incipienttunnelchannels - university of wisconsin–eau claire of...bdepartment of geosciences,...

Quaternary International 90 (2002) 41–56

Incipient tunnel channels

Darren B. Sjogrena,*, Timothy G. Fisherb, Lawrence D. Taylorc, Harry M. Jold,Mandy J. Munro-Stasiuke

aEarth Science Program, University of Calgary, Calgary, Canada AB T2N 1N4bDepartment of Geosciences, Indiana University Northwest, 3400 Broadway, Gary, IN 46408, USA

cDepartment of Geology, Albion College, Albion, MI 49224, USAdDepartment of Geography, University of WisconsinFEau Claire, Eau Claire, WI 54702, USA

eDepartment of Geography, Kent State University, Kent, OH 44242-0001, USA

Abstract

Glaciated terrains in east-central Alberta and south-central Michigan contain channels that have hummocks and transverse ridges

separating depressions along their floors. This association imparts a linked pothole appearance. Similar channels are ofteninterpreted as tunnel channels or subaerial channels, partly filled with sediment from a subsequent glacial advance, a stagnating iceroof, or slumped sediment from the channel margins. However, the truncation of sedimentary packages in the channel walls and

intrachannel hummocks indicates that they are erosional landforms, cut into glacial sediments (till), bedrock, or gravel. Eskersoverlie and are found within a few channels, indicating that these channels formed before the final stagnation that produced theeskers. These two characteristics, combined with the observation that many channels have convex-up long profiles, indicate that the

channels were eroded by pressurized, subglacial water. Because the formative mechanisms for this type of channel are not clear, andmodern environments that could produce this type of landform are inaccessible, we draw on several morphologic analogues topropose mechanisms for channel erosion. We conclude that the erosion of these linked pothole channels (incipient tunnel channels)was the product of the complex interaction between complex turbulent flow structures and various scales of roughness elements.

r 2002 Elsevier Science Ltd and INQUA. All rights reserved.

1. Introduction

Decay of the Pleistocene ice sheets was characterizedby the release of tremendous volumes of meltwater in arelatively short time (e.g., Teller, 1987, 1990). Recon-structions of the timing and style of meltwater releaseare often based on the configuration of glaciofluviallandforms (e.g., Bretz, 1943). Complex glaciofluvialassemblages make it difficult to differentiate betweenpreglacial, subglacial, ice-marginal, and proglacialenvironments; made more difficult by the likelihoodthat channels may be polygenetic (Sjogren, 1999) andrelate to earlier glacial episodes (Klassen, 1989).

Many meltwater channels are closely associated withzones of hummocky terrain, commonly presumed to becomposed of supraglacial debris released from stagnat-ing glacier ice (Gravenor and Kupsch, 1959), although

other hypotheses have been proposed (Hoppe, 1952;Stalker, 1960; Holden, 1993; Rains et al., 1993; Munroand Shaw, 1997; Eyles et al., 1999). Glaciofluvialchannels, associated with hummocky terrain, are inter-preted to have formed prior to the hummocky terrain(i.e., predating the advance that deposited the till), inice-walled stream trenches (Gravenor and Bayrock,1956; Parizek, 1969), as glacial spillways (Kehew,1982), or as subglacial channels (Evans and Campbell,1995; Clayton et al., 1999; Sjogren, 1999; Evans, 2000).Occasionally, it is necessary to invoke multiple glacialadvances to explain the landforms and stratigraphicassociations (e.g., Bayrock, 1958; Kozlowski, 1999).

This paper focuses on the genetic linkages betweentunnel channels and hummocky terrain. We use the termtunnel channel to mean a linear depression or series oflinked depressions formed by subglacial water flowingfrom bank to bank. A tunnel valley is formed by asubglacial river that is narrower than the valley itself(Clayton et al., 1999). Hummocky moraine is usuallydefined as tracts of hillocks and depressions formedduring the latest stages of ice stagnation as debris is

*Corresponding author.

E-mail addresses: [email protected] (D.B. Sjogren), tgfisher@

iun.edu (T.G. Fisher), [email protected] (L.D. Taylor), jolhm@

uwec.edu (H.M. Jol), [email protected] (M.J. Munro-Stasiuk).

1040-6182/02/$ - see front matter r 2002 Elsevier Science Ltd and INQUA. All rights reserved.

PII: S 1 0 4 0 - 6 1 8 2 ( 0 1 ) 0 0 0 9 1 - X

lowered from ice onto the ground (e.g., Gravenor,1955). Because many hypotheses for the formation ofhummocky landforms do not include lowering of debris,we prefer to use the nongenetic term hummocky terrainto describe tracts of hummocky landforms (Munro andShaw, 1997).

In this paper we present morphologic and sedimento-logic evidence from two glaciofluvial channel assem-blages that indicates that they were part of a continuumof erosional landforms. Hummocky terrain representsthe initial erosion by broad, turbulent, subglacialmeltwater flows (Munro and Shaw, 1997), whereaswell-defined tunnel channels represent a later, channe-

lizing phase. Between these two end members is amyriad of channel forms, including aligned depressionsand channels with hummocks on their base. On the basisof these end members, intermediate forms are inter-preted as incipient tunnel channels in various stages ofdevelopment. We conclude that the landforms presentedin this paper are erosional, but fully acknowledge thatnot all hummocky terrain or channels containinghummocks were necessarily formed in this way.

We focus on two study areas affected by theLaurentide Ice Sheet (LIS) (Fig. 1). The first area islocated in east-central Alberta, Canada, within what iscommonly referred to as the Viking moraine (Warren,

Fig. 1. (A) Location of study areas in east-central Alberta and south-central Michigan in relation to the margin of the Laurentide ice sheet. Source:

Dyke and Prest (1987). (B) Configuration of major hummocky zones (moraines) in the Alberta study area. Modified from Shetsen (1990). (C)

Configuration of moraines in the Michigan study area. Modified from Farrand (1982).

D.B. Sjogren et al. / Quaternary International 90 (2002) 41–5642

1937) (Fig. 1B). The second area is located in Michigan,along what has been mapped as recessional moraines ofthe retreating Saginaw Lobe (Fisher and Taylor, 1999,2002) (Fig. 1C).

2. Setting

2.1. East-central alberta

The northern part of the Viking moraine has beeninterpreted as a terminal (Warren, 1937; Bretz, 1943) ora recessional (Klassen, 1989) moraine of the LIS. Thestudy area is located in hummocky terrain superposedon a broad bedrock ridge (Viking upland) that was apreglacial drainage divide (Pawlowicz and Fenton,1995). Tributaries to the preglacial rivers partly dis-sected this upland, creating undulating topography.Quaternary sediments mantling these subtle undulationsare up to 20m thick, although isolated thicker sedimentsare present in some areas.

The Viking moraine is dominated by hummockyterrain and is flanked on the west by undulating terrainlargely devoid of Quaternary sediments (Figs. 1B and 2).In some re-entrants along this transition, the presence ofisolated remnants of glacial sediment, flanked by UpperCretaceous bedrock, indicates that these sediments havebeen eroded. Meltwater channels are present in areaswith thick or thin Quaternary sediments (Sjogren, 1999).The morphologic expression and sediments of theglaciofluvial channels are closely related to the topo-graphic setting, determined largely by bedrock, and tothickness of surficial sediment (Gravenor and Bayrock,1956; Sjogren, 1999). Channels on topographically highareas tend to be less distinct than those in the lowlands.West of the Kinsella Glaciofluvial Complex (Fig. 2),indistinct channels are associated with moraine pla-teaux. Locally, lowlands coincide with relatively thinsurficial sediments.

2.2. South-central Michigan

The study area in south-central Michigan includes aseries of landforms interpreted as recessional morainesof the retreating Saginaw Lobe of the LIS (Zumberge,1960; Farrand, 1982) or as moraines deposited during areadvance (Kozlowski, 1999). Fisher and Taylor (1999,2000, this volume), in contrast, suggest that some ofthese moraines may be advance moraines, or representolder ice advances, because tunnel channels anddrumlins are continuous across them (Figs. 1C and 3),and appear to have formed synchronously (Kozlowski,1999). A regional glacial geomorphology map shows thedissected nature of the landscape (Fig. 3). The uplandareas have steep, streamlined proximal slopes thatcontain numerous drumlins and re-entrants. The

sculpted nature of the northeast-facing escarpmentsindicates that the upland areas predate the sculptingevent.

3. Channel characteristics

3.1. Channel pattern and topographic context

In the Alberta study area, meltwater channels arearranged in a sub-parallel pattern (Fig. 2), which persistseven where the channels cross the Viking upland.Channels extending southward from the Birch Lakedepression are deflected slightly as they cross the upland(Fig. 4A). Channels truncate subtle ridges that aretransverse to channel axes. These ridges consist of

Fig. 2. Major geomorphic elements of the Alberta study area. Tunnel

channels cross high ground dominated by hummocky terrain. A

discontinuous subparallel pattern is deflected to the southeast south of

the Kinsella Glaciofluvial Complex.

D.B. Sjogren et al. / Quaternary International 90 (2002) 41–56 43

in situ bedrock or glaciotectonized sediments (Sjogren,1999). Where channels cross the broad, bedrock-controlled upland, the long profiles are convex-up(Fig. 4B). Channels connect at a variety of angles,imparting ‘‘confused’’ anabranching to an overallparallel pattern. Smaller, connecting segments aresometimes perched above the main channels. Minor,blind-ended channels, oriented parallel to the mainchannel assemblage, are also present. South of the

Kinsella Glaciofluvial Complex, channels are orientednorthwest to southeast (Figs. 2 and 5). They also have asub-parallel pattern with interconnections occurring atvarious angles. Some channel thalwegs have convex-uplong profiles, but such profiles are not as pronounced asthose to the north.

Glaciofluvial channels in the Michigan study areahave a more pronounced anabranching pattern. Larger,well-defined channels delineate the upland areas (Fig. 3).

Fig. 4. (A) Hillshaded relief map of channels in the northern part of the Alberta study area. Minor deflections that occur as channels encounter

subtle transverse ridges. (B) Topographic context of the channels in (A). Channel orientation has been altered slightly as they cross the bedrock

controlled upland. Convex-up thalwegs indicated by the arrow. Data source: 1 : 20 000 Alberta Environmental Protection digital elevation models

with 25m sampling.

Fig. 3. Major geomorphic elements of the Michigan study area. Upland areas contain drumlins and are sculpted on the upflow side. An

anabranching pattern of the tunnel channels dominates. Discontinuities and minor deflections in channel orientation occur where channels partly

dissect upland regions.


Smaller channels bifurcate from these larger channelsand cross the uplands. These smaller channels arecommonly discontinuous or indistinct (Fig. 3) and manyhave convex-up or undulating long profiles (Figs. 6 and7). Isolated channels on some of the upland areas areblind-ended (Fig. 3). In some areas, smaller channelsdissect or delineate the hummocks (arrow on Fig. 7A),whereas in other areas, upland channels truncate bothtransverse linear elements and drumlins (compareFigs. 3 and 6). In both study areas channels cross overpresent drainage divides and are less distinct near thosedivides.

3.2. Channel morphology

Channels in the two study areas have a wide rangeof sizes and shapes. Channels in the Viking area are0.2–1 km wide and up to 30m deep (Figs. 2, 4 and 5),comparable in width to the smallest channels that

delineate hummocks in the Michigan study area. Thelargest Michigan channels are up to 4 km wide and 50mdeep and contain outwash of unknown depth. Smallerchannels typically range from 0.5 to 2 km wide and are15–20m deep. Although individual channels vary inshape, the connectivity between them indicates con-temporaneous formation.

To facilitate descriptions of channel morphology, weidentify three basic channel types. Type 1 ‘channels’,which are best developed in the Alberta study area(Fig. 4), do not possess an obvious channel shape, butrather consist of aligned trains of depressions that areconsiderably deeper than depressions in the adjacenthummocky terrain (Fig. 8A). They are commonlydiscontinuous, situated on locally high ground, andare flanked by high-relief hummocky terrain (Fig. 9A).Depressions range from circular to elliptical in plan viewand are separated by hummocks or transverse ridges.Hummocks within the channels have tops that are app-roximately the same height as those adjacent to the cha-nnels. Most hummocks are composed of till, with minorcomponents of gravel or bedrock. In the Michiganstudy area Type 1 channels are wider than in Albertaand have better defined margins (Figs. 6 and 7).

Type 2 channels are crenulated in plan view, withpronounced expansions and constrictions along theirlengths, giving them a beaded shape (Fig. 8B). Theyresemble linked potholes with lower transverse ridges(Fig. 8B) or symmetrical hummocks along their base(Fig. 9B). The lateral margins of some intrachannelhummocks are separated from surrounding hummockyterrain by a low, narrow ridge oriented parallel to thechannel axis (Fig. 10). Depressions within the channelsare larger than those in Type 1 channels and arecommonly flat-bottomed. This characteristic is particu-larly conspicuous in the Michigan study area (Fig. 6),where steep margins appear to be lateral expansions ofthe more gently sloped margins of Type 1 channeldepressions. Type 2 channels are found both in areas ofthick till and thin, discontinuous till. The relief betweenthe channel bottoms and margins is subtler in areas witha thin till cover. Type 2 channels in the Michigan studyarea dissect the upland surfaces, whereas in the Albertastudy area they are found in both upland and lowlandareas. The Michigan channels are not as well developedas the Alberta examples.

Type 3 channels have sharply defined margins thatlack the pronounced bulbous shape characteristic of thefirst two types, although constrictions and expansionsare common. They may contain hummocks, whichoccur in clusters or are streamlined. In the Alberta studyarea Type 3 channels may contain isolated, low-amplitude (1–3m) hummocks. These channels typicallyhave a flat bottom (Figs. 8C and 9C) with rare convex-up segments. Type 3 channels occupy lowland areas inboth Alberta and Michigan.

Fig. 5. (A) Hillshaded relief map of channels in the southern part of

the Alberta study area. Note the offset character of the depressions

that form the channel (arrow). (B) Topographic context of the

channels in (A). Channel indicated by the arrow has a convex-up long

profile as it dissects a hummocky upland. Data source: 1 : 20 000

Alberta Environmental Protection digital elevation models with 25m

sampling.


Fig. 7. (A) Hillshaded relief map of the channels in the eastern part of the Michigan study area (south of Albion). Transverse ridges and hummocks

form the crenulated margins of the channels (arrows). (B) Topographic context of the channels in (A). The channels have significant uphill sections

(flow toward the south). Data source: United States Geological Survey 7.50 digital elevation models (Level 2) with 30m sampling.

Fig. 6. (A) Hillshaded relief map of channels in the northwestern part of the Michigan study area. Note the bulbous shape of the channel formed by

the depressions (arrows). (B) Topographic context of the channels in (A). The overall convex-up long profile is formed as the channel dissects the

upland area that contains drumlins. Data source: United States Geological Survey 7.50 digital elevation models (Level 2) with 30m sampling.


3.3. Sediment and landform associations

In the Alberta study area there is a relationshipbetween the channel morphology and the sedimentpresent. The sediment that dominates the margins andintrachannel hummocks of Type 1 channels is diamicton(till). In some cases, however, bedrock is exposed at thebase of the intrachannel hummocks and in depressions.Boulder lags are common at the bottom of thedepressions (Fig. 9A). Boulder gravel is present at thetop of some hummocks, overlying an erosional contactwith the till. Type 2 channel margins have a thin cover oftill overlying bedrock. Hummocks within these channelsare composed of either till or bedrock (Fig. 11A).Boulder lags are common both within and adjacent tothe channels. Perched gravel deposits are present alongthe channel margins. The steep margins of most Type 3channels are composed of bedrock (Fig. 9C). Hum-mocks, if present, are generally composed of bedrock; insome cases, however, they are formed of sand and gravel(Figs. 11B and C).

The structure of one intrachannel hummock in theMichigan study area is illustrated in Fig. 12. A ground-penetrating radar (GPR) transect (see Fisher et al., 1995for methodology) shows the stratigraphy of a tunnelchannel island and adjacent upland plain (Fig. 12A). Atransect along the upland plain, not shown here, reveals

only a few parallel reflectors below a zone of closelyspaced, wavy reflectors identical to those at 10–20m inFig. 12B. A roadcut reveals a thin sandy diamicton overglaciofluvial sand and gravel in this area (Kozlowski,1999). Over the tunnel channel island, penetrationincreases (west of 30m in Fig. 12B). Unit 1 containswavy, semi-continuous reflections within a mound-likelandform (Fig. 12B). Units 2 and 3 consist of continuousto semi-continuous, parallel to wavy reflections thatstack upon one another, with short, westward-dippingreflections near the surface associated with unit 2. Atpositions 145–155m, a steeply dipping reflection trun-cates the reflections of unit 2, indicating that anerosional event was followed by deposition of unit 3.Bedrock is deeper than 25m along the transect.

The GPR transect indicates that the upland surface isunderlain by sand and/or gravel. These sediments mayhave been deposited in a braided river environment,consistent with Kozlowski’s (1999) stratigraphic inter-pretation for the area. Unit 1 in the tunnel channelisland may be boulder beds. Units 2 and 3 are inter-preted to be braided river deposits. Unit 3 is similar tounit 2, but separated from unit 2 by an erosional event,indicated by the steeply dipping reflector on the westside of the island. The white dashed lines in Fig. 12Bextend the stratigraphy of unit 2 before road construc-tion and slope recession. Because the GPR stratigraphy

Fig. 8. Three types of channels and their stratigraphic context. The cross sections are idealized. (A) Type 1 channels consist of elliptical depressions

that are larger than depressions in the adjacent hummocky terrain. These depressions are commonly elongated transverse to paleoflow direction. (B)

Type 2 channels have crenulated margins. Channel depressions are separated by ridges or hummocks that are significantly lower than the

surrounding terrain. These separations are commonly composed of till but gravel and bedrock hummocks are present at some locations. Boulder lags

are common. (C) Type 3 channels have flat bottoms and may contain eskers. Boulder lags are common. Data source: 1 : 20 000 Alberta

Environmental Protection digital elevation models with 25m sampling.


of the islands appears to be the same as that of the adja-cent plain, we conclude that the island was once part ofthe adjacent plain and thus is an erosional remnantrather than a constructional landform (e.g., a kame).

Eskers are rare in hummocky terrain, but severaleskers are closely associated with channels in the Albertaand Michigan study areas. One esker, for example,crosses Type 2 channels and their interchannel divide in

Fig. 9. (A) Linked depressions that form part of a Type 1 channel. Bedrock is exposed at the base of some hummocks and boulder lags are common. (B)

Low relief hummocks in a Type 2 channel (arrows). These hummocks are usually composed of till, but some contain gravel and sand (Fig. 11). Boulder lags

are common in these channels. (C) Flat bottom of a Type 3 channel. These channels are cut into bedrock and usually have relatively steep margins.

Fig. 10. Stereogram showing a lateral ridge adjacent to a Type 2 channel (arrow). Such ridges likely formed as sediment was squeezed by ice as the

overlying glacier reattached to terrain adjacent to the channel. This would represent the last significant terrain modification by glacial ice. Source: Air

Photo Services, Alberta Environmental Protection, AS4439, #172 and 173.


the Viking area (Fig. 13). Isolated, discontinuous eskersalso occur within Type 3 channels roughly parallel to thechannel margins. In the Michigan area, eskers are foundwith Type 2 channels.

4. Discussion

4.1. Environment of channel formation

Three morphologic characteristics of the channelsindicate that they are erosional features that wereformed in a subglacial environment by turbulent melt-water flows. First, many channels have pronouncedconvex-up long profiles that cross bedrock divides,which requires erosion by pressurized water (Sjogrenand Rains, 1995). Interpretations that invoke formationof segments subaerially do not explain the continuity ofthe segments or their negligible adherence to topo-graphic gradients. The convex-up long profiles could notbe the result of subsequent crustal movements. Second,channel morphology is linked to the nature of thesubstrate; where there is a thick till cover over bedrock,channels are generally narrower with more pronouncedhummocks along their base (Types 1 and 2). Where tillcover is thin or absent, the channels have relatively flatbottoms, steep margins, and may contain depositionallandforms at their base (Type 3). Third, wheretransverse ridges are present, the channels either

truncate the ridges or are deflected by them. Thisindicates that minor topographic irregularities influ-enced channel shape and orientation. The transverseridges cannot be recessional moraines because they aretruncated by channels with convex-up thalwegs. Thisrelationship indicates that the ridges formed before thechannels.

Stratigraphic and landform associations also provideclues to the timing and environment of channelformation. First, truncation of the stratigraphy in thechannel walls shows that the channels formed afterthe till was deposited. Second, bedrock exposed inchannel walls, adjacent to channel bottom hummockscomposed of till, indicates that some channels predatethe local onset of glaciation. Third, eskers occur within,and drape, some channels, indicating that they formedafter the channels. The nearly orthogonal orientationof eskers and channels indicates a significant reorgani-zation of the glaciohydrologic system after thechannels developed. Eskers are usually interpretedas stagnant ice features (Brennand, 2000), therefore,the channels in Alberta and Michigan formed subgla-cially, under pressure, prior to the eskers. Thestratigraphic and landform associations demonstratethat the channels formed between the time of tilldeposition and esker formation, and thus in a subglacialenvironment.

Ascribing these landforms to a subglacial environ-ment contradicts the traditional view that they are icemarginal features. Gravenor and Bayrock (1956) inter-preted the Viking channels as the result of partialinfilling of ice-walled channels by slumped sediment.Parizek (1969) echoed this interpretation for channels insouthern Saskatchewan. Gravenor and Ellwood (1957)noted that the width of some channels (B1 km),compared to their depth (B25m), precluded simpleslumping as the mechanism for intrachannel hummockformation. They supported the idea that a readvancewas responsible for the partial infilling. However,landform evidence for a readvance in the Viking areais absent. Rather, channels with convex-up thalwegstruncate ridges that could represent a former ice margin.

Landform evidence has been proposed for a read-vance to the Tekonsha Moraine, which extends acrossthe middle of Fig. 6 (Leverett and Taylor, 1915;Kozlowski, 1999). This evidence consists of faint ridges,surface boulder concentrations, and ice-contact fans.The ridges and fans, however, are not visible on ourDEM. Alternatively, these deposits may be crevassefillings and kames that postdate the formation of thedrumlins and tunnel channels. Melt-out of stagnant icewould also explain the thin sandy diamicton describedby Kozlowski (1999) and blanketing the area shown inFig. 6. We agree with Kozlowski (1999), however, thatthe tunnel channels and drumlins are synchronous andformed subglacially.

Fig. 11. (A) Low-amplitude bedrock hummock in a Type 3 channel.

(B) Gravel hummock in a Type 2 channel. (C) Poorly organized gravel

at the location of B. Scale has decimetre divisions.


4.2. Analogous forms

We have demonstrated that glaciofluvial channels inour Alberta and Michigan study areas are complex, withtheir different morphologies linked to regional topogra-phy and sediment type. We have also shown that theyformed in a subglacial environment. However, thecharacteristics of the flows responsible for the variouserosional forms are not obvious. To address this issue,we consider several morphologic analogues, of a widerange of sizes that may provide insight into thecharacteristics of the turbulent flow responsible for thechannels we have described.

Shepherd and Schumm (1974) demonstratedthat river incision in bedrock produced longitudinal

lineations, potholes, and transverse erosional ripples.With time, erosional lineations enlarged to formgrooves. Potholes developed at various places alongthe channel before grooves became the main bedform.

The Channeled Scabland in Washington State, USA,was produced by catastrophic drainage of glacial lakesMissoula and Columbia (e.g., Bretz, 1969). Highlyturbulent flow deeply eroded loess and underlyingbasalt, creating landforms that are unequivocally ero-sional (Fig. 14). These bedforms are, in many ways,similar to the erosional landforms in Alberta andMichigan. Hummocky terrain in our study areasresembles the butte-and-basin topography developed inbedrock in Lenore Canyon (Baker et al., 1987, Fig. 30),but occur at a larger scale. Longitudinal grooves and

Fig. 12. (A) Location of the ground penetrating radar (GPR) transect. Map area is just off the southwest corner of Fig. 6. (B) GPR data and

interpretation of the transect located in (A).


anabranching channels near Dry Falls (Baker et al.,1987, Figs. 28 and 29) (Fig. 15) and longitudinal groovesin the Cheney Palouse scabland tract (Fig. 16), are alsosimilar to the Alberta and Michigan channels. Thecrenulated margins, acute-to-orthogonal interconnec-tions, shallow basins, deflections of longitudinalgrooves, convex-up thalwegs, erosional intrachannelresiduals (hummocks), blind-ended segments, and dis-continuities along segments are all present in our studyareas. Major differences in appearance relate to thesubstrate into which they are eroded (basalt versus tilland/or gravel).

The forms found in the scabland have been attributedto erosion of highly jointed basalts by macroturbulent

eddies called kolks (Matthes, 1947; Baker, 1978a).Width of flow in the Dry Falls cataract complex andthe Cheney Palouse scabland tract is B6 km andB20 km, respectively, compared to B60 km for boththe Alberta and Michigan channel systems (Figs. 2 and3). The only obvious difference between the glacial andscabland examples is that the potholes and longitudinalgrooves are deeper relative to their width where erodedinto basalt.

Wohl (1993) reported longitudinal grooves, similar tothose in the Channeled Scabland that formed in high-velocity channelized flows. Though virtually identicalmorphologically, to the Channeled Scabland features,they are much smaller (28 cm wide and 23 cm deep).Wohl (1993, Fig. 3) also described sinuous grooves thatresemble linked, offset potholes. The grooves join atacute angles and contain transverse depressions. Furthererosion of the potholes results in discontinuous,sinuous-walled inner channels, with reversed gradientsat the down-flow end. Formation of such inner channelssupports the experimental work by Shepherd andSchumm (1974) and the spillway formation sequenceof Kehew (1982). Linking potholes by breaching theseparating ridges is thought to be an importantmechanism for forming the inner channel in bedrockstreams (Shepherd and Schumm, 1974; Wohl, 1993).

Aalto et al. (1999) reported strikingly similar, tsunamierosional forms to those described above. They de-scribed long, narrow, sinuous grooves consisting of aseries of potholes linked by shallow connecting channelsand ranging from 5 to 20 cm wide and 2 to 36 cm deep.The grooves are commonly internally drained. Smaller

Fig. 13. (A) Airphoto of the Type 2 channels (dashed lines) super-

imposed by an esker (arrows). Source: Alberta Environmental

Protection photo, AS 4439, #171. (B) Topographic context of the

channels and esker. Note that the esker occupies an indistinct trough

that separates two zones of hummocky terrain. The esker crosses

terrain with an elevation range of between 15 and 20m. Data source:

1 : 20 000 Alberta Environmental Protection digital elevation models

with 25m sampling.

Fig. 14. Location of Figs. 15 and 16 within the Channeled Scabland in

Washington State. Shaded polygons represent Palouse Formation.

Modified from Baker (1978b, Fig. 5.2).


potholes and transverse ripples occur on intergrooveridges. The intergroove ridges are similar to ridgesbetween channels in the Alberta and Michigan studyareas. Some grooves are discontinuous, and some havehigh-angle connections with adjacent grooves. Aaltoet al. (1999) determined that bedrock joints areimportant, but not essential to the formation of theseand related erosional bedforms.

4.3. Formation of incipient tunnel channels

Several themes emerge from our observations anddiscussion. First, the bedforms that we have describedare primarily controlled by hydrodynamic conditions.Second, bedrock jointing or unusual material character-istics are not required for the formation of the bedforms.Third, the developmental sequence suggested byShepherd and Schumm (1974) appears to be recordedin the arrested development of the bedforms. Fourth,potholes, transverse ridges, linked potholes, and long-itudinal grooves may be associated with progressivechannelization by a highly turbulent flow and may,therefore, represent different stages in that developmen-tal sequence. Fifth, the landforms we have describedfrom glaciated regions have smaller-scale analogues inbedrock river valleys and the Channeled Scablands.

Very little has been written about the hydraulicconditions that form linked potholes, thus the formativeflow structures are unclear. Wohl (1993) noted thesimilarity between the sequence of erosional bedformsand a periodic or von Karman vortex street. Althoughthis type of flow structure is not present in intermediateflows, there is evidence that the growth and shedding ofvortices in this manner occurs at high Reynolds numbers(Karcz, 1973; Clifford et al., 1992) and could beresponsible for the linked pothole channels described

Fig. 15. (A) Hillshaded relief map of longitudinal grooves and

discontinuous anabranching channels cut into basalt near the Dry

Falls Cataract Complex. Grooves on the upland areas and the

intrachannel residuals are similar to those in the Michigan study area

(Fig. 3). (B) Topographic context of the features in (A). Dry Falls is

located immediately north of these maps. Data source: United States

Geological Survey 7.50 digital elevation models (Level 2) with 30m

sampling.

Fig. 16. (A) Hillshaded relief map of longitudinal grooves located in the Cheney Palouse Scabland tract. Although they are cut into basalt, the size,

pattern, and interconnectivity of individual grooves are very similar to the channels in the Alberta study area, compare with Fig. 4. (B) Topographic

context of the features in (A). Data source: United States Geological Survey 7.50 digital elevation models (Level 2) with 30m sampling.


in this paper. Wohl (1993) stressed that vortices shed offirregularities on the bed may be self-perpetuating. Theinteraction between bed roughness elements and theflow has been cited as important for focussing erosion,an idea proposed for other erosional bedforms (Sharpeand Shaw, 1989).

Kolks have been implicated in the formation oflandforms in bedrock channels, including basalt, and itis possible that similar flow structures could exist inhighly turbulent, catastrophic, subglacial flow over tillor poorly consolidated bedrock such as in Alberta andMichigan. Although vortices are known to sponta-neously form in fluid flows (Einstein and Li, 1958),minor topographic elements on the bed could alsospawn vortices. In the subglacial environment there arethree potential roughness elements: irregularities inthe water–sediment interface, irregularities along thewater–ice interface, and irregularities due to interactionof internal waves. Interaction among these three rough-ness components and turbulent, erosive flow would beexceptionally complex. Three-dimensional constrictionsand expansions in the flow could produce spatial andtemporal fluctuations in velocity, vortices, and thuserosion.

4.4. Proposed developmental sequence of incipient tunnelchannels

Catastrophic releases of water, adequate to formtunnel channels, require that vast amounts of water werestored subglacially and/or supraglacially. Recent at-tempts at quantifying subglacial lake volumes inAntarctica (Dowdswell and Siegert, 1999; Siegert,2000) support the contention that large subglacial lakesmay have formed under the Laurentide ice sheet(Shoemaker, 1991, 1992, 1999). Munro-Stasiuk (2000)and Munro-Stasiuk and Sjogren (in press) describedfield evidence for subglacial lakes in Alberta. Drainingof many interconnected or isolated subglacial reservoirssimilar to those described by Munro-Stasiuk andSjogren (in press) could supply the water necessary toform the features described in this paper. Draining ofthe reservoirs could have destabilized others, resulting ina domino-like release, thereby initiating large cata-strophic flows (Shaw, 1996).

Once flow was initiated, roughness elements createdvarious scales of longitudinal and transverse expansionsand constrictions in the flow (Fig. 17). Roughnesselements ranged in size from preglacial valleys and

Fig. 17. Cartoon illustrating the proposed developmental sequence for incipient tunnel channels. (A) Different subglacial roughness elements are

related to glacier and substrate configuration. (B) As catastrophic release of meltwater commences, roughness elements spawn vertical vortices

(kolks) that accentuate these elements. A positive feedback causes the erosion to be concentrated in the depressions. (C) Water is increasingly

concentrated in channels as discharge drops, producing the intrachannel ridges and hummocks. As ice reattaches to the channel margins, it deforms

the substrate creating lateral ridges. (D) The entire flow is concentrated in channels creating a more recognizable channel form.


interfluves to small-scale glaciotectonic bumps (Sjogren,1999) (Fig. 17A). Deflection of flow over bed irregula-rities caused near-vertical return flow in the lee ofobstacles, thereby initiating macroturbulent flow struc-tures such as kolks (Fig. 17B). These flow structuresbecame fixed because kolk erosion created potholes that,once initiated, formed a larger flow impediment thataccentuated the flow structure and thus the erosionalform (Shaw, 1996). Linked potholes (Type 1 channels,Figs. 8A and 9A) began to form as some depressionswere accentuated. Further erosion caused breaching ofdivides between potholes, leading to the intermediateType 2 channels (Figs. 8B, 9B and 17C). This increasedchannelization focused the erosion in the incipienttunnel channels. As the discharge dropped, the channelsconveyed the entire flow. This resulted in the iceadjacent to the channel reattaching to the bed, therebycreating the channel flanking ridges (Fig. 10). The morefamiliar tunnel channel form, with a nearly flat bottomand steep margins (Figs. 8C, 9C and 17D) was createdwhere the flow duration was adequate to remove theintrachannel hummocks or ridges. The presence of thethree channel types represents a developmental sequencewhere the flow duration and bed characteristics con-trolled the shape of the channels.

In the Alberta study area, flow that created theincipient tunnel channels was roughly parallel to theeastern margin of one subglacial flood path identified byRains et al. (1993). Erosional landforms in the easternflood path of Rains et al. (1993) are coincident with thechannels described herein, thus it is likely that they werecontemporaneous. The meltwater flows overtoppedpreglacial valleys and interfluves at right angles. In theAlberta study area, flow was forced over bedrock-controlled uplands that acted as transverse escarpments,albeit subtle ones. The connection between the floodpath and the channels indicates that the incipient tunnelchannels are the result of overbank-like subglacial flow.Their incomplete formation indicates that flow durationwas insufficient to fully develop channels.

Flow in south-central Michigan encountered escarp-ments along the edges of the upland surfaces. Theproximal ends of the uplands (northeast edges on Fig. 3)show greater sculpting than the distal ends (see alsoTinkler and Stenson, 1992; Brennand and Shaw, 1994;Pollard et al., 1996; Shaw et al., 2000). Between thetunnel channels, smaller incipient tunnel channelsdelineate some hummocks. The smaller tunnel channelsare characterized by sinuous depressions similar to theupper dissected terrain described by Shepherd andSchumm (1974) and Kehew and Lord (1986). The onlydifference is that the channels in the Michigan studyarea cross over topographic divides (Figs. 6 and 7).Adjacent well-developed tunnel channels indicate atemporally dependent focusing of flow and developmentof an inner channel (Shepherd and Schumm, 1974).

5. Conclusions

1. Many glaciofluvial channels in both the Albertaand Michigan study areas cross upland areas andpossess convex-up or undulating long profiles. Thischaracteristic requires that they were formed underpressure. This could only occur in a subglacialenvironment.

2. Truncation of the stratigraphies in the channelmargins and intrachannel hummocks indicates thatthe depressions are erosional. Boulder lags attest tothe preferential erosion of the fine-grained material.

3. Eskers that superpose tunnel channels indicate thatthey formed prior to the final stagnation thatproduced the eskers.

4. Comparison with similar-scaled landforms suggeststhat, at a simple level, vortices shed from flow-perimeter irregularities produced the linked potholesof incipient tunnel channels.

5. Tunnel channels exist in a variety of developmentalstages that follow a logical sequence. The variety offorms reflects the duration of the formative flow,the intensity of the erosion, and the presence oftransverse roughness elements such as bedrockescarpments and/or glaciotectonic ridges.

6. The incipient tunnel channels, described in this paper,are but one component of a continuum of landforms,ranging from erosional hummocky terrain (Munroand Shaw, 1997) to more typical tunnel channels(Rains et al., 2002). The regional connectivity of thesedifferent landforms strongly suggests a regional scaleevent that required the catastrophic drainage of largereservoirs of meltwater (Munro-Stasiuk and Sjogren,in press).

Acknowledgements

We thank L. Clayton, R. Gilbert, A. Kehew, A.Kozlowski, Y. Martin, B. Rains, and J. Shaw forreviewing earlier versions of this paper. We also thankthe attendants of the 1999 Midwest Friends of thePleistocene who visited the Michigan study area with usand provided many constructive comments, the Carrbrothers for access to their gravel pits, and TheGeological Society of America and Indiana UniversityNorthwest for research funding.

References

Aalto, K.R., Aalto, R., Garrison-Laney, C.E., Abramson, H.F., 1999.

Tsunami(?) sculpturing of the Pebble Beach wave-cut platform,

Crescent City area, California. Journal of Geology 107, 607–622.

Baker, V.R., 1978a. Paleohydraulics and hydrodynamics of Scabland

floods. In: Baker, V.R., Nummedal, D. (Eds.), The Channeled

Scabland: A Guide to the Geomorphology of the Columbia Basin,


Washington. National Aeronautics and Space Administration,

Washington, DC, pp. 59–79.

Baker, V.R., 1978b. Large-scale erosional and depositional features of

the Channeled Scabland. In: Baker, V.R., Nummedal, D. (Eds.),

The Channeled Scabland: A Guide to the Geomorphology of the

Columbia Basin, Washington. National Aeronautics and Space

Administration, Washington, DC, pp. 17–35.

Baker, V.R., Greeley, R., Komar, P.D., Swanson, D.A., Waitt Jr.,

R.B., 1987. Columbia and Snake River plains. In: Graf, W.L.

(Ed.), Geomorphic Systems of North America. Geological Society

of America, Centennial Special Volume 2, pp. 403–468.

Bayrock, L.A., 1958. Glacial Geology AllianceFBrownfield District

Alberta. Research Council of Alberta Report 57-2, 56pp.

Brennand, T.A., 2000. Deglacial meltwater drainage and glaciody-

namics: inferences from Laurentide eskers, Canada. Geomorphol-

ogy 32, 263–293.

Brennand, T.A., Shaw, J., 1994. Tunnel channels and associated

landforms, south-central Ontario: their implications for ice-sheet

hydrology. Canadian Journal of Earth Sciences 31, 505–522.

Bretz, J.H., 1943. Keewatin end moraines in Alberta, Canada.

Geological Society of America Bulletin 54, 31–52.

Bretz, J.H., 1969. The Lake Missoula floods and the Channeled

Scabland. Journal of Geology 77, 505–543.

Clayton, L., Attig, J.W., Mickelson, D.M., 1999. Tunnel

channels formed in Wisconsin during the last glaciation. In:

Mickelson, D.M., Attig, J.W. (Eds.), Glacial Processes Past and

Present. Geological Society of America, Special Paper 337,

pp. 69–82.

Clifford, N.J., Robert, A., Richards, K.S., 1992. Estimation of flow

resistance in gravel-bedded rivers: a physical explanation of the

multiplier of roughness length. Earth Surface Processes and

Landforms 17, 111–126.

Dowdswell, J.A., Siegert, M.J., 1999. The dimensions and topographic

setting of Antarctic subglacial lakes and implications for large-scale

water storage beneath continental ice sheets. Geological Society of

America Bulletin 111, 254–263.

Dyke, A.S., Prest, V.K., 1987. Late Wisconsinan and Holocene

Retreat of the Laurentide-Ice Sheet. Geological Survey of Canada

Map 1702A, scale 1 : 5,000,000.

Einstein, H.A., Li, H., 1958. Secondary flow in straight channels.

Transactions, American Geophysical Union 39, 1085–1088.

Evans, D.J.A., 2000. Quaternary geology and geomorphology of the

Dinosaur Provincial Park area and surrounding plains, Alberta,

Canada: the identification of former glacial lobes, drainage

diversions and meltwater flood tracks. Quaternary Science Reviews

19, 931–958.

Evans, D.J.A., Campbell, I.A., 1995. Quaternary stratigraphy of the

buried valleys of the lower Red Deer River, Alberta, Canada.

Journal of Quaternary Science 10, 123–148.

Eyles, N., Boyce, J.I., Barendregt, R.W., 1999. Hummocky moraine:

sedimentary record of stagnant laurentide ice lobes resting on soft

beds. Sedimentary Geology 123, 163–174.

Farrand, W.R., 1982. Quaternary geology of Southern Michigan. State

of Michigan Department of Natural Resources, map scale

1 : 500,000.

Fisher, T.G., Taylor, L.D., 1999. Glacial landforms and sediment

landform assemblages, Saginaw Lobe. In: Brown S.E., Fisher,

T.G., Taylor, L.D., Kehew, A. (Eds.), Guidebook for the 45th

Midwest Friends of the Pleistocene Field Conference. Indiana

Geological Survey, Open File Study 99-1, pp. 3–23.

Fisher, T.G., Taylor, L.D., 2000. Boulder beds, a product of subglacial

outburst flooding, south-central Michigan. Geological Society of

America, Abstracts with Programs 32 (7), A116.

Fisher, T.G., Taylor, L.D., 2002. Stratigraphic evidence for subglacial

flooding, south-central Michigan, USA. Quaternary International,

90 (1), 87–115.

Fisher, T.G., Jol, H.M., Smith, D.G., 1995. Ground-penetrating

radar used to assess aggregate in catastrophic flood deposits,

northeast Alberta, Canada. Canadian Geotechnical Journal 32,

871–879.

Gravenor, C.P., 1955. The origin and significance of prairie mounds.

American Journal of Science 253, 475–481.

Gravenor, C.P., Bayrock, L.A., 1956. Stream-trench systems in East-

central Alberta. Research Council of Alberta Preliminary Report

56-4, 11pp.

Gravenor, C.P., Ellwood, R.B., 1957. Glacial geology: Sedgewick

District, Alberta. Research Council of Alberta Preliminary Report

57-1, 43pp.

Gravenor, C.P., Kupsch, W.O., 1959. Ice-disintegration features in

Western Canada. Journal of Geology 67, 48–64.

Holden, K.M., 1993. Glacial environments of the Edmonton Region,

Alberta. M.Sc. thesis, University of Alberta, Edmonton, AB,

108pp.

Hoppe, G., 1952. Hummocky moraine regions, with special reference

to the interior of Norrbotten. Geografiska Annaler 34, 1–72.

Karcz, I., 1973. Reflections on the origin of small-scale longitudinal

streambed scours. In: Morisawa, M. (Ed.), Fluvial Geomorphol-

ogy. Allen & Unwin, Binghamton, NY, pp. 149–173.

Kehew, A.E., 1982. Catastrophic flood hypothesis for the origin of the

Souris spillway, Saskatchewan and North Dakota. Geological

Society of America Bulletin 93, 1051–1058.

Kehew, A.E., Lord, M.L., 1986. Origin and large-scale erosional

features of glacial lake spillways in the northern Great Plains.

Geological Society of America Bulletin 97, 162–177.

Klassen, R.W., 1989. Quaternary geology of the southern Canadian

Interior Plains. In: Fulton, R.J. (Ed.), Quaternary Geology of

Canada and Greenland. Geological Survey of Canada, Geology of

Canada No. 1, pp. 138–173.

Kozlowski, A., 1999. Three dimensional mapping of the East Leroy

and Union City 7.5 Minute Quadrangles in Southwest Michigan.

M.S. Thesis, Western Michigan University, Kalamazoo, MI,

113pp.

Leverett, F., Taylor, F.B., 1915. The Pleistocene of Indiana and

Michigan and the history of the Great Lakes. United States

Geological Survey Monograph 53, 527pp.

Matthes, G.H., 1947. Macroturbulence in natural stream flow.

Transactions, American Geophysical Union 28, 255–262.

Munro, M.J., Shaw, J., 1997. Erosional origin of hummocky terrain in

south-central Alberta, Canada. Geology 25, 1027–1030.

Munro-Stasiuk, M.J., 2000. Rhythmic till sedimentation: evidence for

repeated hydraulic lifting of a stagnant ice mass. Journal of

Sedimentary Research 70, 94–106.

Munro-Stasiuk, M.J., Sjogren, D.B., in press. Evidence for water

storage and catastrophic release beneath the Laurentide Ice Sheet,

Alberta, Canada. International Association of Hydrological

Sciences Red Book Series: Proceedings of the Conference on The

Extremes of the Extremes: Extraordinary Floods, Iceland.

Parizek, R.R., 1969. Glacial ice-contact rings and ridges. In: United

States Contributions to Quaternary Research. Geological Society

of America Special Paper 123, pp. 49–102.

Pawlowicz, J.G., Fenton, M.M., 1995. Bedrock Topography of

Alberta. Alberta Geological Survey Map 226, scale 1 : 2,000,000.

Pollard, A., Wakarani, N., Shaw, J., 1996. Genesis and morphology of

erosional shapes associated with turbulent flow over a forward-

facing step. In: Ashworth, P.J., Bennet, S.J., Best, J.L., McLelland,

S.J. (Eds.), Coherent Flow Structures in Open Channels. Wiley,

New York, pp. 249–265.

Rains, B., Shaw, J., Skoye, B., Sjogren, D., Kvill, D., 1993. Late

Wisconsin subglacial megaflood paths in Alberta. Geology 21,

323–326.

Rains, R.B., Shaw, J., Sjogren, D.B., Munro-Stasiuk, M.J., Skoye,

K.R., Young, R.R., Thompson, R.T., 2002. Subglacial tunnel


channels, Porcupine Hills, southwest Alberta, Canada. Quaternary

International, 90 (1), 57–65.

Sharpe, D.R., Shaw, J., 1989. Erosion of bedrock by subglacial

meltwater, Cantley, Quebec. Geological Society of America

Bulletin 101, 1011–1020.

Shaw, J., 1996. A meltwater model for Laurentide subglacial

landscapes. In: McCann, B., Ford, D.C. (Eds.), Geomorphology

Sans Fronti"eres. Wiley, West Sussex, UK, pp. 181–236.

Shaw, J., Faragini, D.M., Kvill, D.R., Rains, R.B., 2000. The

Athabasca fluting field, Alberta, Canada: implications for the

formation of large-scale fluting (erosional lineations). Quaternary

Science Reviews 19, 959–980.

Shepherd, R.G., Schumm, S.A., 1974. Experimental study of river

incision. Geological Society of America Bulletin 85, 257–268.

Shetsen, I., 1990. Quaternary Geology, Central Alberta. Alberta

Research Council, map, scale 1 : 500,000.

Shoemaker, E.M., 1991. On the formation of subglacial lakes.

Canadian Journal of Earth Sciences 28, 1975–1981.

Shoemaker, E.M., 1992. Water sheet outburst floods from the

Laurentide ice sheet. Canadian Journal of Earth Sciences 29,

1250–1264.

Shoemaker, E.M., 1999. Subglacial water-sheet floods, drumlins and

ice-sheet lobes. Journal of Glaciology 45, 201–213.

Siegert, M.J., 2000. Antarctic subglacial lakes. Earth Science Reviews

50, 29–50.

Sjogren, D.B., 1999. Formation of the Viking Moraine, East-central

Alberta: geomorphic and sedimentary evidence. Ph.D. Thesis,

University of Alberta, Edmonton, AB, 219pp.

Sjogren, D.B., Rains, R.B., 1995. Glaciofluvial erosional morphology

and sediments of the Coronation-Spondin Scabland, east-central

Alberta. Canadian Journal of Earth Sciences 32, 565–578.

Stalker, A. Mac, S., 1960. Ice-pressed drift forms and associated de-

posits in Alberta. Geological Survey of Canada Bulletin 57, 38pp.

Teller, J.T., 1987. Proglacial lakes and the southern margin of the

Laurentide Ice Sheet. In: Ruddiman, W.F., Wright Jr., H.E. (Eds.),

North America and Adjacent Oceans during the Last Glaciation.

Geological Society of America, The Geology of North America,

Volume K-3, pp. 39–69.

Teller, J.T., 1990. Volume and routing of late-glacial runoff from the

southern Laurentide ice sheet. Quaternary Research 34, 12–23.

Tinkler, K.J., Stenson, R.E., 1992. Sculpted bedrock forms along the

Niagra Escarpment, Niagra Peninsula, Ontario. G"eographie

physique et Quaternaire 46, 195–207.

Warren, P.S., 1937. The significance of the Viking Moraine.

Transactions, Royal Canadian Institute 21, 301–305.

Wohl, E.E., 1993. Bedrock channel incision along Piccaninny Creek,

Australia. Journal of Geology 101, 749–761.

Zumberge, J.H., 1960. Correlation of Wisconsin drifts in Illinois,

Indiana, Michigan, and Ohio. Geological Society of America

Bulletin 71, 1177–1188.


incipienttunnelchannels - university of wisconsin–eau claire of...bdepartment of geosciences,...

Documents