pause will last even longer whileice thickens at the valley mouth. The second con-sequence of valley-wall configuration is reflectedindirectly in its contribution to the location of thesubglacial spillways (discussed below).

(4) Interdependence of ice thickness and waterdepth. - Although an ice margin and groundingline may stabilize under the special conditionsmentioned above, these conditions cannot ac-count for deposition of all embankments alongthe Cascade front, because they do not all satisfyone or more of the preceding requirements. Themorphological and sedimentological data fromthese embankments can be understood, however,by considering the mechanics and interaction ofice, subglacial topography, ice-marginal water,and transported sediment.

As thickening ice seals subaerial drainageroutes out of a basin, rising lake water, unavoid-able in a blocked basin, must eventually float theglacier snout prior to (or simultaneous with) lakedrainage. The location ofthe grounding line thenbecomes a function of the depth of impoundedwater and the hydraulic potential at the bed.Wherever the altitude of the lake surface equals

Fig. 11. Calligan and Hancock embmkments (outlined). To-pography from USGS Mount Si 15' quadrangle. Ice-maximumlimit shown by black dots; altitude in feet.

Wherever the altitude of the lake surface equalsor exceeds the hydraulic potential at the bed, theice floats. The exact extent and configuration ofthe lake ice is irrelevant, for although the icefront will advance as a shelf if it is sufficientlythick (Weertman 1974; Sanderson 1979), the lo-cation of the grounding line does not depend onthe position of the ice front.

Rising water can float ever-increasing thick-nesses of ice, causing the grounding line to mi-grate upglacier. As lake level approaches the hy-draulic potential of the subglacial spillway anddrainage becomes imminent, the grounding linelies across the ice tongue just downglacier of thisspillway (Fig. 12). Subsequent drainage may al-low the grounding line to migrate upvalley somedistance, but repeated cycles of deposition duringlake filling will tend to stabilize the ice front. Thisis partially due to both the effect of shoals (Post1975) and the shape of natural reservoirs. Waterlevel rises most slowly, and so the grounding linemigrates most gradually, when these lakes arefilled near to their maximum depths. Thus thegrounding line will remain longest at positions

,u\fuoe

4i.3

2km

260 Derek B. Booth BOREAS 15 (1986)

Aconto?;,1 ine

trI'3J;riff,]acier [r.Erl'?i;.iiff'1..grounding 1 ine and/or 1 akez

tq Ar:l H:',"llF [7'lsu'u?]ii'll,ni.Ice margi n

Subglacialsoi 1 I way (ep' isodic)

Flg. 12. Reconstruction of subglacial water-flow paths and spillway in a hypothetical valley dammed by active glacial ice. A: Sub-glacial bed topography; contour interval 100 m. B: Ice-surface topography; contour interval 10 m. C: Subglacial hydraulic equipo-tentials; contour interval 10 m. Grounding line for a given lake level defined by the position of the equipotential contour whosevalue equals the lake-surface altitude. The maximum lake altitude will be determined by the potential of the subglacial spitlway.Deposition of embankment sediments occurs in the shaded area, at greatest rates near to the maximum upglacier grounding-lineposition. Note gradient of the hydraulic potential adjacent to ice margin on the steep sideslopes; extensive marginal drainage notphysically possible in these areas. D: Cross-sectional view ofice dam, impounded lake, and subaqueous embankment sediment.Observed ranges of embankment heights and maximum ice thicknesses in the eastern Puget Lowland are shown (from Thble 1).

BOREAS 15 (1986)

closest to that of the spillway. Continued sedi-mentation there will tend to stabilize that lo-cation.

Subglacial spillways are located in the Lowlandjust south of the mouth of most valleys (Figs. 8and 1.2). This geometry focuses subaqueous de-position at the valley mouths and explains thetendency for embankments to form in these posi-tions. These spillway locations, and thus thegroundingJine positions, are quite insensitive tochanges in the ice thickness (which can be easilyif somewhat tediously checked by alternative re-constructions with thinner ice using eqn. 1). De-position during advance, maximum, and retreatstages therefore will be at virtually the same lo-cation.

The Sultan River, by contrast, has a spillwaythat is not in the Lowland proper (Fig. 7A). Theembankment in this valley also occupies an atyp-ical position. As predicted by this analysis, thediamicton (unit 2 of Fig. 78) of this embankmentin the main Pilchuck-Sultan valley was depositedat a point just upvalley of where the subglacialpotential (in the main valley) equals the potentialof the spillway (located in the Sultan gorge; cf.Fig. 7A).

Other models

The physical processes described here are similarto those invoked by Hillaire-Marcel et al. (1'981)for formation of 're-equilibration moraines', lo-calized deposits at ice-water margins during uni-form retreat of the eastern Laurentide ice sheet.Their model implicitly assumes sedimentationonly during periods when' the ice is fullygrounded in shallow water, with neither a float-ing nor an actively calving margin. The processesthat actively contribute sediments to this environ-ment, however, will be active at the groundingline regardless of the configuration of the ice far-ther downglacier. This resolves their acknowl-edged difficulty (Hillaire-Marcel et al. 1981':212)in explaining large sediment volumes, hypoth-esized to accumulate only during the few yearsthat the ice front actually terminates at a specificlocality.

Other settings proposed for deposition of flu-vial and ice-contact sediment in other localitiesare clearly not applicable to the initial formationof the Puget embankments. Deltaic sedimenta-tion into a proglacial lake, described by Orom-belli & Gnaccolini (1978), lacks a significantbasal-meltout component because water depth in

Ice-marginal embankments 261

their example was controlled by a subaerial spill-way at an altitude too low to float the ice. Thisprocess, however, was active following the ice-dammed stage of deposition at several of the Pu-get embankments (e.g. Pilchuck-Sultan, Sky-komish, North Fork Tolt, and North Fork Sno-qualmie). as reflected by planar topset surfacesand rarely observed upvalley-dipping forsets.Other generalized ice-contact lacustrine environ- -ments, such as described by Koteff (197a), allhave subaerial spillways distinct from the dam-ming ice. The location of deposition in such casesis therefore determined by the position of the icefront itself during temporary stillstands. Thiscontrasts with the Puget embankments, whose lo-cations were determined by a grounding linefixed by the pattern of subglacial hydraulic po-tentials and resultant subglacial spillways. Al-though focused on a particular geographic re-gion, this analysis should apply to any localitywhere modern or Pleistocene glaciers have termi-nated against ice-dammed, subglacially drainingwater bodies.

Acknowledgments. - The field data for this study were collected

as part of a regional mapping project for the U.S. Geological

Survey (Booth 1986; Frizzell et al. 1984). I would like to par-

ticularly thank Stephen C. Porter and Fred Pessl, Jr., who led

me to more careful consideration of many of the ideas pre-

sented here. and Bernard Hallet, Mark E. Shaffer, Rowland

W. Tabor, Richard B. Waitt, Jr., Charles F. Raymond, and

Garry K. C. Clarke for invaluable discussions. B. Hallet, S. C.

Porter, J. J. Clague, and M. E. Shaffer crit ically reviewed the

manuscnpt.

ReferencesAnderson. J. B. , Kurtz, D. D., Domack, E. W. & Balshaw, K.

M. 19U0: Glacial and glacial marine sediments of the Ant-

arctic continental shelf. "/. Geol. 88,399-414.Armstrong, J. E. , Crandel l , D. R., Easterbrook, D. J. & No-

ble. J. B. 1965: Late Pleistocene stratigraphy and chronologyin southwestern Brit ish Columbia and northwestern Wash-ington. Geol. Soc. Am. Bull.76,321-330.

Barnett. D. M. & Holdsworth, G. L974: Origin, morphology,and chronology of sublacustrine moraines, Generator Lake,Baffin Island, Northwest Territories. Canada. Can. J. Earth

Sci. 11. 380-408.Barnett. P. J. 19tl-5: A model of glacial-glaciolacustrine sedi-

mentation, Lake Erie basin. Geol. Soc. Am., Abstr. with

Progr. 17(5),279.Bindschadler, R. A. & Rasmussen, L. A. 1983: Finite-differ-

ence model predictions of the drastic retreat of ColumbiaGlacier, Alaska. U.S. Geological Suruey Professional Paper1258-D. 17 pp.

Booth, D. B. 1984a: Glacier dynamics and the formation of gla-

cial landforms in the eastern Puget Lowland, Washington.Seattle, University of Washington, Ph.D. thesis. 217 pp.

262 Derek B. Booth

Booth, D. B. 1984b: Ice-sheet reconstruction and erosion bySubglacial meltwater in the eastern Puget l,owland, Wash-ington. Geol. Soc. Am., Abstr. with Progr. 16(6), 450.

Booth, D. B. 1985: Surficial geology of the Granite Falls 15-minute quadrangle, Snohomish County, Washington. U.S.Geological Survey Open-File Report 85-517, scale 1:50,000.

Booth, D. B. 1986: Surficial geology of the Skykomish andSnoqualmie Rivers area, Snohomish and King Counties,Washington. U.S. Geological Survey Miscellaneorc Investi-gations Map I - 1745, scale 1:50,000.

Cary. A. S. & Carlston, C. W. 1937: Notes on Vashon stage

glaciation of the South Fork of the Skykomish River valley,

Washington. Northwesl Science I I, 6l-62.

Christe-Blick, N. 1985: Depositional processes and environ-

ments inferred from glacial rocks: interpretive ambiguity, ob-

servational inadequacy, inappropriate models and strategies

for future research. Geol. Soc. Am., Abstr. with Progr.

r7(5),283.Clarke. G. K. C. 1982: Glacier outburst f loods from 'Hazard

Lake', Yukon Territory, and the problem of f lood magnitudeprediction. J. Glac. 28,3-21.

Converse Ward Davis Dixon 1979: Final geotechnical invest!gation. raising of Culmback Dam, Sultan River Project,

Stage ll. Unpublished report for Public Util i ty District No. 1

of Snohomish County, Washington.

Crandell. D. R. 1963: Surficial geology and geomorphology of

the Lake Tapps quadrangle, Washington. U.S. GeologicalSurvev Professional Paper 388A. 84 pp.

Dethier. D. P.. Safioles, S. A. & Pevear, D. R. 1981: Com-position of t i l l from the Clear Lake quadrangle, Skagit and

Snohomish Couoties. Washington. U.S. Geological Survey

Open-File Report 8l-517. 55 pp.

Dreimanis. A. 1976: Til ls: their origin and properties. 1n Leg-get. R. F. (ed.): Glacial Tills, an Interdisciplinary Study, 1l-

49. Ottawa. Ontario.Dreimanis. A. 1979: The problem of waterlain ti l ls. In Schluch-

ter. C. (ed.): Moraines and Varves, 167-177. A. Balkema,

R0ttcrdam.Dreimanis. A. 1982: Two origins of the stratif ied Catfish Creek

Til l of Plum Point. Ontario, Canada. Boreas 11, 173-180.

Edwards. M. B. 1978: Glacial environments. 1n Reading. H.

G. (ed.): Sedimentary Environments and Facies, 416_4'38.

Blackwell. Oxford.Evcnson. E. B.. Cl inch, J. M. & Stephens. G. C. 1985: Debr is

delivery. deposition and reworking at Alaskan glacier mar-gins. Geol. Soc. Am. Abstr. with Progr. 17(5), 286.

Evenson. E. B., Dreimanis, A. & Newsome, J. W. 1977: Sub-

aquatic t low ti l ls: a new interpretation for the genesis of some

laminated ti l l deposits. Boreas 6, 115-133.

Eyles. C. H. & Eyles. N. 1983: Sedimentation in a large lake: a

reinterpretation of the Late Pleistocene stratigraphy at Scar-

borough Bluffs. Ontario. Canada. Geology 11,146-152.

Eyles. C. H. & Eyles. N. 1984: Glaciomarine sediments of the

lslc of Man as a key to late Pleistocene stratigraphic investi-

gations in the Irish Sea Basin. Geology 12,359*364.

Eyles. N.. Sladen. J. A. & Gilroy, S. 1982: A depositional mo-

del for stratigraphic complexes and facies superimposition in

lodgement tills. Boreas 1 1, 317-333.Fecht, K. R. & Tallman, A. M. 1978: Bergmounds along the

western margin of the channeled scablands, south-central

Washington. Geol. Soc. Am., Abstr. with Progr. 10(7),400.

Fr izzel l .Y. A. . Jr . . Thbor. R. W., Booth, D. B. & Wait t , R.

B.. Jr. 1984: Preliminary geologic map of the Snoqualmie

I : 100,000 quadrangle. Washington. U. S. Geological Survey

Open-File Report 84493.Funder. S- 1972: Deglaciation of the Scoresby Sundfjord re-

gion. northeast Greenland. /n Price, R. J. & Sugden, D. E.

BOREAS ls (1986)

(eds.): Polar geomorphology. Inst. Brit. Geogr. Spec. pub. 4,3HZ.

Gibbard, P. 1980: The origin of stratified Catfish Creek till bybasal meft ing. Boreas Q,7l--85.

Heiny. J. S. 1985: Glaciogenic sediment sequences and inferredice margin fluctuations in the Lake Michigan basin- Geol-Soc. Am., Abstr. with Progr. 17(5),292.

Hicock. S. R., Dreimanis, A. & Brosten, B. E. 1981: Subma-rine flowtil ls at Victoria, Brit ish Columbia. Can. J. Earth Sci/8, 7l-80.

Hil laire-Marcel, C., Occhietti, S. & Vincent, J. S. 19g1: Sa-kami moraine, Quebec: a 50O-km-long moraine without cli-matic control. Geology 9,210-214.

Hirsch. R. M. 1975: Glacial geology and geomorphology of theupper Cedar River watershed, Cascade Range, WashingtonSeattle. University of Washington, M.S. thesis. 48 pp.

Holdsworth. G. 1973: Ice calving into the proglacial GeneratorLake. Baffin Island, Northwest Territories, Canada. J- Glac.r2.235-250.

Karrow, P. F., Dreimanis, A., Sharpe, D. R., Gravenor, C. P.,Eyles, C. H. & Eyles, N. 1984: Comment and reply on'Sedi-mentation in a large lake: A reinterpretation of the late Pleis-tocene stratigraphy at Scarborough Bluffs, Ontario, Canada.'Ceology 12, 185-190.

Knoll. K. M. 1967: Surficial geology of the Tolt River area,Washington. Seattle, University of Washington, M.S. thesis.9l pp.

Koteff, C. 1974: The morphologic sequence concept and degla-ciation of southern New England. /n Coates, D. R. (ed.):Clacial Geomorphology, 121-144. State University of NewYork. Binghampton.

Kulig. J. J. 1985: Detailed sedimentological analysis of glaci-genic sediments deposited in central Alberta. Geol. Soc.Am., Abstr. with Progr. l7(5),297.

Mackin. J. H. t94l: Glacial geology of the Snoqualmie-Cedararea. Washington. J. Geol. 49, 449481.

May. R. W. 1977: Facies model for sedimentation in the glacio-lacustrine environment. Boreas 6. 175-180.

McCabe, A. M.. Dardis. G. F. & Hanvey, P. M. 1984: Sedi-mentology of a late Pleistocene submarine-moraine complex,County Down, Northern lreland. J. Sed. Pet. 54,716-730.

Mercer, J. H. l96l: The response of fjord glaciers to changes inthe firn limit. L CIac. 3,850-858.

Miall, A. D. 1977: A review of the braided river depositionalenvironment. Earth Sci. Rev. 13,142.

Mottershead, D. N. & Collin, R. L. 1976: A study of glacier-dammed lakes over 75 years - Brimkjelen, southern Norway.J. Clac. 17.491-505.

Mullineaux. D. R. 1970: Geology of the Renton, Auburn, andBlack Diamond quadrangles. King County, Washington.U.S. Geologkal Survey Professional Paper 672.92 pp.

Mullineaux, D. R., Waldron, H. H. & Rubin, M. 1965: Stra-tigraphy and chronology of late interglacial and early Vashonglacial time in the Seattle area, Washington. U.S. GeologicalSurvey Bulletin 11944. lO pp.

Nye, J. F. 1965: The flow of a glacier in a channel of rec-tangular. elliptical, or parabolic cross-section. J- Glac. 5,661-690.

Nye. J. F. 1976: Water flow in glaciers: jokulhlaups, tunnels,and veins. l. Glac. 17,181-207.

Orheim. O. & Elverhoi, A. 1981: Model for submarine glacialdeposition. Annals of Glaciology 2, 123-128.

Orombelli, G. & Gnaccolini, M. 1978: Sedimentation in pro-glacial lakes: a Wurmian example from the Italian Alps. Z.Geomorph. N. F. 22, 417-425.

Pessl. F., Jr. & Frederick. J. E. 1981: Sediment source for melt-water deposits. Annals of Glaciology 2, 92-96.

BOREAS 15 (1986)

l)orter. S. C. 1976: Plcistrxcne glaciation in the soulhcrn partof thc North Cascade Range. Washington. Ceo!. Soc. Am.Bullcrin U7.6l-75.

Post. A. S. t975: Preliminary hydrography and historic termi-nal chungcs of Columbia Clacicr. Alaska. U.S. GeologicalSurrct Hvdrologic Inrcstigations Atlas 559. 3 sheets.

Posl . A. S. & Mayo. L. R. l97l : Clacier dammed lakes andoutburst floods in Alaska. U.S. Gcological Surve-v H;"dro"logic lnrestigarions Arlas 155. lll pp,

Poucll. R. D. l9til: A model for scdimentation by tidcwatergfacicrs. ; rncls of Glociologl' 2, 129-134.

Rigg. G. B. & Gould. H. R. 1957: Age of Clacier Peak erup-tion and chronology of postglacial pcal dcposils in Washing-ton and surrounding areas. Am. J. $ci.255. 341-363.

Robin. C. de Q. 1979: Formation. flow. and disintegralion ofice shefves. l. Gloc. 24.259-271.

Rust. B. R. & Romanel l i . R. l9?5: Late Quaternary suhaque-ous oulwash dcposits ncar Ottawa. Canada, /, Jopling. A.V. & McDonald. B. C. (eds-): Glaciofluvial and glaciola"cuslrinc sedimcnlation. Soc. Econ. Paleo. Min. Spec. Publ.2-t. t77-192.

Sandcnon. T. J. O. 1979: Equilibrium profile of ice shelves. J.Clac. 22.43-5*460.

Shaffcr. M. E. l9tl3: Repon on geohydrologic investigation ofPilcbuck Plug. raising of Culmback Dam, Sultan River Pro-jcct. Stagc ll. Scatlle. Convcrse Consultants. lnc., unpub-lishcd report for R. W. Beck and Associales and Public Util-it1, District No. I of Snohomish County. Washington. 54 pp.

Shaq'. J. l9ti2: Mcltout till in the Edmonton ar€a. Alberta.Canada. Can. t. Earrh Sci. /9, l-5411-1569.

Shreve. R. L. 1972: Movemcnt of $,ater in slaciers. "1. Glac. I I,l(1.5-214.

l\i A) BOREAS BOOK REVIEWS\\-</t7

Boreas, Yol. 15. pp. 26T264. Oslo, 1986 09 0l

Ice age mammals and environments

MATTI ERONEN

Sutcliffe. Antony J. 1985: On the Tiack of lce Age Mamnab.224 pp..Brirish Museum (Natural History). Price GBP 12.95'

This book is a comprehensive treatment of the biological andphysical developments during the Quaternary period. Althoughthe lce Age mammalian fauna referred to in the title fiSureprominently rhroughout the entire book, many different as-pects of environmental changes during the Quaternary time arealso discussed. Both the title and tbe aPpearanc€ of the bookare 'selling': on the dust jaclet, for example, there is a col-ourful picture of an lce Age landscape with a herd of woollymammoths. Five such colour plates depicting different Pleisto-ccne sccnes can be found inside the book, which as a whole is

Ice-marginal embankments 263

Shreve. R. L. l9li5: Eskcr characleristics in tcrms of glacierphysics. Katahdin esker sl'slcm. Mainc. Geol. Soc. Ant. Bull.9(). 639446.

Stonc. K. H. 1963: Alaskan ice-dammcd lakcs. Assoc. ofAnrcrican Geogr., Annals J.l, 332-3.19.

Sturm, M. & Benson, C. S. 1985: A history of jSkulhlaups fromStrandline Lake, Alaska. l. Glac.3l,27?-28o.

Thomas. R. H. t973: The crecp of ice shclves: thcory. J. C/oc.,2. 4.s-53.

Thomas. R. H. 1979: The dynamics of marinc ice shccts. J:Glac. 24. 167-177 .

Thorson. R. M. l9ft0: lce shcct glaciation ofthe Puget hrrvland.Washington. during the Vashon stade. Quat. Rcs. I'1, tt13'321.

Thorson. R. M. l9lll: Isostatic effccts of the last glaciation inthe Pugel lowland. Washington. U.S. Geological SurvevOpen-Filc Report 81-370. l|.X..l pp.

Vrrren. T. O.. Hald. M.. Edvardsen. M. & Lind-Hanscn, O.W. 1983: Glacigcnic sediments and sedimenlary cnviron'mcnts of continental shelves: general principles with a casestudy from the Norwegian shelf. /n Ehlcn. J. (ed.l; GlacialDcposits in North-west Europc. 6l-73- A. Balkcma, Rot-lerdam.

Wecnman. J. 1957: Dcformation of floaring ice shelves' J.GIac. 3.3tA2.

Weertman. J. 1974: Stability of rhe junction of an ice sheel andan ice shelf. J. Glac. 13. Tll.

Williams. V. S. l97l: Glacial geology of the drainage basin ofthc Middlc Fork of the Snoqualmie River. Scattle. Universit)of Washington. M.S. thesis. 4-5 pp.

well illustrated with many black-and-*'hite pholographs, draw-ings, maps, and diagrams. The book is aimed primarily at sPe-cialists working on different fields related 1o Quaternary re-search, but it should also appeal to a wider audience-

Although Antony J. Sutcliffe is an expert in mammalian pal-aeonrology, he skilfulty provides an oven'ieu of all the mainlines of modern Ouaternary research in this book. He openswith a discussion of Quatemary climatic variations and theirconsequences and follows with a fascinating description of theearly interpretations of fossil remains. For example, earlyworkers identified animal bones unearthed in rhe Past cenluriesas remains of such fanciful creatures as dragons and unicorns-The next chapler is concerned with our prescnt knowledgc of