PALEOGLACIOLOGICAL CONTEXT OF ROGEN , NORTHEASTERN MINNESOTA

A THESIS SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL OF THE UNIVERSITY OF MINNESOTA BY

MARGRETTA SOPHIA MEYER

IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE

May, 2008

i

© Margretta Sophia Meyer 2008

ii Table of Contents

Table of Contents i

List of Illustrations ii

Acknowledgements iii

Introduction 1

Study Site 2

Glacial History 5

Rogen Moraine 11

Hypotheses for Formation 24

Methods of Investigation 41

Results 43

Discussion 49

Conclusion 61

References 64

i List of Illustrations

Fig. 1. Location of study area 4

Fig. 2. Phases of glaciation in Minnesota 9

Fig. 3. End and other features in northeastern Minnesota 9

Fig. 4. Geographic distribution of Rogen moraine 13

Fig. 5. Rogen moraine morphological characteristics 16

Fig. 6. Jigsaw-like matching of Rogen moraine 17

Fig. 7. Association with other 22

Fig. 8. Shear and stack model 26

Fig. 9. Two-step/precursor ridge model 28

Fig. 10. Thrust stacking and lee-side cavity fill model 29

Fig. 11. Catastrophic subglacial flood model 31

Fig. 12. Extensional fracturing model 33

Fig. 13. Field map 39

Fig. 14. Photos of in field area 40

Fig. 15. Rose diagrams of moraine and orientations 42

Fig. 16. Grain size classification 43

Fig. 17. GPR profile 45

Fig. 18. Coulomb fracture criteria 48

Fig. 19. Isabella ribbed moraine extension 49

Fig. 20. Plan and cross-sectional view of boudins 58

ii Acknowledgements

Thank you to my examining committee, Howard Mooers, Stacey Stark, and Keith

Brugger. Thank you particularly to Howard for being patient. Tim Demko provided valuable GPR field assistance and I am grateful to Nigel Wattrus for processing the GPR data. Thanks to Vicki Hansen for references on boudins. The UMD Geology

Department provided funds for fieldwork expenses, and the UMD Visualization and

Digital Imaging Lab provided summer support in 2005.

Many thanks to Sarah Davidson for camping out, chipping through compacted subglacial till, enduring repetitive banjo practice, and doing controlled experiments on botanical tick repellent. My family and friends, especially my fellow graduate students, without whom I certainly would not have finished this, have kept me sane through your support and encouragement. And lastly, T.C. Chamberlin for inspiration.

iii Introduction

Subglacial landforms are important features of continental glaciations, yet have traditionally posed problems for researchers simply because of the difficulty of physically investigating or modeling the subglacial environment. Modern continental ice sheets hide the formation of any analogous structures. Active alpine settings feature recently exposed glacier beds but do not behave like larger scale continental ice sheets.

We are therefore confined to working with the geomorphological records left by the most recent continental glaciation; these landforms are the key to understanding not only their mechanics, but the behavior of the glacier in creating them.

Rogen, or ribbed, moraines are enigmatic ridges found throughout the northern parts of areas covered by the Laurentide, Fennoscandian, and Irish Ice Sheets.

Transverse to ice flow and of uncertain origin, they are generally assumed to be subglacial landforms. Many hypotheses for their formation have been suggested over the last century. These range from marginal features, ripple marks from catastrophic subglacial floods, to a wide variety of subglacial deformation.

Although prevalent throughout Canada, there are few examples of Rogen moraine in the United States. Rogen moraine in northeastern Minnesota thus offers a unique perspective on both potential modes of formation and paleoglaciology of the Late

Wisconsin Rainy Lobe. The formation of this area of ribbed landscape is constrained to a narrow window of time compared to other areas located more centrally within the

Laurentide (LIS). Ice advanced over this area of northeastern Minnesota shortly before the (LGM) and retreated shortly thereafter, also offering further constraint on the glaciological conditions.

1 and striated bedrock provide little evidence that cold-based conditions existed in this area. The Rogen landscape is developed in an area of thin till, down- glacier from ice-scoured bedrock and up-glacier from a thicker drumlinized till sheet. The abrupt change in basal boundary conditions along this flowline may have resulted in initiation of extensional flow and Rogen moraine formation. This transition poses intriguing questions as to the behavior of the Laurentide

Ice Sheet along this flowline and suggests that in this setting, a transition in the glacial substrate may be responsible for Rogen moraine genesis.

Study site

The study area is located in central Lake County, MN, immediately north of the town of Isabella and Minnesota State Highway 1 within the Tofte district of the Superior

National Forest. This area was glaciated repeatedly by the Rainy Lobe of the LIS during the Late Wisconsin and lies to the west of the complex interlobate junction of the Rainy and Superior Lobes (Fig. 1). The Rogen moraine trends northwest-southeast at an angle of approximately 300˚ and is primarily confined to the northern halves of the Mitawan

Lake and Sawbill Landing 7.5" 1:24,000 USGS quadrangles, with a few ridges in the

Gabbro Lake, Quadga Lake, and Slate Lake East quadrangles. These ridges are visible on both 1:24,000 scale topographic maps and on shaded-relief DEM of the area. The landscape is heavily vegetated and marked by wetlands or parallel streams between the

Rogen moraine ridges. Although the glacial landforms are the predominant features there are scattered patches of scoured bedrock and extremely large boulders incorporated into the till, primarily locally-derived gabbro and granophyre from the Middle

2 a b

Duluth

c

w o e fl ic

Fig. 1. (a) DEM of field area with ice flow direction. White line is Minnesota State Highway 1. (b) Field area (rectangle) in relation to regional geography. (c) Example of topographic map expression of Rogen moraine ridges , section lines indicate scale.. Proterozoic Duluth Complex. The northern border of the study site is evident from

digital elevation models as it essentially ends at the bedrock-controlled geomorphology

of the Boundary Waters Canoe Area Wilderness, whose scoured and jointed bedrock

provides for the abundant lakes (Fig. 1.). Up-ice of the Isabella area the ice was flowing

over crystalline bedrock of the Canadian Shield and Duluth Complex. Ice moving over

bedrock must move via the process of regelation and enhanced creep and thus flows slowly. The Rogen moraine is developed on thin till between the striated bedrock and the thick till of the Toimi drumlin field, which lies further down ice (Wright,1969, 1972).

The till of the Toimi ranges from 50 to over 200 feet thick as determined by rotasonic drilling (Hobbs, 1992). The till that makes up both the Rogen moraine and the drumlins is extremely bouldery and its texture is rocky sandy loam to loamy sand, with a grain size distribution of the less than 2mm fraction at 69% sand, 29% silt, and 2% clay

(Lehr and Hobbs, 1992). It is known as the Independence Till (Wright, 1969). Late

Glacial chronology is controlled primarily by the relative dating of landforms, with a few radiocarbon dates as anchors. It is uncertain when the Rogen moraine was formed because of the improbability of obtaining a date on subglacial sediments. The positions behind the Vermillion and Highland moraines can therefore be used to constrain a maximum formation to the period during the last glacial maximum in northeastern

Minnesota.

4 Glacial History

Historical Glacial Investigations in Northeastern Minnesota

Continental ice sheets advanced and retreated over the upper Midwest for much of the last million years. Most of Minnesota was covered by ice at some point except for the southeastern portion, part of the . Pre-Wisconsin glacial deposits are found in northeastern Minnesota, but not much is known about their distribution and history.

Northeastern Minnesota was continuously covered by ice from the earliest Late

Wisconsin ice advance at approximately 27-29ka (Clayton and Moran, 1982; Mooers and

Lehr, 1997) until about 11ka by the Rainy and Superior lobes of the LIS (Fig. 2.).

The earliest formal studies of glacial deposits in northeastern Minnesota were conducted by Upham (1894) who identified a series of moraines across the state. He identified the Vermillion moraine, as the 12th moraine, although he did not define its entire length. Winchell, as the first Minnesota state geologist, organized systematic mapping of the glacial geology of Minnesota. Along with Upham and others, Winchell

(1899) was one of the first to map large portions of northeastern Minnesota and describe the surficial deposits. Todd (1898) postulated two lobes of ice, the Lake Superior lobe, which flowed along the axis of Lake Superior, and the Red River lobe, which advanced from the west. Elftman (1898) suggested two lobes for the northeastern portion of

Minnesota because of observed till differences and provenances; he named these the

Superior and Rainy lobes. The Rainy lobe refers to the ice flowing from the Rainy River area.

Leverett (1932) proposed that northeastern Minnesota was glaciated by three separate lobes of ice. He suggested that the earliest drift in the area was the result of ice

5 from the Patrician ice center located in the Hudson Bay Lowlands between the Keewatin

and Labradorean ice accumulation centers, and that this drift was deposited by a lobe of

ice which was a combination of the Rainy and Superior lobes. He also proposed that a

later advance of both lobes resulted in the interlobate area to the east of Isabella.

In terms of the overall glacial history of Minnesota, the modern understanding

began with Wright, who identified multiple phases of glaciation (Wright, 1964) and was

the first to identify and interpret the tunnel valleys, drumlins, and eskers (Wright and

Ruhe, 1965). The Rainy and Superior lobes originated in the Labradorean sector of the

Laurentide Ice Sheet (Wright, 1972); The Superior Lobe followed the present Lake

Superior basin, which was filled with easily eroded sedimentary rocks derived from the

Midcontinent Rift System, whereas the Rainy Lobe flowed over northeastern Minnesota

(Fig. 2, 3). The overall advance and retreat of the Late Wisconsin Rainy Lobe was divided into two phases: the Saint Croix-Itasca and Vermillion, whereas the Superior lobe advanced and retreated in four major phases: the St. Croix, Automba, Split Rock, and

Nickerson. Wright (1972) placed the ages of the St. Croix and Vermillion phases at 20ka and 18ka respectively. Mooers and Lehr (1997) significantly revised the Late Glacial

chronology based on a synthesis of recent work and the regional and local radiocarbon

chronology (Fig. 2).

Hewitt phase

The earliest Late Wisconsinan advance of ice was that of the Rainy and Superior

Lobes and occurred at about 23 ka. The Rainy Lobe terminated at the Alexandria

moraine (Goldstein, 1986; Mooers and Lehr, 1997). This advance is interpreted to be

6

erior

Lake Sup Lake

Superior Lobe Superior aine

Inner Moraine

oraine Mor d

Rainy Lobe Rainy Highlan

Outer M aine

1992.

Isabella Mor Isabella

Toimi Drumlins Toimi Wampus Lake Moraine Lake Wampus

Arrows indicate ice flow direction. Modified from Lehr and Hobbs, and Hobbs, Lehr Modified from direction. flow ice indicate Arrows

Fig. 3. End moraines and other features in northeastern Minnesota. in northeastern and other features Minnesota. End moraines 3. Fig. Giants Range Giants Vermillion Moraine 5 miles 10 kilometers

Allen Moraine N

Wahlsten Moraine Wahlsten

erior ior Lobe ior

ession

N

N e

o N

Rec Super

o o Lake Sup Lake

49 49 e W 49 W o W o o

WI

WI

e 92

92 Lob Rainy 92

ior Lob ior

e

ny Lob ny

Lob 100km

ON Super 100km 100km ession Rai N ON N o N

ON

o o Rec Rainy W 43

W 43 o W 43

o o Itasca Lobe Itasca 95 95 95 IA MN IA IA MN MN c b a oines Lobe SD ND Itasca, (c) Vermillion- (c) Itasca, SD WI ND Red River - Des M Mooers and Lehr, 1997. Mooers and Lehr, SD Fig. 2. Wisconsin glacial Wisconsin 2. Fig. ND phases in Minnesota: (a) phases in Minnesota: Highland Modified from Highland Modified from Alexandria, (b) Saint Croix- Alexandria, contemporaneous with the Emerald Phase of the Superior Lobe (Mooers and Lehr, 1997;

Johnson, 2000). The relatively large size of the Alexandria moraine indicates that the

Rainy Lobe occupied this area for an extended period of time. While the retreat from the

Alexandria is not well documented, it associated with the formation of the Wadena drumlins (Wright, 1957; Goldstein 1986). The extent of ice recession is unknown but may have retreated as far as northern and northeastern Minnesota (Wright and Ruhe,

1965).

St. Croix Phase

The next major advance of the Rainy and Superior lobes terminated at the St.

Croix moraine, while the Itasca Lobe was at the Itasca moraine (Mooers and Lehr, 1997)

(Fig. 2). Associated with the St. Croix moraine are the Pierz and Brainerd drumlin fields.

Wright suggested that they were formed while the ice stood at the St. Croix moraine

(Wright and Ruhe, 1965; Wright, 1972). However, Mooers (1990) suggested that the drumlins were formed as the ice retreated, much the same as Goldstein (1986) proposed that the Wadena drumlins formed as the ice retreated from the Alexandria moraine. The recession from the St. Croix moraine is well constrained by both radiocarbon dates and landforms. A series of dates from an inter-drumlin depression suggests that the age of the

St. Croix phase is around 16-15.5ka (Clayton and Moran, 1982; Mooers and Lehr, 1997).

The retreat of the Superior Lobe resulted in the Pierz drumlin field, ice-marginal fans, tunnel valleys, and eskers; the retreat of the Rainy Lobe created the Brainerd drumlin field and hummocky end moraines (Mooers, 1988). It is clear that the Rainy Lobe

8 retreated in stages, each marked by a recessional moraine: Pleasant Lake, Stewart Lake,

Outing, Sandy Lake, Big Rice, and Wahlsten (Mooers, 1988; Lehr and Hobbs, 1992).

As the Rainy Lobe retreated further into northeastern Minnesota, the topographic high of the Giants Range influenced the ice margin and requires correlating the end moraines on either side. South of the Giants Range, the Big Rice moraine is correlated with the Allen moraine (Hobbs and Lehr, 1992; Lehr, 2000). The Allen moraine is interpreted (Lehr and Hobbs, 1992; Lehr, 2000) as a possible re-advance or reorientation of the Rainy Lobe margin because it truncates the Toimi drumlins. The Wahlsten moraine can be tenuously correlated with the Wampus Lake moraine (Lehr and Hobbs,

1992; Lehr, 2000). The ice retreated as far as the Canadian border before the next major advance (Wright, 1975).

Vermillion phase

The last major advance of the Rainy Lobe during the Late Wisconsin was the

Vermillion phase. During this phase the Rainy Lobe was at the Vermillion moraine, which can be traced over much of northeastern Minnesota, while the Superior Lobe was at the Highland moraine occupying the present-day Lake Superior basin. An approximate age for this phase is 11ka based on radiocarbon ages and correlations with other ice margins (Bjork, 1990; Clayton and Moran, 1982; Mooers et al., 2005a, 2005b).

East of the Giants Range, the Vermillion Moraine is possibly correlated with the

Isabella moraine of the Isabella sublobe (Lehr and Hobbs, 1992, Lehr 2000). The

Isabella sublobe has been postulated as an intermediate lobe between the Rainy and

Superior lobes to explain the mixture of traditionally Superior lobe materials into the till

9 near Isabella, but is better explained by the Rainy lobe incorporating sediment from the

Superior lobe. The major moraines attributed to the Isabella sublobe are the Inner, Outer, and Isabella moraines. The Outer moraine may be slightly younger than the Wampus

Lake moraine because it appears to truncate it (Lehr, 2000, Lehr and Hobbs, 1992).

Previous investigations of Isabella ribbed moraines

Friedman (1981) and Fenelon (1986), respectively, mapped the Isabella and

Cramer quads, which are located south and east of the Isabella moraines. These studies assessed the history of the Rainy and Superior lobes in the area and their associated end moraines, but do not cover the area in which the Rogen moraine occurs. Lehr and Hobbs

(1992) describe the Isabella moraines but do not comment as to a preferred mode of formation. Lehr (2000) concentrated on identifying ribbed moraines further west, north and just to the south of the Giants Range, but did not discuss the Isabella moraines.

According to Lehr (2000), ice encountering the topographic high and solid bedrock obstruction of the Giants Range would have flowed compressively and was also probably frozen to the bed at the margin (Clayton and Moran, 1982; Moran et al, 1980). Lehr

(2000) uses the original hypothesis of Shaw (1979) that Rogen moraine are the result of the development of folds in the debris-rich near-basal ice, and therefore as indicators of compressive flow, although no detailed explanation of mechanics is given. The moraines near the Giants Range are not well-defined and do not show up on topographic maps or

DEMs in the way that the Isabella area moraines do. While these other moraines are interesting they do not fit into the unique context along the Rainy lobe flowline that may

10 have led to the formation of the Isabella landforms, and I have to finish this thesis someday, so they are not considered in this study.

Rogen moraine

Rogen, or ribbed, moraine have intrigued researchers since the late 19th century, when they were first described by Hogböm (1885) in central Sweden. He did not postulate a method for formation, but indicated the varied structure and composition. The term “Rogen moraine” was coined by Hoppe (1959) after the type location near Lake

Rogen in Sweden. The first modern studies began with Lundqvist (1969) who was one of the first to suggest a subglacial origin for ribbed moraines and to note their association with drumlins. Since subglacial origins were first proposed, many different hypotheses have attempted to explain the details of this process. These hypotheses will be covered in detail later on.

There are a wide variety of shapes and forms that fall into the general category of

“subglacial ridges transverse to glacier flow.” Hättestrand and Kleman (1999) compile from the literature the most commonly cited characteristics of Rogen moraine (Table 1).

Most recently, Dunlop and Clark (2006) have assessed a large dataset of moraines from the Fennoscandian, Laurentide, and Irish Ice Sheets to determine whether they hold to the generally ascribed morphological characteristics of ribbed moraine (Table 1). Their analyses suggest that ribbed moraines have a far greater breadth of morphologies and geographic settings than previously described.

Geographic and topographic distribution

11 Rogen moraine are commonly found in, but not restricted to, the interior areas of continental ice sheets in North America and Europe (Hättestrand and Kleman, 1999;

Dunlop and Clark, 2006) (Fig. 3). Digital datasets revolutionized subglacial studies of remote continental regions that are hard to access otherwise. High-resolution datasets covering large expanses allow researchers to notice and analyze patterns that otherwise may not have been evident.

Rogen moraine were first observed in Scandinavia and much current research continues on these landforms. Moraines are found in Sweden, Norway, and in Finland

(Lundqvist, 1981; Hättestrand 1997, Hättestrand and Kleman, 1999; Sollid and Sørbel,

1994; Moeller, 2005; Punkari, 1984, Lundqvist, 1981; Sollid and Torp, 1982;

Hättestrand, 1997). Rogen moraine occur inland as well as in coastal settings

(Hättestrand and Kleman, 1999; Linden et al, 2008).

Rogen moraine in North America have been described in Canada and to a lesser extent in the United States. Rogen moraine are found throughout Canada, often in the interior regions covered by the (Bouchard, 1989; Dyke et al, 1992;

Shaw, 2002; Fisher and Shaw, 1992; Boulton, 1987; ). In the United States, only a few instances are identified in Maine and northern Wisconsin and Minnesota (Thompson and

Borns, 1985; Attig, 1985; Lehr, 2000; Lehr and Hobbs, 1992). These instances of Rogen moraine in the U.S. are much smaller in extent and occur closer to former ice sheet margins.

12 ? ?

Fig. X. Global distribution of known ribbed moraines. Light grey is extent of LGM; dark grey is areas of common ribbed moraine occurrence; triangles show outliers. Redrawn from Dunlop and Clark, 2006, and Hättestrand and Kleman, 1999.

Ribbed moraine formed by the Irish Ice Sheet Ireland differ from the classic type

found in Scandinavia and were only recently realized to be a significant aspect of the

glacial geomorphology of Ireland after digital datasets provided more high-resolution

imagery of landforms (Knight and McCabe, 1997). These ribbed moraines are larger

than other previously-described examples, in both the size of the landforms themselves and the extent of coverage (Knight and McCabe, 1997; Clark and Meehan 2001; Knight

2006).

There are no known Rogen moraine occurrences in the southern hemisphere, perhaps due to the lack of expansive continental ice sheets beyond Antarctica. There may very likely be Rogen moraine below the Antarctic Ice Sheet and improved remote sensing techniques may shed additional light on their formation.

Ribbed moraine fields distributed around the globe also are varied in their size, morphology, and topographic setting. Dunlop and Clark (2006) observe that the size of ribbed moraine fields can vary between a few square kilometers to thousands of square kilometers: these fields can be extensive and continuous, in packed or dispersed clusters, formed in ribbons or narrow tracks, isolated, or can even cross-cut one another. Rogen moraine are often cited as occurring in lowlands or depressions and on upstream-facing slopes (Hättestrand and Kleman, 1999). The topographic setting is considered in several formation hypotheses (i.e. Bouchard, 1989). However, Dunlop and Clark (2006) found that in their study areas the moraines were not confined to topographic lows and occurred on both upstream- and downstream-facing slopes.

14 Morphology

As compiled by Hättestrand and Kleman (1999), Rogen moraine ridges typically range in size from 300-1200m in length, 150-300m in width, and 10-30m in height.

Dunlop and Clark (2006) provide evidence from four different areas that ribbed moraine in fact have a far greater range in size, between 17 and 1116 m wide (mean 278m), 1-

64m high (mean 17m), 45-16214m long (mean 688m), and had wavelengths ranging between 12 and 5800m (mean 505m). Hättestrand and Kleman (1999) do not provide a range for wavelengths.

Hättestrand (1997a) classifies ribbed moraines into four types: “Classic” Rogen moraine (after Lundqvist, 1969a) features well-developed parallel arcuate ridges that are frequently curve down-glacier, fluted, and have steeper distal slopes; hummocky ribbed moraine are poorly-developed transverse ridges with less consistency in shape and size;

Blattnick moraine (Markgren and Lassila, 1980) are short and broad, convex up-glacier, may be drumlinized, and have less overall relief on the landscape; minor ribbed moraine are small and closely-spaced, straight and asymmetrical, and often found on slopes.

Dunlop (2004) and Dunlop and Clark (2006) describe a number of ribbed moraine shapes beyond the commonly cited “anastomosing, distal-side-steeper”: downstream curving, upstream curving, broad rectangular, blocky and angular, anastomosing, barchan shaped, broad arcuate, jagged, lumpy, lattice-structured, hummocky, minor, poorly developed, cross-cutting, and mega-scale (Fig. 5). Hättestrand and Kleman (1999) note that ribbed moraine ridges also seem to have an interlocking or “jigsaw” pattern (Fig. 6).

15 1 km 1 km 1 km 1 km Downstream-curving Upstream-curving Broad rectangular Blocky, angular ribbed ribbed moraine ridges ribbed moraine ridges ribbed moraine ridges moraine ridges

1 km 1 km 1 km 1 km Anastomosing ribbed Barchan-shaped ribbed Broad arcuate ribbed Jagged type ribbed moraine ridges moraine ridges moraine ridges moraine ridges

1 km 1 km 1 km 300 m Lumpy ribbed moraine Lattice-structured Hummocky ribbed Minor ribbed moraine ridges ribbed moraine ridges moraine ridges

5 km 5 km 1 km 1 km Poorly-developed Poorly-developed Mega-scale ribbed Cross-cutting ribbed moraine ridges ribbed moraine ridges moraine ridges moraine ridges

Figure X. Ribbed moraine morphological classifications within study areas. Arrow shows ice flow direction. Modified from Dunlop and Clark, 2006 A1 B1

35% extension 65% extension

A2 B2

Fig. x. Jigsaw matching of ribbed moraine ridges near Lake Rogen, Sweden showing hypothetical amounts of extension. Modified from Hättestrand and Kleman, 1999. Composition and structure

Ribbed moraines also possess a great variation in internal structure and composition (Hättestrand and Kleman, 1999). Many exhibit typical unsorted and unstratified till morphology, but others have glaciotectonic structures, slightly stratified layers of till, or may even contain interbedded fluvial deposits and shattered bedrock.

Hättestrand (1997a) summarizes many of the sedimentological and structural descriptions attributed to Rogen moraine (Table 3). Recently, work by Möller (2006) and Linden et al. (2008) concentrates on the interiors of Rogen moraine as the key to determining their mode of formation. Moeller (2006) describes the interior of moraines in Dalarna,

Sweden as debris flow and fluvial deposits which had been subsequently deformed into the Rogen moraine shape. Linden et al. (2008) excavate coastal ribbed moraines and describe a mixture of massive and stratified diamict and fluvial materials, with shear planes and other indicators of subglacial folding and thrusting.

Depending on the hypothesis of formation, the internal composition and structure may or may not matter. In hypotheses where the deformation mechanism comes from the overall behavior of the ice sheet, the exact nature of the bed material does not matter as it will deform independently. In other formation hypotheses where the local subglacial environment controls the development of the , internal structure and stratigraphy are important in determining the method of formation. Möller (2006) and others (Linden et al, 2008) stress the importance of the internal structure and composition on ribbed moraines as integral to determining their mode of formation and point to the varied interiors as evidence of polygenesis. Other researchers (Hättestrand and Kleman, 1999) state that mode of formation acts independently of the nature of the material because the

18 interiors of the moraines are so varied and there would need to be a way to explain them via an overarching theory.

Material Locality Reference Medium to coarse sand Quebec, Canada Ives (1956) Glaciofluvial sand, interbedded in Värmland, Sweden Lundqvist (1958) and basal till Gravely sediments Keewatin, Canada Aylsworth and Shilts (1989) Glaciofluvial gravel Lake Rogen area, Sweden Lundqvist (1937) Sandy-muddy gravel Newfoundland, Canada Fisher and Shaw (1992) Kalix till, slightly disturbed, Norrbotten, Sweden, Hoppe (1948); Fromm water-deposited sediments Värmland, Sweden (1965); Lundqvist (1969a) Sveg till, diamicton with layers of Jämtland, Sweden Shaw (1979) sorted material ('subglacial melt- out till') Stratified, matrix-supported, fine- Quebec, Canada Bouchard (1989) gained diamicton ('basal melt-out till') Sandy-silty till with sorted Lake Rogen area, Sweden Wastenson (1983) sediment lenses Sandy till (clayey till underneath Jämtland, Sweden Lundqvist (1969a) and in surrounding areas) Loose ablation till Värmland, Sweden Lundqvist (1958) Very hard-packed basal till Jämtland, Sweden Rasmusson and Tarras- Wahlberg (1951) Silty till with folded beds of Jämtland, Sweden Minell (1977) shattered bedrock Broken rock with little matrix Manitoba, Canada Dredge et al. (1986) Disaggregated sandstone, without Keewatin, Canada Aylsworth and Shilts fines (1989) 'Great variations in internal Sweden in general Lundqvist (1969a, 1995) structure and bedding' Table 1. Ribbed moraine compostion. From Hättestrand, 1997a.

Associations with other landforms

Ribbed moraines often occur with or near other subglacial landforms. The most common of these are drumlins, which was first noted by Lundqvist (1969). Because of this association with drumlins and other subglacial lineations, many researchers have hypothesized about the connection between the two landforms. Drumlins often occur down-ice from ribbed moraine. In some cases there are Rogen ridges which seem to be

19 in the process of elongation into drumlins, whereas in others there is no obvious transition between the two. Occasionally, drumlins may appear to be ribbed. Eskers always superimpose Rogen moraine, whether running through meltwater cuts in ridges, in between and parallel to ridges, or on top of ridges (Hättestrand and Kleman, 1999;

Dunlop and Clark, 2006). Because eskers and drumlins are features which occur some distance from the ice margin, this is considered by some as an indicator of ribbed moraine formation far from the ice margin (Hättestrand, 1997; Kleman et al, 1994). Relationships between ribbed moraine and other glacial landforms are summarized by Dunlop and

Clark (2006) (Fig. 7).

Glaciological setting and implications for ice dynamics

Rogen moraine are cited as a number of indicators for glacier behavior based on their geographic setting and orientation. Rogen moraine are assumed to be transverse to ice flow and thus, like drumlins, can be used as indicators of ice flow direction. They also are generally agreed to have formed under warm or poly-thermal basal regimes

(Möller, 2005). Some researchers (Hättestrand, 1997; Hättestrand and Kleman 1999) note their proximity to core areas of glaciation and inferred cold-based ice, but not all ribbed moraines follow this pattern (Dunlop and Clark, 2006). Ribbed moraines can be located anywhere between ice divides and ice margins, but not directly under ice divides

(Dunlop, 2004). Dyke et al (1992) showed evidence that ribbed moraine fields on Prince of Wales Island in Arctic Canada are located in the onset zone of an , and

Stokes and Clark (2003) and Dunlop (2004) note ribbed moraines superimposed on mega-scale lineations within the main track of ice streams. This superposition of the

20 landform suggests that the ribbed moraines were formed during or after the shut-down of the ice stream, and that they form at slower ice velocities than lineations (Dunlop and

Clark, 2006). Depending on the hypothesis for ribbed moraine formation, the implications for ice sheet behavior may vary considerably.

21 2km 2km 2km Drumlinized ribbed Abrupt lateral transition Abrupt lateral transitions moraine and exclusivity of and overprinting of landforms landforms

2km 2km 2km Downstream transition with Abrupt lateral and Ladder-type association overprinting of landforms downstream transition with mega-scale glacial lineations

2km 2km 2km 2km Mega-scale glacial Minor ribbed moraine superimposed Crag-and-tails with lineation "ribbed" into superimposed on on ribbed moraine tail "ribbed" into sequence of minor mega-scale glacial sequence of minor ribbed moraine lineations ribbed moraine

Illustration of downstream transition from mega-scale ribbed moraine to drumlinized ribbed moraine to clasic-type drumlins

Fig X. Relationships between landforms in study sites. Arrows show direction of ice flow. Modified from Dunlop and Clark, 2006.

Commonly Cited Characteristics as Compiled by Hättestrand and Kleman (1999) 1. Ridges are generally curved or concave downstream, or are anastomosing 2. Ridges often have horns that point down-ice 3. Ridges in cross-section are asymmetric with commonly a steeper distal side 4. Ridges have accordant summits 5. Ridges commonly have multiple subcrests, or flat crests giving ridges a tabular appearance 6. Have regular spacing between ridges which are typically of 300-1200 m, ridges are 150-300 m wide and 10-30 m high and are of similar size throughout a field 7. Ridges often display a “jigsaw puzzle matching” of their planform shape 8. Are most common on concave or flat surfaces in their terrain 9. Occur only in core areas of former glaciation 10. Commonly occur close to and distal of inferred frozen-bed areas 11. Are commonly drumlinized but are rarely found super-imposed on drumlins 12. Are composed of a variety of materials and commonly consist of locally derived sediment and which often display glaciotectonic structures

Revised Morphological and Spatial Characteristics from Dunlop and Clark (2006) 1. Ribbed moraine fields are formed at a wide range of scales and vary between a few square kilometers to extremely large continuous expanses of several thousand square kilometers. 2. Ribbed moraine fields manifest themselves as extensive continuous fields of large size; elongate ribbons and narrow tracks; densely packed or dispersed clusters; or as isolated fields, and in some instances as cross-cutting patterns. 3. Ribbed moraine ridges have complex plan view morphologies. They may be arcuate and concave down-ice, featuring downstream-pointing horns, but also arcuate and concave up-ice. May also be anastomosing, straight, rectangular, barchanoid, broad, hummocky, or can be poorly formed and lack a distinct morphology. 4. Ribbed moraine are nearly always asymmetric in cross-section and are just as likely to have a steep proximal side as a steep distal side. 5. Ribbed moraine ridges have undulating longitudinal crest profiles with multiple subcrests and resemble waves on the landscape. 6. Ribbed moraine fields consist of ridges whose length and height vary within the field, although neighboring ridges usually have similar dimensions. 7. In sample sets, ribbed moraine ridges varied between 17 and 1116 m wide (mean 278m), 1- 64m high (mean 17m), 45-16214m long (mean 688m), and had wavelengths ranging between 12 and 5800m (mean 505m) 8. Ribbed moraine ridges are commonly drumlinised and have close spatial associations with drumlins and mega-scale glacial lineations. There are spatial transitions between these forms, and ribbed moraines can be superimposed on glacial lineations and vice versa. In some rare instances, lineations may be “ribbed.” 9. Ribbed moraines are not necessarily restricted to core areas of glaciation or in a periphery around ice divides. They may possibly be generated at various times during a glacial cycle in broad swathes between the and some distance behind the margin. They are also found in the onset zones and within the main trunks of ice streams. 10. In general, ribbed moraines are not associated with any particular topographic setting (e.g. depressions) or preferentially occur on up or downstream facing slopes. Table 2. Ribbed moraine characteristics summarized from Hättestrand and Kleman (1999) and Dunlop and Clark (2006).

23 Formation Hypotheses Early hypotheses: Ice marginal and stagnant ice

Much early work (Frodin, 1913; Hogbom, 1920; Beskow, 1935) suggested that

Rogen moraine were ice-marginal structures, perhaps cyclical or annual end moraines

(Frodin, 1925; Henderson, 1959; Cowan, 1968) or as a result of calving margins in glacial lakes (Hughes, 1964). Lundqvist (1937) postulated that the ribbed pattern formed during dead ice wasting, as supraglacial sediment slumped into transverse .

Modern hypotheses: a subglacial landform

The evolution of thought that Rogen moraine were formed subglacially derived from observations on the nature of the till and associations with other landforms such as eskers and drumlins (Holmsen, 1935; Rasmusson and Tarras-Wahlberg, 1951; Kujansuu,

1967; Hoppe, 1952, 1959, 1968; Lee, 1959; Prest, 1968). Lundqvist (1969a) was the first to suggest a subglacial hypothesis for their formation, and ribbed moraines are generally agreed to have formed in the subglacial environment. The hypotheses for their formation can be divided into several categories: some favor one process that generates the moraines while others suggest that the moraines are pre-existing landforms that have been re-shaped into the ribbed moraine shape. Additionally, most hypotheses can be categorized into a formation process that involves extensional ice flow or one that relies on compressional ice flow. Dunlop and Clark (2006) note that as a collective landform, they must either be part of a bedform analogy, all subglacial transverse ribbed moraines form from the same process, or are all different landforms with different modes of formation and to lump them under one category is inaccurate.

24 Shearing and stacking due to compressive flow

Bouchard (1989) proposed that Rogen moraine are formed a few kilometers behind the ice margin under compressive ice flow by the shearing and stacking of till slabs (Fig. 8). As an ice sheet flows into a bedrock basin, it is compressed locally and erodes the basin, shearing off material that is entrained in the glacier bed and stacked up in slabs as the ice nears the margin, eventually deposited as basal melt-out till. After deglaciation the upwards-facing ends of these stacked till slabs would result in the ribbed topography. Bouchard (1989) suggests that ribbed moraines are characteristic landforms of glaciated areas like the Canadian Shield and similar scoured-rock surfaces and are related to the predominance of bedrock basins and resulting topography.

Two –step process/precursor ridges/polygenetic landform

Boulton (1987) suggested that the Rogen landscape’s common occurrence with drumlins indicates that it is formed in different stages of subglacial deformation from the glacier bed and that ribbed moraine and drumlins are just different end members of the subglacial landform assemblage. According to this hypothesis, Rogen moraine are thus an early stage of drumlin formation and occur due to previous ridges in the glacier bed, perhaps from an earlier ice flow direction. Lundqvist (1981, 1989, 1997), noting the large variation in Rogen moraine shapes, locations, internal structures, and sedimentological data, suggested that a two-step model for formation would allow for multiple methods of generating a pre-existing landform which would then be modified into the “classic” Rogen moraine shape (Lundqvist 1969).

25 a ice ice flow b facies 2 till bedrock c d stagnant ice facies 2 facies 3 rogen moraine fluted hummocky décollement local blocks moraine

Fig. x. Ribbed moraine formation (Bouchard 1986, 1989) by shearing and stacking of till slabs in bedrock basins. Redrawn from Bouchard, 1989. Möller (2005) suggests that Rogen moraine are reshaped ice-cored moraines from the ice-marginal zone. Similar to Hättestrand and Kleman, the hypothesis requires preservation of pre-Late Weichselian landforms by cold-based ice. As the ice-cored moraines wasted, the depressions between the moraines were filled in by the materials.

During a later phase of glaciation, the ridges were reshaped by advancing ice, elongating parts of the ridges into the horns and anastamosing ridges observed in many ribbed moraines. Eventually, the ridges would be elongated into drumlins (Fig. 9).

Linden et al. (2008) propose a two-step/three facies process for Niemisel-type

Rogen moraine and De Geer moraine. The origins of the pre-existing ridge differ from that of Möller (2005) and the proximal part of the moraines (“Proximal Element”) derive from folding and thrust stacking of englacial and subglacial materials. The distal part of the moraines (“Distal Element”) results from lee-side cavity fill with glaciofluvial and gravity-flow sediments. The Proximal Element and Distal Elment are said to originate from changes in bed rheology near the ice margin as the glacier encounters differing amounts of basal meltwater pressure due to changes in the melt season. The increase or decrease in water pressure results in a less mobile bed and initialization of a compressive flow regime, and the lee-side cavities open as a result of the glacier becoming separated from its bed. The third facies, the “Draping Element,” is a deforming bed till that covers the ridges in the last stage of formation, contributing to their final form and creating unconformities from presumed erosion of ridge tops. (Fig. 10)

27 Active glacier

Area of supraglacial re-sedimentation

Lodgement till Fluvial deposits debris band in stagnating/stagnant ice Debris-flow deposits Lacustrine deposits Buried stagnant ice Basal melt-out till Supraglacial melt-out till

Landscape inversion, producing moraine ridges and hummocks

Fig. X. Two-step model for Rogen moraine formation, redrawn from Moller (2005) based on Boulton and that other paper. T4

a deposition of draping facies

T3 stacking of proximal ridge facies and continuous lee-side cavity deposition

T2A T2B

Time

thrusting of pre-ridge and proximal sediments and cavity deposition

T1A T1B

initial phase of ridge formation - or thrusting of pre-ridge sediments folding A: ductile deformation B: brittle deformation

T0 ice/glacier subglacial deposition of pre-ridge formation sediments

ice flow direction

Architectural facies draping facies (DF)

distal ridge facies (DRF)

pre- and proximal ridge facies (PRF)

Conceptual model of seasonal b porewater pressure variability and implications for deforming bed mobility and formation of ribbed A a moraine. A. the principal summer piezometric surface and its annual b variability; (a) steep porewater

gradien tduring peak melt season pressure water c when the meltwater influx from winter the supraglacial catchment area is high; (b) the intermdiate state during start-up or closure of the sea surface melt season/drainage system, and winter (c) a low porewater gradient and B deforming bed - e low porewater pressure during the summer xtension winter season. B, D, The seasonal De Geer moraine variability of the deforming bed extension compression mobility (grey = deforming bed; C Ribbed moraine white = stiff bed) according to the porewater pressure in A. Ribbed moraine ridges are formed in the D transition zone between winter deforming bed and stiff bed <10km

Fig. 10. Rogen moraine formation after Linden et al., (2008). (a) shows facies diagrams; (b) models seasonal variations in pore water pressure and effect on landform generation. Polygenetic landform

Because the variety of landforms labeled “Rogen moraine” or “ribbed moraine” encompasses so many shapes, sizes, compositions, and glaciological settings, there is also a question of whether these are all actually the same landform. Some researchers suggest that because of this variety, not all “transverse subglacial ridges” form in the same manner (Lundqvist, 1997; Fisher and Shaw, 1992). Lundqvist (1997) notes that the two- step process is able to explain the variety of sediments that comprise Rogen moraine by allowing the original ridge structures to have formed in a number of ways, with the subsequent deformation by an overriding ice sheet being the “Rogenization” that defines the landform. Dunlop and Clark (2006) look to describe the many types of Rogen or ribbed moraine in order to draw attention to the variety that must be accounted for when attempting to apply any overarching theory.

Subglacial floods

Shaw (1983, 1996, 2002) and others (Fisher and Shaw, 1992) suggest that Rogen moraine and other subglacial features may be the result of subglacial megafloods. The flood waters, via sheet flow, carve the undersides of the ice sheet into streamlined and ripple patterns. (Fig. 11) Shaw (1983, 2002) cites the Channeled Scablands and other landforms from the catastrophic drainages of Missoula as analogous structures. In Shaw’s original hypothesis (1983), sediment infills the resulting cavities in the ice, creating drumlins and Rogen moraine. Fisher and Shaw (1992) describe internal stratification in Rogen moraine ridges from the Avalon peninsula that is contained within each landform, supporting a cavity fill hypothesis. Later, Shaw (2002) and Fisher and

30 ice bed with inverted erosional marks

subglacial meltwater sheetflow

Rogen moraine cavity fill bedrock erosional marks drumlins lodgement till

Fig. 11. Drumlins and ribbed moraine formation from catastrophic subglacial sheet flow. Modified from Shaw, 2002. Shaw (1992) suggest that while some instances of Rogen moraine may be depositional,

drumlins and other forms may be the result of water eroding previously-existing till

sheets.

Extensional fracturing (Hättestrand and Kleman)

Lundqvist (1969) was the first to attribute Rogen moraine to a extensional or brittle fracture mechanism and suggested that till-loaded ice, by nature less plastic, fractured into the ribbed moraine pattern due to tensional forces from local topography.

Because it is difficult to explain the source of the tensional forces given the varied topographic settings of ribbed moraine fields, he later disassociated himself from the theory.

Hättestrand (1997) and Hättestrand and Kleman (1999) propose that Rogen moraine are essentially “boudinized” till sheets that have fractured due to extensional forces at the base of the ice sheet (Fig. 12). During retreat, the transition from a cold- based (ice frozen to bed) to warm-based (ice sliding over bed) basal thermal regime, because of the shrinking of the cold-based area, generates ice acceleration and extensional flow. The frozen till sheet would thaw from the bedrock up as a phase- change surface moves through the subglacial bed separating thawed and frozen sediments. This would result in a still-frozen section of till sandwiched between the deformable thawed till and the deforming glacier bed. The brittle-ductile transition would occur at the phase-change surface between the thawed and frozen till. Thus, the still-frozen till sheet would develop extensional fractures as it was extended along the thinning and accelerating glacier. As these fractures opened due to continuing

32 0 + 0 + 0 + u u u a * * *

b ice * *

drift sheet

phase change surface

bedrock

ice flow Deformation a. Ice flow velocity profiles in the lower * behavior part of the ice mass for stages 1-3 in b inflow of ice Ductile c b. Time slice boxes showing successive Brittle evolution from a pre-existing drift Ductile sheet to a ribbed moraine

inflow of thawed till (shear zone) c. Close-up of fracturing zone showing Phase change surface/bedrock Bedrock fracturing process and the surface intersection (undeformable) deformational behavior ofthe layers Zone of extension

Fig. x. The extensional fracturing hypothesis for Rogen moraine formation. Redrawn from Hättestrand and Kleman, 1999. extensional flow, they would be filled by inflowing ice, leaving ribs of till. The ribs of till could be modified by the now warm-based glacier, perhaps eventually becoming drumlinized.

Subglacial deformation

Boulton (1987) suggested a deforming bed mechanism for subglacial landforms.

Since then other formation theories have expanded upon the idea. These formation theories are based mostly on numerical modeling of ice sheets and rely on flow instabilities as the trigger for landform genesis, many focusing on drumlins (Hindmarsh,

1998a, b, 1999, 2000). Aario (1977, 1987) proposes Rogen moraine formation under undulating ice motion at base of ice sheet, with drumlins and streamlined features forming from spiral ice flow. The changes in the type of ice flow would derive from a non-uniform slowing of the ice velocity as it entered the inner marginal zone, creating a drag effect. Other work in development indicates that Rogen moraine and other subglacial landforms may result from subglacial deformation that acts on a particular wavelength (Hindmarsh et. al, 2003; Dunlop and Hindmarsh, 2007). This hypothesis is known as the Bed Ribbing Instability Explanation (BRIE) and states that when internally deforming ice is sliding across till and the till is in turn deforming and sliding across the bedrock beneath, this creates small perturbations in the till surface. As these perturbations grow they migrate downstream and create the ribbed pattern in the till. The model predicts conditions and wavelengths for ridge formation. The model can then be applied to existing ribbed moraine knowledge to determine its validity.

34 Methods of Investigation

Field truthing/mapping

Fieldwork completed during the summer of 2004 set out to determine the extent and character of Rogen moraine ridges as identified from topographic maps and DEMs.

One goal of this exploration was to determine the extent at which the subglacial landscape was bedrock-controlled, thin till/Rogen moraine-controlled, thick till/drumlin- controlled, and any overprinting by ice-marginal features or eskers. Forest service roads and trails were investigated for likely exposures where Rogen moraine structure could be viewed in situ and the till could be sampled. Gravel pits, while often useful in most glacial geology fieldwork, proved mostly useless as they were only located in outwash or end moraines and not actually part of any Rogen ridges. Excavation of Rogen ridges proved extremely difficult due to the compact nature of the till. The most successful method of exposing a profile in the till was by chipping with the chisel end of a rock hammer—in effect brittle fracture. Two attempts at excavating a moraine in cross section were attempted; one at an exposure on an abandoned railroad grade, and another at the end of a suspected Rogen moraine form. Additionally, till in the Toimi drumlin field, while largely inaccessible, was sampled when roadcuts allowed.

GIS/spatial analyses

Digital datasets were an extremely useful tool in planning field explorations and visually determining the extent of geomorphological controls on the landscape, as well as relationships between glacial landforms in the area. Axes of Rogen moraine and drumlins were delineated in ArcView in order to determine a general ice flow direction

35 and the angular relationships between the landforms. An Avenue script was used to find the orientation of the line data for the Rogen moraine, bulk Toimi drumlins, and northern, middle, and southern sections of the drumlins. The orientations were saved as spreadsheet data and plotted in rose diagrams using Stereoplot.

Structural

Basic structural analyses were performed to determine the feasibility of Rogen moraine forming from extensional fractures and to produce an estimate for the amount of extension in the Isabella area. Mohr circles and Coulomb fracture criteria were created using estimates for basal shear stresses at the base of the Rainy Lobe and till cohesion.

Similar to analyses made by Ramsay and Huber (1983) using boudinized quartz veins, attempts to “unfracture” the Isabella moraines and estimate the amount of extension were made by fitting outlines of the moraines back together in Adobe Illustrator®. Assuming that this reconnected till sheet was in its original undeformed state, an undeformed strain ellipse was drawn over the “unfractured” till sheet. After moving the moraines back to their current locations, the new deformed strain ellipse was measured compared to the original to calculate percentage extension.

Grain size

Grain size parameters were determined using the pipette method for clay fraction and a sonicator for sorting out the coarse fraction (Folk 1974). Weight percentages for grain sizes were used to assign the till a textural classification. Till matrix grain sizes were plotted on ternary diagrams.

36

GPR

During the summer of 2006, ground-penetrating radar was used to characterize a subsurface transect of approximately 40m across one of the most easily accessible Rogen moraine crests on an old railroad grade using a Pulse Ekko 200 with 100kHz antennae.

This is the same exposure that was excavated, described, and sampled during the summer of 2004. The resulting profile was processed and then visually assessed for subsurface structures.

Results

Field

The field area can be divide up into four approximate geomorphic areas or landscape controls—scoured bedrock, thin till cover, end moraine, and thicker till cover

(Fig 13). The Rogen moraine are confined to the thin till cover and are mostly inaccessible from any roads—only the railroad grade provides access to the subsurface.

South of Kelly Landing Rogen moraine are less obvious and seem to widen in ridge wavelength, however they become obscured by the end moraine. Eskers follow depressions between moraine ridges and in a few places are draped over them, enforcing that they were deposited at a later date. ˚

Along the abandoned railroad grade one pit was excavated and sampled, which revealed the expected massive sandy subglacial till. At another location, which appeared to be the end of a ribbed moraine ridge, the resulting profile was entirely coarse sand and

37 gravel fluvial deposits. This most likely was a mantle of outwash at the end of a till ridge or may have been a completely different landform, as the site was located at the eastern edge of the Rogen landscape, nearing the complex junction of the Rainy and Superior lobes.

38 Fig. 13. Map of field area bedrock showing general geomorphic controls and sample sites.

ice flow Roads Grain size sample site

thin till 15 km 10 mi Rogen moraine end moraine

Minnesota State Hwy 1

thick till Isabella

drumlins a

c

Fig. 15. Photos of till from field area. (a) End moraine till showing significant red component. (b) and (c) Subglacial till from railroad grade cross-section of b Rogen moraine. GIS

The Isabella Rogen moraine have a mean azimuth of 297˚ and are nearly perpendicularly oriented to the drumlins (Fig. 14). The Toimi drumlin field gradually curves westward with a mean azimuth of 33˚ in the northern part and 45˚ in the southern part. The drumlins become more aligned further down glacier.

Grain Size

The till collected from the Isabella ribbed moraines and the Toimi drumlins is similar to other samples of Independence till, ranging from sand to silty sand. The average grain size for all till samples was 78% sand, 20% silt, and 3% clay (Fig. 15). The till collected from the Rogen moraine profiles was 82% sand, 16% silt, and 2% clay, with a slightly higher clay percentage from samples closer to the surface. Overall, the grain size is similar to other analyses of Rainy lobe subglacial till; Lehr (2000) presents average values of 76% sand, 22% silt, and 2% clay, with textures ranging from sand to silty sand.

41 Isabella ribbed moraines Northern third of Toimi drumlins

Southern third of Toimi Middle third of Toimi drumlins drumlins Clay

Silt

Sand

Fig. x. Grain size classification for till collected from Isabella ribbed moraines and Toimi drumlins. Triangle indicates average of all samples, shaded circle average of Rogen moraine samples. GPR

The extremely coarse-grained till proved to be an excellent medium for GPR.

The profile is riddled with reflectors due to the bouldery nature of the till. The profile contains 300 ns of data, with a velocity of 110 m/s, and yields a depth of 33 m. The velocity was determined by calculating a Common Mid-Point (CMP) from which the reflections originated. The results (Fig. 16) seem to show some sort of angular pattern on the down-ice end of the profile. These down-ice dipping structures are on the order of 10 m in height and could show bedding, draped glaciofluvial sediments, or migration of the landform.

Structural

If we assume Rogen moraine to be extensional features, we can apply the same techniques used to assess lithified extensional histories. The Mohr-Coulomb criteria should be able to predict what stresses are needed to fracture till. If Rogen moraine are essentially boudins, we should be able to put them back together and estimate a strain ellipse, as shown by Ramsay and Huber (1983) with filled quartz veins.

The driving force in subglacial deformation is the basal shear stress, or τb. Basal shear stress (τb ) is a function of the density of ice (ρ), gravitational constant (g), glacier height (h), and glacier slope (θ):

τb = ρghsinθ

The density of ice is 900kg/m3. Assuming an average basal shear stress for an ice sheet can also be used to back out its profile. If we assume a basal shear stress for the Rainy

44 Fig.X. (a) GPR profile for Rogen moraine crest. Ice flow is R - L. (b) Stratigraphic interpretations. Lobe in northeastern Minnesota at 26,000Pa, we can back out the height at a particular distance from the margin (x) by using a variant of the shear stress equation:

h = √(2τbx/ρg)

It is unsure at what time during the Rainy Lobe’s history that the Rogen moraine were forming, but if we assume that they formed during the lobe’s maximum extent, we can estimate a distance from the margin of 250km, which results in a glacier height of around

1200m. The other state of stress at the glacier bed is the normal force, σn.

σn = ρgh

The normal stress at the base of the Rainy Lobe near Isabella is 105,584,000Pa,

significantly higher than the shear stress. We can also determine the Coulomb criteria for this environment by estimating a cohesion (C) and angle of repose (θ) for a strong till:

τ = C + N(tanθ) where N = ice overburden

τ = C + N(ρi – ρw)(tanθ) ice pressure – pore fluid pressure

Cohesion for a strong till is about 250kPa, and an angle of repose typical for this value is about 35˚ (Benn and Evans, 1998). The normal stress must be tensile for extensional or

oblique fractures to open up. With such a high normal stress, in order for Rogen moraine

47 to be the result of extensional fractures, there must be an extremely high pore fluid

pressure to overcome the normal stress (Fig. 17).

Hättestrand and Kleman (1999) propose that one of the most convincing

characteristics that supports an extensional model for Rogen moraine formation is the

jigsaw puzzle-like matching of the ridges. They trace the outlines of Rogen ridges and attempt to fit them back together. However, they do not offer much insight as to a standardized technique for “unfracturing” till sheets, including what base level they used to define a ridge. Estimated amounts of extension from their work, depending on the area they investigated, are 35% and 60% (Fig. 6).

Similar work was attempted on the Isabella moraines by tracing a “best fit” ridge outline, placing them back together, constructing an undeformed strain ellipse, and moving the ridges back to their deformed position. This technique resulted in lower values on the order of 13% to 24% extension (Fig.18). Although not a precise calculation, it reinforces spatially that these ridges could have once been a continuous sheet, although in the case of the Isabella area the ridges do not match up as well as in the examples presented by Hättestrand and Kleman (1999). The ridges that appear to widen were also included in these calculations, although this apparent widening in ridge wavelength most likely is a geographic illusion and instead related to stream erosion of the end moraine. In any case, this technique is highly subjective.

48 +Ts, kPa

35o

Rainy Lobe basal shear stress (Tb) = 26kPa Rainy Lobe normal stress (tn)=10,558kPa Mohr-Coulomb criteria = 5253kPa

500

1000 5000 -tn, kPa +tn, kPa

zone of tensile failure

-Ts, kPa

Fig. 18. Mohr-Coulomb fracture criteria for subglacial till in Isabella ribbed moraines. a

b

24% 13%

Fig. 19. Determining hypothetical amounts of extension in Isabella ribbed moraines. (a) shows moraines re-assembled into a best-fit till sheet with undeformed strain ellipses. In (b) the moraines are moved back to their current positions and the ellipses are deformed in the direction of maximum extension. Discussion: applying current Rogen moraine knowledge to Isabella area

Now that we have covered the many physical and spatial characteristics for ribbed moraines and their hypotheses for formation, the task remains to assess if they can explain the features in the Isabella area and if this particular set of moraines is typical of those described globally.

The effort is to bring up into view every rational explanation of new phenomena, and to develop every tenable hypothesis respecting their cause and history. The investigation thus becomes the parent of a family of hypotheses; and, by his parental relation to all, he is forbidden to fasten his affection unduly upon any one. … The investigator at the outset puts himself in cordial sympathy and in parental relations (of adoption, if not of authorship) with every hypothesis that is at all applicable to the case under investigation. Having thus neutralized the partialities of his emotional nature, he proceeds with a certain natural and enforced erectness of mental attitude to the investigation, knowing well that some of his intellectual children will die before maturity, yet feeling that several of them may survive the results of final investigation, since it is often the outcome of inquiry that several causes are found to be involved instead of a single one. (Chamberlin, 1897)

The Isabella moraine dimensions are within those described by Hättestrand and

Kleman (1999) and Dunlop and Clark (2006), with a wavelength estimated from maps at about 350-400m, and some ridges appear to “fit” together. The ridges appear to have a shallower slope up-glacier and steeper distal slope, as also noted by Hättestrand and

Kleman (1999). Rogen moraine near Isabella, from the morphological classification scheme of Dunlop and Clark (2006) (Fig. 5), can be described as anastomosing, broad arcuate to rectangular, some with a slight upstream curve. Spatially related to other landforms, the Toimi drumlin field is downstream but does not appear to have an observable transition into drumlinized Rogen ridges, although the transition area may be obscured by ice-marginal sediments. If the drumlin directional data do indicate that the drumlins become better aligned further ice, there may be a transition from the Rogen moraine to less aligned to more aligned drumlins.

51 The Isabella rogen moraine are composed of extremely bouldery subglacial till of the Rainy Lobe. Some ribbed moraines interiors are massive subglacial till with little internal structure, and some have glaciotectonized sediments. The GPR data suggest that there may be some organization to the till in the Isabella moraines although when excavated the resulting profile suggested massive till.

The Isabella moraines thus do fit with other ribbed moraines worldwide and most likely share a similar mode of formation. Characteristics that may distinguish the Isabella moraines from “classic” Rogen moraine include the lack of drumlinized moraines and

“horns,” and no strong association with cold-based ice.

Ice marginal/recessional

The Isabella moraines show no evidence of being ice-marginal features. End moraines generally show an arcuate shape, representing the lobate nature of ice sheets.

The arcuate forms of the end moraines are not seen in the Isabella rogen moraine, although the Vermillion moraine in this area is not as classically rounded such as the margins at the Alexandria and Saint Croix moraines because it is subaqueous for much of its length. The Rogen forms show a general trend of mean azimuth of 297˚ and are nearly perpendicularly oriented to the drumlins, instead reflecting the ice flow direction over the greater region. Even if the Rogen moraine exhibited an arcuate shape, their angle is still not the same as that of the end moraines. Additionally the till texture and composition between end moraines and Rogen moraine in this area differ greatly. The till in the end moraines is much less compact, compared to the extremely dense and compact subglacial till. Furthermore the lithological differences suggest that these landforms are not analogous. Till in the obvious end moraines is thought to incorporate

52 outwash from the Superior lobe, which gives it a much higher content of reddish material.

Overall the orientation and till differences suggest that it would be highly unlikely for the

Rogen moraine in northeastern Minnesota to be ice marginal deposits from any late

Wisconsin stages. We cannot completely rule out that they were not ice-wasting features from earlier stages of glaciation that were then modified by an overriding ice sheet, as in

Möller (2005), but there is no evidence that any earlier glacier stagnated in the Isabella area. Additionally, because the till in the ribbed moraines and the till in the drumlins is the same from both grain size and field observations, the dense subglacial till is not likely to have originated in a marginal setting.

Two step process/precursor ridges

Boulton (1987) indicated that drumlinized Rogen moraine might be from a different ice direction. In northeastern Minnesota the ice-flow direction is well-defined by other landforms as northeast – southwest; thus neither the Isabella moraines nor the

Toimi drumlins are likely to be modified precursors to one another. The Isabella moraines lack the “horns” described on other examples of Rogen moraine (Lundqvist,

1989; Möller, 2005) and often cited when suggesting Rogen moraine are transitional landforms to drumlins. Additionally, if the Isabella moraines are a function of the bedrock to till transition, it seems unlikely that the drumlins are reshaped Rogen ridges, or at least ribbed moraine forming in the same manner due to the difference in size.

Although the Isabella moraines do not currently show any indication of being ice- marginal features and because relatively little is known as to their internal structures, it is possible that any evidence was erased by the deformation process. In the case of

Möller’s (2005) study sites, he was able to document a structure that was indicative of an

53 ice-marginal origin for the deposition of the sediments. The Isabella moraines seem to be composed of the same dense subglacial till as the Toimi drumlins, and at least in the few exposures available, do not show any stratification that would suggest the collapse of an ice-cored marginal feature. The till composition is also dissimilar to the end and interlobate moraines in the area, which have a component of rocks from the Lake

Superior basin. Again, as in the case of purely recessional features, the lithology points to the subglacial Independence till, but the ridges could have been deposited as ice-cored marginal features before the Isabella sublobe was incorporating Superior lobe materials.

Similarly, Linden et al. (2008) describe precursor ridges that would not be supported by the Isabella moraines based on their internal structure or bed conditions of the Rainy Lobe. Folding and thrust stacking of subglacial and englacial materials is not supported by the GPR profile, nor would the glacier, flowing off of scoured bedrock, encounter compressive flow conditions. Furthermore, precursor ridges are hard to generate in any situation because the glacier was flowing over bedrock and did not have a significant sediment source up-ice from which to deposit or shape ridges.

Shear and stack

This method would warrant more consideration if there was significant evidence of large shear planes and other significant structural indicators in the till. As it is, the potentially imbricated clasts would show the correct direction for shear failure, although the forms reflected in the GPR profile do not suggest the correct geometry for thrust planes along slabs of till. The primary issue in using this method of formation for the

Isabella landforms is that there needs to be a source of compressive flow. Lehr (2000) uses the association of Rogen moraine near the Giants Range as indicators of

54 compressive flow. Ice encountering the large topographic high of resistant bedrock might be subjected to compressive flow. However, in the Isabella area it would be difficult to rationalize with the landscape in the context of Bouchard’s (1989) model, where the bedrock basin provides the environment that generates compressive flow.

However, compressive flow is also common near the margin of the glacier. Depending on the timing of formation for the Isabella moraines, the shearing and stacking of till slabs could theoretically work if they were forming near the margin.

Polygenetic landform

There is no reason why different kinds of ribbed moraines cannot form in different manners. Perhaps there is a certain unifying theory that remains elusive to researchers thus far, but we cannot say that the Isabella Rogen moraine provide evidence for a universal mode of formation. They may only represent one type of ribbed or transverse moraine formation that involves a distinct change in glacier bed material. The unique bed material transition may characterize a certain type of ribbed moraine that may occur in other locations where an ice sheet moves from scoured bedrock to a thin till sheet. Alternatively they may show evidence of that unifying theory we have yet to discover. Similarly, there are many other ribbed patterns that occur transverse to types of flow besides ice: string-and-flark topography in wetlands, washboard road, and clouds.

As is noted by Lundqvist (1997), the two-step process allows for differing mechanisms to generate the initial till ridges. The origins of the Isabella moraines may not be entirely reflected by their current morphology.

Subglacial floods

55 While Shaw (2002) does include numerous examples of subglacial landforms that look very much like landforms or smaller-scale bedforms known to have been carved or deposited by water, there is still the question as to where such mass quantities of water would originate. There is no evidence for any large subglacial lakes in the Isabella area.

Proglacial lake sediments and local kettles are the only lacustrine areas.

Fisher and Shaw’s (1992) depositional model for Rogen moraine formation is also not a likely model for the Isabella moraines because it relies on stratification and water- deposited sediment, when the sediment that comprises the ribbed moraine and drumlins is subglacial till.

As Benn and Evans (1998) note, even though forms may be similar in shape, they may occur at much different scales and from different processes—streamlined and ripple marks occur in many environments: sand and snow dunes, as well as stream and glacier beds. This is important to keep in mind in examining Hättestrand and Kleman’s (1999)

“boudinage” of a pre-existing till sheet. Although Rogen moraine may share many of the same characteristics as boudins, they occur in vastly different environments.

Frozen bed transition, extensional fracturing

Hättestrand and Kleman’s hypothesis requires cold-based glaciation in order to provide the extensional flow necessary for the boudinage of a till sheet. However, there are instances where Rogen moraine do not follow this geographic distribution, as in the case of northeastern Minnesota. If Rogen moraine are the result of a boudinage of a pre- existing till sheet, there must be another mechanism to generate extensional flow in these instances. The Isabella area was very likely warm-based at the time of Rogen moraine formation. Eskers and striated bedrock would support this hypothesis, and there are no

56 landforms that indicate that cold-based conditions preceded warm-based ice. Therefore it

would seem that these indicators do not reflect a later phase of warm-based glaciation

following a bedform-preserving cold-based period, as in the case of an area noted by

Knight and McCabe (1997).

Within this narrow band of a few kilometers, the ice of the Late Wisconsin Rainy

Lobe would have experienced a change in the substrate and thus in the flow regime.

Bedrock slows down the velocity of a glacier and causes it to thicken. The glacier must slide via the process of regelation and enhanced creep, whereas when the Rainy lobe traversed from the hard bed to the thin till, be resistance would have decreased dramatically (Benn and Evans, 1998), and to maintain continuity of flow the ice would accelerate and thin.

Boudins are formed by the fracturing of a competent layer between weaker layers.

In the case of fractured Rogen moraine, the competent layer would have to be the till sheet. It would be hard to envision the till sheet alone as the competent layer between an ice sheet and bedrock. However, boudins can form a variety of geometries from both extension and shearing (Price and Cosgrove, 1990). Given the possibility of shear boudins, this does not require Rogen formation to be purely extensional. The core of the till sheet could be competent with two weaker shear planes at its top and base.

One feature of boudins that bear striking resemblance to the morphology displayed by Rogen ridges include the “map view” of boudins. One does not often see plane views of boudins, as they are usually pictured in the motion plane. The surface view offers the interesting observation that the ridges are not completely linear—they anastamose and look much like the curved nature of Rogen ridges (Fig. 20). It is also interesting to note that not all boudins are symmetric in cross-section—neither are Rogen

57 Top plan view of cut

Right side view of block

Left side view of block

Front view of block

hornblendite 0 2 4in hornblende-biotite- 0 5 10cm garnet gneiss

calcite

Fig. x. Boudinage of a hornblendite layer showing plan view similar to ribbed moraine pattern. Mineral lineation in host rock is parallel to the front face of the block. Redrawn from Price and Cosgrove, 1990 after Jones, 1959. moraine, which have a shallower up-ice slope. The morphologies and formation of rhomboidal boudins—mostly through shear failure—should be investigated more closely in terms of analogues for Rogen moraine. As in other formation hypotheses, it is unwise to rely too heavily on the mere appearance of similar structures.

In Hättestrand and Kleman’s (1999) hypothesis the bed is frozen, which would imply no pore fluid pressure, although it could exist below the frozen part of the till sheet.

In the Isabella area we are assuming that the bed is not frozen, which would allow for pore water to contribute to reduced shear stress. However, given the extreme compactness of the subglacial till and that in some places only the top foot or so has been leached, how much pore water could have been present?

Möller (2005) also attempts to determine whether the extensional fracturing hypothesis is valid from a soil mechanics perspective and comes to a similar conclusion, which is that extremely high pore fluid pressure would be necessary for failure to occur.

However, he does not appear to use a value for till cohesion, assuming frictionless and unfrozen till. Additionally, he notes that crevasses on a glacier surface heal as the ice flows out of the extensional environment and that similarly once the zone of thawing and extensional flow had moved up-glacier, the flow regime would likely change and could close up the fractures. In the case of the Isabella area, the zone of extensional flow is relatively stationary. As long as ice was flowing, it would remain extensional because of the change in substrate.

The main problem in comparing the Isabella ribbed moraine with boudins is simply accounting for the brittle-ductile transition zone. As in Hättestrand and Kleman

(1999) the ductile flowing ice would provide the upper weaker area. A ductile zone would also be required below the till sheet, which is problematic due to the crystalline

59 bedrock. To account for the till sheet as a competent layer, one must provide another incompetent layer. Could the till sheet have been permafrost with a thawed base or have had high pore water pressure? Lian and Hicock (2000) suggest that in examples of subglacial till from the , high pore water pressures can result in ductile deformation as a till sheet thickens. They also cite examples where the till has been so saturated that it may have flowed as a viscous slurry, entraining pieces of local bedrock. Dewatering of till previously deformed in a ductile manner results in stiffening and brittle deformation. Lian and Hicock (2000) do not provide a quantitative mechanical analysis to demonstrate their results, but cite a large number of directional data from till pebble orientations and deformation planes. If these observations can be extended to other subglacial environments such as the Isabella area, they suggest that a high pore water pressure is not needed to fracture a till slab, but it may contribute to ductile deformation. Cutler et al. (2000) also indicate permafrost in northeastern

Minnesota.

Boudins formed from shear failure are often rhomboidal and asymmetric compared to their extensional counterparts, and may appear as mini horst-grabens (Price and Cosgrove, 1990). Rogen moraine are typically asymmetric and not the flat

“chocolate bar” tablet structures of a traditional boudins. However, continued deformation in the direction of ice flow most likely affects their morphology, regardless of the initial mode of genesis.

Additionally, Dunlop and Clark (2006) note that ribbed moraine ridges occur on slopes independent of whether the slope was experiencing extending (downstream facing) or compressive (upstream facing) ice flow. In the Lake Rogen area 43% occur on slopes that would have experienced compressive flow, and 47% occur on slopes experiencing

60 extending flow. The Isabella moraines are on an adverse slope, and while this might suggest a compressive origin (Bouchard, 1987; Linden et al., 2008) it is clear that the aspect of the slope alone is not an indicator of compressional or extensional origin.

Subglacial deformation

These hypotheses, while fascinating, are still in early stages of development, and often do not have accessible ways of looking at data or using the model to create Rogen moraine in the Isabella areas. Because of the limited availability of specific information on these models, it is not yet possible to fully assess the Isabella moraines within this context. We can note with interest that in the BRIE model (Dunlop and Hindmarsh,

2007) the deforming till causes instabilities by sliding on underlying bedrock. In the

Isabella area this scenario is relatively easy to imagine, if not quantitatively or qualitatively evaluate, due to the thin till cover and evidence for warm-based ice. The

GPR profile could also support a migrating landform.

Additional Research and Questions

Further research would ideally include additional GPR of the Isabella area. The unique sandy character of the Independence till provides an opportunity that other with higher clay content would probably not provide. GPR transects would be ideally taken both up- and down-ice from the current profile. Superficial field examination from roads in the area seems to match up less-defined Rogen-like ridges with bedrock-cored ridges. Knowing whether the bedrock also shows a definite transverse ridge pattern would shed additional light on the nature of the deformation. Till pebble orientations would also be valuable in determining a variety of directional data.

61 Constraining the timing of formation might also help in determining the method of formation by indicating the behavior of the glacier or the proximity to the margin.

While in the Isabella area the overall nature of the subglacial bed would indicate extensional flow, Rogen moraine forming near the margin could potentially be a result of or at least modified by compressive flow. However, the dating of subglacial landforms is difficult and even if one were to determine a basal date from inter-moraine depressions, it would most likely only indicate the timing of the overall retreat of the ice sheet.

An ideal addition to this project would be to compile a global database of the bed materials and thicknesses at transitions between drumlins and Rogen moraine. Several studies (Hättestrand, 1997; Lundqvist, 1997) cite that ribbed landforms often contain large boulders of the local bedrock or may even in fact be cored by shattered bedrock.

This suggests a shallow till depth and it would be interesting to see if areas of scoured bedrock were present up-glacier, and consequently greater till depths down-glacier from the Rogen landscape—often also containing drumlins.

Conclusions

The Isabella ribbed moraines can be interpreted best as a transitional landform that forms at a significant change in the basal conditions. In the case of the Rainy Lobe, the change occurs as the glacier’s bed transitions from striated Canadian Shield bedrock to a thicker till cover and results in acceleration, eventually forming drumlins down-ice from the transition zone. Rogen moraine as a landform related to a slower ice velocity is also suggested by Dunlop, 2004; Dunlop and Clark, 2006; and Stokes and Clark, 2003, in these instances forming during or after a shut-down of a paleo-ice stream. Drumlins in

62 both these instances and in the Isabella area are linked to a faster-moving ice sheet. The change in bed rheology in the Isabella area might also cause a localized flow instability, much like in the BRIE model (Dunlop and Hindemarsh, 2007). Ribbed moraine formation in this area is unique because of this transition, which may be likely related to larger-scale behavior of the Laurentide Ice Sheet in eroding and cannibalizing its bed, resulting in a migrating bedrock-to-drumlinized till transition (Larson, 2007). With scoured bedrock up-ice, the sources for till to form the moraines are limited to the till at hand or any that has been transported from the erosion of the thin till sheet up-ice.

Overall the origins of the Isabella moraines are speculative, but can be used to rule out a few hypotheses for their formation as well as possibly support a few others.

What we do know regarding the moraines is that they feature an extensional flow environment/ice acceleration zone that is either constant due to a change in glacier bed material, or is moving up-glacier as a pre-existing till sheet is stripped. Any formation model must account for a thin till cover, dense and compacted sandy subglacial till, clinoform-like structures on the distal end of the moraines, large boulders, and a thawed bed. The modes of formation that are unlikely are any which suggest that the Rogen moraine are somehow related to ice-marginal structures, or that they are carved or deposited by large subglacial sheet floods. While there is no evidence against a two-step process for formation, there is no good known explanation for a precursor landform in the case of the Isabella ribbed moraines. The shear and stack method is not supported by the

GPR profile and does not work unless there is a bedrock basin and source of compressive ice flow to scrape out slabs of till, so this hypothesis is not likely. While the extensional fracturing theory is intriguing, it is difficult to make it viable in light of mechanics and the structural setting required for boudinage, although the glaciological conditions would

63 predict extensional flow. Theories of till deformation due to instabilities sound promising, especially in the case of the BRIE model (Dunlop and Hindemarsh, 2007), since a thin till sheet would allow for interaction with the underlying bedrock, but these models are still in development and it is thus hard to apply them to the Isabella moraines.

Worldwide, ribbed or Rogen moraine are a widely varying set of landforms with an equally varied set of hypotheses for their formation. There is evidence for ice-sheet scale control on formation, as opposed to local topographic or marginal processes, because there is no knowledge of them in alpine or smaller ice sheet settings. There probably exist similar conditions as described in the theories that describe local topographic or marginal settings (Möller, 2005; Linden et al, 2008) in alpine settings— why don’t we see Rogen moraine or significant drumlin swarms? If there are multiple ways that ribbed moraines are formed, there must still exist some reason that they only seem to form in continental settings.

I suggest that the formation of the Isabella moraines is directly related to the bedrock – thin till transition and may form as follows. The local acceleration and erosive behavior of the ice sheet results in transverse linear deformation of local till, through a combination of erosion and deposition. Previously-entrained thin till cover within the glacier from just up-ice contributes to the down-ice slope of the Rogen moraine ridges and their growth and migration downstream. This is supported by the abundance of local rock types and the structures shown by the GPR profile.

64 References Cited Aario, R., 1977. Classification and terminology of morainic landforms in Finland. Boreas v. 6, pp. 87–100.

Aario, R., 1987. Drumlins of Kuusamo and Rogen-ridges of Rauna, northeastern Finland. In: Menzies, J., Rose, J. (Eds.), Drumlin Symposium. A.A. Balkema, Rotterdam, pp. 87–101.

Attig, J.W., 1985, geology of Vilas County, Wisconsin. Wisconsin Geology and Natural History Survey, Informational Circular, v.50, 32pp

Aylsworth, J.M., Shilts, W.W., 1989. Bedforms of the Keewatin Ice Sheet, Canada. Sedimentary Geology 62, 407–428.

Benn, D.I, and Evans, D.J.A., 1998. and Glaciation. New York: Oxford University Press.

Beskow, G., 1935. Praktiska och kvartärgeologiska resultat av grusinventeringen i Norrbottens län. Geologiska Föreningens i Stockholm Förhandlingar, v.57, pp.120-123.

Bjork, S, 1990. Late Wisconsin history north of the Giants Range, northern Minnesota, inferred from complex stratigraphy. Quaternary Research, v. 33, pp.18-36.

Bouchard, M.A., 1989. Subglacial landforms and deposits in central and northern Quebec, Canada, with emphasis on Rogen moraines. Sedimentary Geology., v. 62, pp. 293-308.

Boulton, G.S., 1987. A theory of drumlin formation by subglacial sediment deformation. In Menzies, J. and Rose, J. (eds), Drumlin Symposium. Balkema, Rotterdam, pp.25-80.

Chamberlin, T.C., 1897, The method of multiple working hypotheses, Journal of Geology, v.5, pp. 837-848.

Clark, C.D., 1999, Glaciodynamic context of subglacial bedform generation and preservation, Annals of , v.28, pp.23-32.

Clark, C.D., and Meehan, R.T., 2001, Subglacial bedform geomorphology of the Irish Ice Sheet reveals major configuration changes during growth and decay. Journal of Quaternary Science, v.16, pp.483-496.

Clayton, L, and Moran, S.R., 1982, Chronology of late in middle North America. Quaternary Science Reviews, v.1, pp55-82.

Cowan, W., 1968. Ribbed moraine: till fabric analysis and origin. Canadian Journal of Earth Sciences v.5, pp.1145–1159.

65

Cutler, P.M., MacAyeal, D.R, Mickelson, D.M., Parizek, B.R., and Colgan, P.M., 2000, A numerical investigation of ice-lobe—permafrost interaction around the southern Laurentide ice sheet. Journal of Glaciology, v.46, pp.311-325.

Dunlop, P., 2004. The characteristics of ribbed moraine and assessment of theories for their genesis. Ph.D. Thesis, Department of Geography, The University of Sheffield, 363pp.

Dunlop, P and Clark, C.D, 2006. The morphological characteristics of ribbed moraine. Quaternary Science Reviews, v.25, pp.1668-1691

Dunlop, P., Clark, C. D., Hindmarsh, R.C.A, 2007, The Bed Ribbing Instability Explanation (BRIE) - Testing a Numerical Model of Ribbed Moraine Formation Arising from Coupled Flow of Ice and Subglacial Sediment. Geophysical Research Abstracts, v.9, European Geosciences Union.

Dyke, A.S., Morris, T.F., Green, D.E.C., and England, J., 1992. Quaternary geology of Prince of Wales Island, Arctic Canada, Geological Survey of Canada Memoir, 433, 142pp.

Dyke, A.S., 1993, Landscapes of cold-centred late Wisconsinan ice caps, Arctic Canada, Progress in Physical Geography, v.17, pp.223-247.

Elftman, A.H. 1898, The geology of the Keweenawan area in northeastern Minnesota, part I, glacial geology: The American Geologist, v. 21, p. 90-109.

Evans, David J.A., 2001. Glaciers. In Progress in Physical Geography, v.25, 3, pp.428-439.

Fenelon, J.M., 1986. Glacial geology of the Cramer quadrangle, northeastern Minnesota [M.S. Thesis]: Milwaukee, University of Wisconsin, 76pp.

Fisher, T.G., and Shaw, J., 1992. A depositional model for Rogen moraine, with examples from the Avalon Peninsula, Newfoundland. Canadian Journal of Earth Sciences. v. 29, pp. 669-686.

Folk, R.L., 1974, The petrology of sedimentary rocks: Austin, Texas, Hemphill Publishing Co., 182 pp.

Friedman, A.L, 1981, Surficial geology of the Isabella quadrangle, northeastern Minnesota [M.S. thesis]: Minneapolis, University of Minnesota, 66pp.

Frodin, G., 1913, Bidrag till vastra Jämtlands senglaciala geologi. Sveriges Geologiska Undersökning C, v.246, 236pp.

66 Frödin, G., 1925. Studien über die Eissheide in Zentralskandinavien. Bulletin of the Geological Institution, University of Uppsala, v.19, 214 pp.

Frödin, G., 1954. De sista skedena av CentralJämtlands glaciala historia. Geographica v.24, pp.1–251.

Goldstein, B.S., 1989, Lithology, sedimentology, and genesis of the Wadena drumlin field, Minnesota, U.S.A.: Sedimentary Geology, v. 62, pp. 241-277.

Hättestrand, C., 1997. Ribbed moraines in Sweden distribution pattern and palaeoglaciological implications. Sedimentary Geology, v.111, pp.41-56.

Hättestrand, C. and Kleman, J., 1999. Ribbed moraine formation. Quaternary Science Reviews., v. 18, pp. 43-61.

Hindmarsh, R.C.A., 1998a. The stability of a viscous till sheet coupled with ice flow, considered at wavelengths less than ice thickness. Journal of Glaciology 44, 288–292.

Hindmarsh, R.C.A., 1998b. Drumlinization and drumlin-forming instabilities: viscous till mechanisms. Journal of Glaciology 44, 293–314.

Hindmarsh, R.C.A., 1999. Coupled ice-till dynamics and the seeding of drumlins and bedrock forms. Annals of Glaciology 28, 221–230.

Hindmarsh, R.C.A, Dunlop, P., Clark, C.D., 2003. Modeling the geomorphological effects of till redistribution; assessing a dynamic theory for Rogen moraine formation and drumlin formation. Congress of the International Union for Quaternary Research., v. 16, pp. 123.

Hobbs, H.C., 1992, Description of rotasonic core from the Toimi drumlin field area, northeastern Minnesota. In Field trip guidebook for the glacial geology of the Laurentian Divide area, St. Louis and Lake Counties, Minnesota. St.Paul, Minnesota Geological Survey, Guidebook Series No. 18., pp.65-73.

Hobbs, H.C, and others, 1988, Surficial geologic map of the Greenwood lake, Isabella, and Cramer quadrangles, Minnesota: Minnesota Geological Survey Open File Report 88-2, scale 1:62,500, 9pp. text.

Högbom, A.G, 1885, Glaciala och petrografiska jakttagelser I Jemtlands län. Sveriges Geologiska Undersökning C, v.70, 39 pp.

Hogbom, A.G., 1920, Geologisk beskrifning öfver Jemtlands län. Sveriges Geologiska Undersökning C, v.140, 107 pp.

Hoppe, G., 1952, Glacial morphology and inland ice recession in northern Sweden. Geografiska Annaler 41, 193–212.

67

Johnson, M.D., 2000, Pleistocene geology of Polk County, Wisconsin. Wisconsin Geological and Natural History- Survey Bulletin, v.92, 70 pp.

Knight, J., 2006, Geomorphic evidence for active and inactive phases of Late Devensian ice in north-central Ireland. Geomorphology, v.75, pp.4-19

Knight, J. and A.M. McCabe, 1997, Identification and significance of ice-flow-transverse subglacial ridges (Rogen moraines) in northern central Ireland, Journal of Quaternary Science, v.12, pp.519-524.

Larson, P., 2007. Sediment transport cycles of the Laurentide Ice Sheet. (in prep)

Lehr, J.D., 2000, Pleistocene geology of the Embarrass area, St. Louis County, Minnesota. [M.S. thesis]: Duluth, University of Minnesota – Duluth, 157 pp.

Lehr, J.D. and Hobbs, H.C, 1992, Glacial Geology of the Laurentian Divide Area, St. Louis and Lake Counties, Minnesota. In Friends of the Pleistocene Field Trip Guidebook for the Glacial Geology of the Laurentian Divide Area, St. Louis and Lake Counties, Minnesota. Minnesota Geological Survey, St Paul, Minnesota. Guidebook Series No. 18

Leverett, F., 1932, Quaternary geology of Minnesota and parts of adjacent states: U.S. Geological Survey Professional Paper 161, 149 p.

Lian, O.B., and Hicock, S.R., 2000, Thermal conditions beneath parts of the last Cordilleran Ice Sheet near its centre as inferred from subglacial till, associated sediments, and bedrock. Quaternary International, v.68-71. pp.147-162.

Lundqvist, G., 1937. Sjösediment från Rogenområdet i Härjedalen. Sverige Geolgiska Underskoning C, v.408, 80 pp.

Lundqvist, J., 1969, Problems of the so-called Rogen moraine. Sveriges Geologiska Undersökning C, 648, 32 pp.

Lundqvist, J., 1997. Rogen moraine—an example of two-step formation of glacial landscapes. Sedimentary Geology v.111, pp.27–40.

Lundqvist, J., 1981. Moraine morphology – Terminological remarks and regional aspects. Geografiska. Annaler., 63 A: 127-138.

Lundqvist, J., 1989. Rogen (ribbed) moraine – identification and possible origin. Sedimentary Geology, v.62, pp. 281-292.

Markgren, M. and Lassila, M., 1980. Problems of moraine morphology: Rogen moraine and Blattnick moraine. Boreas, 9: 271-274.

68 Meyer, G.N., 1996, Geologic Atlas Stearns County, Minnesota, Part C, Text supplement, in Meyer, G.N., and Swanson, L., eds., County atlas series, Volume C-10, St Paul , Minnesota Geological Survey.

Möller, P., 2005, Rogen moraine: an example of glacial reshaping of pre-existing landforms. Quaternary Science Reviews, v.25, pp.362-389

Mooers, H.D. and Lehr, J.D., 1997, Terrestrial record of Laurentide Ice Sheet reorganization during Heinrich events. Geology, v.25, pp. 987-990.

Mooers, H.D., 1988, Quaternary history and ice dynamics of the St. Croix phase of Late Wisconsin glaciation, central Minnesota. Ph.D. Thesis, University of Minnesota, Minneapolis, 200pp.

Mooers, H.D., Marlow, L.M. and Larson, P.C., 2005a, Glacial lakes in northern Minnesota:drainage relationships, chronology, discharge, and isotopic composition of meltwater. Geological Society of America Abstracts with Program, Vol. 37, No. 5, p. 22.

Mooers, H.D., Larson, P.C., and Marlow, L.R., 2005b, Ice advances in the western Lake Superior region: a reevaluation of the St. Louis Sublobe and the Marquette Phase of the Superior lobe. Geological Society of America Abstracts with Program, Vol. 37, No. 5, p. 92.

Price, N.J., and Cosgrove, N.J., 1990, Analysis of geologic structures. Cambridge University Press, New York, 502pp.

Punkari, M., 1982. Glacial geomorphology and dynamics in the eastern part of the Baltic shield interpreted using Landsat imagery. The Photogrametric Journal of Finland, v.9, pp.77-93.

Punkari, M., 1984, The relations between glacial dynamics and the tills in the eastern part of the Baltic shield. In L.K. Königsson (Ed.), Ten Years of Nordic Till Research. Striae, v.20, pp.49-54

Ramsay, J.G., and Huber, M.I., 1983. The techniques of modern structural geology, volume 1: strain analysis. Academic Press, San Diego.

Shaw, J., 1977. Till body morphology and structure related to glacier flow: Boreas, v. 6, pp. 189-201.

Shaw, J., 1979. Genesis of the Sveg tills and Rogen moraines of central Sweden, a model of basal melt out: Boreas, v.8, pp. 409-426.

Shaw, J., 1983, Drumlin formation related to inverted meltwater erosional marks. Journal of Glaciology, v.29, pp.461-479.

69 Shaw, J, 2002, The meltwater hypothesis for sublgacial landforms. Quaternary International, v.90, pp.5-22.

Sollid, J.L. and Torp, B., 1984. Glacialgeologisk kart over Norge, 1:1,000,000. Nasjonalatlas for Norge. Geografisk Institut, University of Oslo.

Sollid. J. L. and Sørbel, L. 1994: Distribution of glacial landforms in southern Norway in relation to the thermal regime of the last continental ice sheet. Geografiska Annaler. Series A, Physical Geography, v.76, pp. 25-35.

Stokes, C.R., Clark, C.D., 2003. The Dubawnt Lake Palaeo-Ice Stream: evidence for dynamic ice sheet behaviour on the Canadian Shield and insights regarding the controls on ice-stream location and vigour. Boreas, v.32, pp.263–279.

Thompson, W.B, and Borns H.W., 1985, Surficial geologic map of Maine, 1:500:000. Maine Geological Survey, Department of Conservation.

Todd, J.E., 1898, A revision of the moraines of Minnesota: American Journal of Science (ser. 4), v. 6, p. 469-477.

Upham, W., 1894, Preliminary report of the field work during 1893 in northeastern Minnesota, chiefly relating to the glacial drift, in Winchell, N. H., Geological and Natural History Survey of Minnesota, 22nd Annual Report, for the year 1893, pp.18-86. van der Meer, J.J.M., Menzies, J., and Rose, J., 2003, Subglacial till: The deforming glacier bed: Quaternary Science Reviews, v. 22, pp.1659–1685.

Winchell, N.H., 1899, The geology of the north part of St. Louis County, in Winchell, N.H., The geology of Minnesota, Volume IV of the final report: Geological and Natural History Survey of Minnesota, p. 222-265.

Wright, H.E., Jr., 1957, Stone Orientation in Wadena Drumlin Field, Minnesota. Geografiska Annaler, v.39, pp.19-31

Wright, H.E. Jr., and Ruhe, R.V., 1965, Glaciation of Minnesota and Iowa, in Wright, H.E., Jr., and Frey, D.G., eds., The Quaternary of the United states: Princeton, Princeton University Press, pp.29-41.

Wright, H.E., Jr, 1972, Quaternary History of Minnesota in Sims, P.K. and Morey, G.B., eds, Geology of Minnesota, a centennial volume: St. Paul, Minnesota Geological Survey, pp.515-547.

70