Ribbed moraine formation

Quaternary Science Reviews 18 (1999) 43—61

Ribbed moraine formation Clas Ha¨ttestrand, Johan Kleman Department of Physical Geography, Stockholm University, S-106 91 Stockholm, Sweden

Abstract Ribbed (Rogen) moraines are conspicuous landforms found in interior parts of formerly glaciated areas. Two major theories for ribbed moraine formation have been suggested in recent years: (i) the shear and stack theory, which explains ribbed moraine formation by shearing and stacking of till slabs or englacially entrained material during compressive flow, followed by basal melt-out of transverse moraine ridges, and (ii) the fracturing theory, according to which ribbed moraines form by fracturing of frozen pre-existing till sheets, at the transition from cold- to warm-based conditions under deglaciating ice sheets. In this paper, we present new data on the distribution of ribbed moraines and their close association with areas of frozen-bed conditions under ice sheets. In addition, we show examples of ribbed moraine ridges that fit together like a jig-saw puzzle. These observations indicate that fracturing and extension of a pre-existing till sheet may be a predominant process in ribbed moraine formation. In summary, we conclude that all described characteristics of ribbed moraines are compatible with the fracturing theory, while the shear and stack theory is hampered by an inability to explain many conspicuous features in the distribution pattern and detailed morphology of ribbed moraines. One implication of the fracturing theory is that the distribution of ribbed moraines can be used to reconstruct the extent of areas that underwent a change from frozen-bed to thawed-bed conditions under former ice sheets.  1998 Elsevier Science Ltd. All rights reserved.

1. Introduction Ribbed moraines are characteristic features of the interior parts of formerly glaciated areas in the Northern Hemisphere. Despite this, ribbed moraines are rarely used in reconstructions of past ice sheets, unlike for example, drumlins that are used to infer flowlines of wet-based ice sheets, or end moraines to infer ice-marginal positions. This results from insufficient knowledge of the glaciological conditions and processes involved in the ribbed moraine formation. Therefore, we have not been able to interpret the distribution pattern and other characteristics of ribbed moraines in terms of glacier dynamics. Given a more credible model of formation, ribbed moraines are a potentially valuable source of information when reconstructing paleo-ice sheets. In this paper we will review proposed theories of formation, describe the characteristics of ribbed moraines, and finally evaluate the theories against available observations on ribbed moraine characteristics. Ribbed moraine is defined here as an area with large, regularly and closely spaced, moraine ridges consisting of

 e-mail: [email protected]

glacial drift, usually till (Fig. 1). The ridges are mostly curved or anastomosing, but their general orientation is transverse to ice flow. The term ‘ribbed moraine’, first used by Hughes (1964), is preferred over Rogen moraine, because the latter, according to the original definition by J. Lundqvist (1969a), only applies to those ribbed moraines that show the presence of drumlinoid elements or superimposed fluting (see also J. Lundqvist, 1981). However, Lundqvist’s definition excludes all non-drumlinised ribbed moraines, and as it is hitherto not shown that the drumlinisation is directly linked to the ridge construction, all ribbed moraines will here be treated as one group. Ha¨ttestrand (1997) showed that different morphological varieties of ribbed moraines in Sweden, such as Rogen moraine and Blattnick moraine have very similar distribution patterns and orientations. Ha¨ttestrand therefore suggested that the basic controlling factors of their formation were similar and that various types of ribbed moraines should be treated as one genetic type of landform.

2. Formation hypotheses We will in this section review papers dealing with morphologies that we interpret (or for Sweden, have

0277-3791/99/$ — see front matter  1998 Elsevier Science Ltd. All rights reserved. PII: S 0 2 7 7 - 3 7 9 1 ( 9 7 ) 0 0 0 9 4 - 2

44

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

Fig. 1. Ribbed moraine in Ha¨rjedalen, west-central Sweden. Ice movement was from right to left. Photo: Arne L. Philip

mapped) to be ‘true’ ribbed moraines, although the authors of these papers frequently have used other terminologies. For a review of the terminology of ribbed moraines, see J. Lundqvist (1969a, 1981, 1989), J. Bouchard (1989), Bouchard et al. (1989), and Ha¨ttestrand (1997). Ribbed moraines were first described in central Sweden by Ho¨gbom (1885, 1894), who did not offer an explanation to their formation, but noted the highly varied internal composition. In the first theories on the formation of ribbed moraines, it was suggested that they had a frontal origin, as series of end moraines (Fro¨din, 1913, 1925; Ho¨gbom, 1920; Beskow, 1935). Several later studies have also proposed a marginal or near-marginal formation (Fro¨din, 1954; Lindqvist and Svensson, 1957; Henderson, 1959; Craig, 1965; Fromm, 1965; Cowan, 1968), possibly in association with calving ice margins in glacial lakes (Hughes 1964). A marginal explanation was rejected by G. Lundqvist (1935, 1937, 1943, 1951), who argued that the large quantities of material found in ribbed moraines were unlikely to be deposited by active ice close to the last small ice remnants, where the most proximal ribbed moraines are found (with respect to deglacial ice flow). Lundqvist instead favoured a dead-ice explanation where supraglacial material slumped into transverse crevasses, forming the ribbed pattern of the moraines. Later, Mannerfelt (1942, 1945), Granlund

(1943), Tanner (1944), and Kurimo (1980) followed this idea of ribbed moraine formation during areal stagnation towards the end of glaciation. Hoppe (1948) invoked a near-marginal deposition of material followed by remoulding by smaller ice-front oscillations for the Kalixpinnmo ridges in northern Sweden, as their interiors clearly showed signs of glacial deformation. Hoppe (1959) was also the first to introduce the term ‘Rogen moraine’, named after the type locality around Lake Rogen in west-central Sweden. Theories of a subglacial origin were put forward as further investigations showed subglacial characteristics of ribbed moraines (J. Lundqvist, 1969a). Such observations included (i) lodgement character of the till (Holmsen, 1935; Rasmusson and Tarras-Wahlberg, 1951; Kujansuu, 1967), (ii) eskers overlying ribbed moraine (Hoppe, 1952, 1959), and (iii) drumlinisation and fluting of the ridges (Lee, 1959; Hoppe, 1968; Prest, 1968). More recent studies, over the past 20 years, almost exclusively invoke a subglacial origin. Most of these studies explain ribbed moraines as formed by shearing and stacking of slabs of near-base englacial or subglacial debris, due to localised compressive stresses, followed by subglacial melt-out of the till ridges. Boulton’s (1970a, b, 1971, 1972) model of entrainment and deposition of englacial debris in modern glaciers is often used as a reference for this process. The shearing and stacking theory is

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

45

Fig. 2. Formation of ribbed moraine, according to Bouchard (1986, 1989), by processes of shearing and stacking of till slabs under compressive ice flow, as basal ice flows towards the down-ice end of rock basins. (Reprinted from Sedimentary Geology, 62, Bouchard, M., Subglacial landforms and deposits in central and northern Quebec, Canada, with emphasis on Rogen moraines, 293—308, 1989, with kind permission of Elsevier Science—NL, Sara Burgerhartstraat 25, 1055 KV Amsterdam, The Netherlands.)

most extensively outlined in Shaw (1979) and Bouchard (1980, 1989) (Fig. 2), but is also suggested by Lee (1959), Kurimo (1977), Minell (1977, 1980), Shilts (1977), Markgren and Lassila (1980), Punkari (1982, 1984), Sollid and S+rbel (1984, 1990, 1994), Dredge et al. (1986), Aylsworth and Shilts (1989a, b), Bouchard and Salonen (1989) and Dyke et al. (1992). These papers differ primarily in terms of the origin of the compressive flow and the glaciodynamic environment in which ribbed moraines are suggested to form. Bouchard (1980, 1989), Minell (1980), and Sollid and S+rbel (1984) suggested compressive flow when the ice meets a topographic obstruction at the down-stream end of rock basins. Dredge et al. (1986) and Aylsworth and Shilts (1989b) argued that high concentrations of near-basal debris cause a decrease in the plastic behaviour of the ice, inducing compression and basal shearing of debris-rich ice, possibly in association with a low ice-surface gradient and climatic deterioration. Sollid and S+rbel (1984, 1990, 1994) put the ribbed moraine formation in an environment far from the ice margin, where patches of trapped, water-soaked debris are entrained by freezing on to the glacier sole and, subsequently, sheared up into ridges during an expansion of a frozen core area under the ice sheet. Similar conditions, but closer to the ice margin, are proposed by Punkari (1984) and Bouchard and Salonen (1989), who infer a frozen outer margin of the ice sheet, causing the compressive stresses leading to shearing and stacking. Dyke et al. (1992) inferred ribbed moraine formation in a transition zone from cold- to warm-based conditions. In this zone, alternate sticking and sliding conditions were suggested to cause infolding and stacking of basal debris. A few other recent formational theories differ from the above-described shearing and stacking model. J. Lundqvist (1989, 1997) suggested that the primary ridge structure in ribbed moraines reflects pre-existing features, later reshaped to various degrees by the last ice

sheet into drumlinised elements of Rogen moraine. However, the origin of the primary ridges is not specified. Boulton (1987) put forward a similar theory, and suggested that the pre-existing ridges could either have been subglacial folds structures, or possibly, drumlins oriented at right angle to the secondary ice-flow direction. Aario (1977a, b, 1987) related ribbed moraine formation to primary till deposition under a wavy ice motion at the base of the ice sheet, while the drumlinoid elements are thought to result from basal spiral ice flow. Fisher and Shaw (1992) suggested that ribbed moraines are part of a whole meltwater-flood landform assemblage, formed during subglacial outburst sheet floods. The ribbed moraine is suggested to form as ripple-like cavity fills. Finally, a few theories have been put forward that link ribbed moraine formation to extensional flow at the base of the ice sheet. In his original treatise on Rogen moraine, J. Lundqvist (1969a) suggested that less plastic till-loaded basal ice fractured into transverse elements by tensional forces induced primarily by topography. J. Lundqvist (1981, 1989) later dissociated himself from this theory because it was difficult to explain the origin of tensional forces. However, Ha¨ttestrand (1997) proposed that tensional stresses were coupled to the contraction of a frozen-bed core area of a retreating ice sheet. At the transition from proximal frozen (non-sliding) conditions to distal melting (sliding) conditions, high tensional stresses and extensional ice flow will occur, as the basal ice velocity increases across the boundary of basal thermal regime. Ha¨ttestrand (1997) suggested that these tensional stresses lead to detachment and ‘boudinage-like’ fracturing of a pre-existing drift sheet into ribbed moraine (Fig. 3). A key element in this theory is the presence of alternating competent and incompetent materials, exhibiting brittle and ductile deformation characteristics, respectively. In contrast to J. Lundqvist (1969a), who inferred the formation of ribbed moraines from basal

46

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

Fig. 3. Model of ribbed moraine formation from Ha¨ttestrand (1997), showing the fracturing of a pre-existing till sheet during the transition from coldto warm-based conditions under a deglaciating ice sheet. At this transition, basal ice velocity increases as the ice starts to slide over and deform its bed. This highly localised increase in ice velocity (acceleration) induces extensional ice flow. The phase change surface (a pressure melting isotherm), separating frozen material above from thawed below, rises through the bedrock/drift/ice sequence during thawing of the ice sheet bed. When the phase change surface is located in the lower part of the drift sheet, still-frozen drift is underlain by a deforming layer of thawed drift and overlain by deforming basal ice. Under extensional flow, the frozen upper part of the drift sheet breaks up into ribs in a boudinage-like fashion. (a) Ice flow velocity (u) profiles in the lower part of the ice mass for the stages 1—3 in b. (b) Time slice boxes (1—3), showing the successive evolution from a pre-existing drift sheet to a ribbed moraine. Detachment of frozen drift ribs will start to occur when the pressure melting isotherm intersects the bedrock surface. (c) Close-up of the fracturing zone showing the fracturing process and the deformational behaviour of the layers. During, and after, deglaciation, mass movement processes decrease the slope angles of the ridges and degrade their tabular morphology. The ribbed moraine fields form successively as the cold-/warm-based transition zone migrates up-glacier. (Reprinted from Sedimentary Geology, 111, Ha¨ttestrand, C., Ribbed moraines in Sweden—distribution pattern and paleoglaciological implications, 41—56, 1997, with kind permission of Elsevier Science—NL, Sara Burgerhartstraat 25, 1055 KV Amsterdam, The Netherlands.)

debris-loaded ice, Ha¨ttestrand (1997) explained ribbed moraine ridges as fractured subglacial sedimentary sequences.

3. Characteristic of ribbed moraine 3.1. Distribution 3.1.1. Global distribution Ribbed moraines are restricted to the core areas of the late Pleistocene glaciations in the Northern Hemisphere

(Figs 4 and 5). Specifically, they are found in the Keewatin and Quebec sectors of the Laurentide ice sheet, on Newfoundland, and in Norway, Sweden and Finland (Prest et al., 1968; J. Lundqvist, 1981; Punkari, 1984; Sollid and Torp, 1984; Ha¨ttestrand, 1997). In addition, a few outliers of ribbed moraine have been found outside these areas, such as in Maine, USA (Thompson and Borns, 1985), Wisconsin, USA (Attig, 1985), and on Prince of Wales Island, Arctic Canada (Dyke et al., 1992). So far, ribbed moraines appear to be absent in other formerly glaciated areas, and they have not been found in association with present-day glaciers or ice sheets.

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

Fig. 4. World-wide distribution of ribbed moraines (modified from Ha¨ttestrand, 1997). Light shaded—Last Glacial Maximum ice sheet distribution. Dark shaded—areas where ribbed moraine occurs commonly (after Prest et al., 1968; Sollid and Torp, 1984; Nordkalott project, 1986; Ha¨ttestrand, 1997; Mid-Norden Project, in press). In addition, a few outliers of ribbed moraine have been found outside the core areas (marked with triangles; e.g. Attig, 1985; Thompson and Borns, 1985; Dyke et al., 1992). The position of the last ice spreading centres of the Fennoscandian and the Laurentide ice sheets are marked with an L.

As seen in Fig. 4, ribbed moraines occur in only about 10% of the area covered by the Laurentide Ice Sheet and in 20% of the area covered by the Fennoscandian Ice Sheet. The limited distribution, often with a sharp boundary to areas lacking ribbed moraines (Fig. 5), does not appear to coincide with a specific topography, bedrock geology, surficial geology, the position of the marine limit, or the ice-marginal position at a specific time (Ha¨ttestrand, 1997). For example, the ribbed moraine limit coincides approximately with the 12 ka BP ice marginal position in Newfoundland, the 10 ka ice margin in Fennoscandia, the 9 ka ice margin in Keewatin, and the 8 ka ice margin in Quebec. Therefore, it is unlikely that the ‘start’ of the formation of ribbed moraines at the last deglaciation was induced by regional or global climatic changes, as has been suggested by Aylsworth and Shilts (1989a). Rather, it appears likely that the distribution pattern is controlled by specific conditions or events in the subglacial environment. 3.1.2. Proximity to cold-based areas The most abundant and best developed ribbed moraine areas occur in close connection to areas where the ice sheet was cold-based during the last glaciation (Kleman et al., 1994, in press; Sollid and S+rbel, 1994; Ha¨ttestrand, 1997), whereas areas of glaciation that bear no sign of frozen-bed conditions lack ribbed moraines. Fig. 6 shows the ribbed moraine distribution in Fenno-

47

scandia (from Fig. 5) compared with the reconstruction by Kleman et al. (in press) of the minimum extent of the frozen-bed core area during the last glacial maximum (LGM). The Kleman et al. reconstruction is based on the distribution of pre-late Weichselian landforms and deposits. The match between these two areas is striking. On an ice-sheet scale, no other glacial feature is nearly as closely correlated spatially with the ribbed moraines. On a regional scale, ribbed moraines are located in areas that experienced a transition from cold- to warm-based conditions, as the frozen-bed core areas diminished in size and warm-based zones migrated inwards towards the deglaciation centres (Kleman et al., 1994, in press). In areas that were continuously cold-based during the final retreat of the ice sheet, no ribbed moraines exist, apart from some with deviating directions and which are most likely correlated with earlier deglaciations (Ha¨ttestrand, 1997). Fig. 7 shows the glacial geomorphology in an area in northern Quebec. The westerly section shows a classical deglaciation landscape with well-developed drumlins and eskers, indicating a west—southwesterly ice-flow direction. In the easternmost area, large north-trending drumlins appear, but overprinted flights of lateral channels show that the receding ice sheet continued to retreat with an ice flow towards the west—southwest. The general lack of subglacial landforms overprinted on the pre-existing large north-trending drumlins is interpreted to indicate that the ice sheet was cold-based during its retreat across this area. Ribbed moraines are absent in the westernmost (distal) area and increase in density towards the change in basal temperature regime, and are absent again in the easternmost, cold-based, area. Kleman et al. (1994) argued that the frozen-bed core area of the Quebec dome was in a shrinking phase during the deglaciation, and that ribbed moraines therefore are situated in the zone that underwent a transition from cold- to warm-based conditions. A similar relation between ribbed moraines and frozen-bed zones was observed by Dyke et al. (1992), on Prince of Wales Island in Arctic Canada (Fig. 8). The geomorphological record of this island indicates that substantial and rapid shifts in ice-flow directions and basal thermal regime occurred. In the southeastern part of Prince of Wales Island, a spectacular convergent drumlin field, the Transition Bay drumlin field, indicates the presence of a late-glacial ice stream. The margins of this ice stream were not confined by topography, because the whole area has a relatively low relief. Dyke et al. (1992) interpreted the margins to be related to cold-based zones bordering the wet-based ice stream. Drumlins decrease in size towards the head of the Transition Bay field, where older underlying drumlins, oriented at oblique angles, start to appear (western part of Fig. 8). Around the head of the Transition Bay drumlin field, a number of ribbed moraines are located at the contact

48

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

Fig. 5. Ribbed moraines in Fennoscandia. Sources of information: Norway—Sollid and Torp (1984), Sweden—Ha¨ttestrand (1997), Finland—Nordkalott project (1986) and Mid-Norden project (in press).

between cold- and warm-based conditions. West of the drumlin field head, a glacial landscape formed by a previous ice flow still remains intact. This association of landforms lead Dyke et al. (1992) to suggest that ribbed

moraines form by infolding and stacking of basal debris at the transition from frozen- to thawed-bed conditions. On a local scale, at Hassavare in northern Sweden, Ha¨ttestrand (1997) observed the following succession of

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

49

Fig. 6. The distribution of ribbed moraines compared with the LGM ice sheet margin and frozen bed extent in Fennoscandia. The ribbed moraine distribution, enclosed by the heavy line, is based on Fig. 5. The minimum extent of the LGM frozen bed area (shaded area) is based on the distribution of pre-late Weichselian landforms and sediments (from Kleman et al., in press).

Fig. 7. Transect through the Quebec sector of the Laurentide Ice Sheet, featuring a distal classic warm-based deglaciation landscape (west) and a proximal zone where cold based conditions prevailed until deglaciation (shaded area). The north-trending drumlins in the shaded zone are overprinted by flights of lateral channels draining towards the west (open arrows). This indicates that cold based conditions prevailed during deglaciation, leaving the older north-trending drumlins more or less intact. The ribbed moraines increase in size and density towards the former change in basal thermal regime. (Sources of information: Prest et al., 1969, Kleman et al., 1994).

50

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

Fig. 8. The Transition Bay drumlin field and associated ribbed moraines, Prince of Wales Island, Arctic Canada (redrawn after Dyke et al., 1992). The eastward flowing ice stream forming this drumlin field was surrounded by zones of cold-based conditions, and partly truncated an older, northward trending drumlin field in its proximal part. The location of the ribbed moraines, at the head of the drumlin field, indicates that they formed at the boundary between frozen- and thawed-bed conditions.

Fig. 9. Transition from frozen-bed conditions to ribbed moraine at the mountain Hassavare, northern Sweden (from Ha¨ttestrand, 1997). (Reprinted from Sedimentary Geology, 111, Ha¨ttestrand, C., Ribbed moraines in Sweden—distribution pattern and paleoglaciological implications, 41-56, 1997, with kind permission of Elsevier Science— NL, Sara Burgerhartstraat 25, 1055 KV Amsterdam, The Netherlands.)

morphology from the top of the mountain to the valley basin: Thick continuous soil cover—lee-side till scarp— exposed bedrock—small-scale moraine ridges—fractured till sheet—ribbed moraine (Fig. 9). Kleman and Borg-

stro¨m (1994) argued that lee-side till scarps are diagnostic of the distal ends of frozen-bed patches. Therefore, this succession of landforms is interpreted to be the result of an up-glacier migration of a phase change boundary

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

51

Fig. 10. Direction of ice flow associated with the ribbed moraine formation (short arrows), compared with (a) the LGM ice flow pattern, and (b) the pattern of recession of the Late Weichselian Ice Sheet (from Kleman et al., in press). Sources of information regarding ribbed moraine orientation: see Fig. 5.

(separating frozen and melting conditions at the glacier bed), where material is detached in segments at the downice end of the cold-based zone, and successively forming ribbed moraine ridges. The location of the lee-side till scarp is interpreted to mark the position of the phase change boundary at the time of deglaciation in the area. 3.1.3. Direction of ribbed moraine ridges The direction of the ribbed moraine ridges is almost invariably parallel to the retreating ice margin at deglaciation. As seen in Fig. 10, the direction of ice flow associated with the ribbed moraines in Fennoscandia does not at all match the LGM ice flow pattern (Fig. 10a), but fits well with the receding ice margin of the Late Weichselian Ice Sheet (Fig. 10b). Because ice divides, and hence, ice-flow directions, shifted continuously during the decay phase, the implication is that the ribbed moraines must have formed in a late stage of the deglaciation. Generally this refers to the last (Late Wisconsinan/Weichselian) deglaciation. However, some ribbed moraines in northeastern Sweden (Ha¨ttestrand, 1997) are most likely associated with earlier deglaciations as they are integrated in a deglaciation landscape which dates to the Early Weichselian (Lagerba¨ck and Robertsson, 1988). 3.1.4. Local distribution On a local scale, ribbed moraines are commonly confined to plains, basins, and wide upland plateaux. It is

often stated that ribbed moraines occur in low positions, while drumlins are found on higher ground (e.g. J. Lundqvist, 1969a, 1989; Sugden and John, 1976), and that these observations must be incorporated in theories of formation (e.g. Menzies and Shilts, 1996). However, these topographic relationships are not universal (cf. Aario, 1987). Aylsworth and Shilts (1989a) noted that the ribbed moraines in Keewatin are distributed largely independent of topography. There are also examples of ribbed moraine ridges forming islands in the sea, such as in the Replot area, at the western coast of Finland, and at a few places along the Swedish east-coast (Ha¨ttestrand, 1997). These ribbed moraines must be situated on convex parts of the sea floor, protruding above the present sea surface. 3.2. Relation to other landforms 3.2.1. Drumlin—ribbed moraine relation The original definition of Rogen moraine (J. Lundqvist, 1969a, 1981) included the presence of drumlinoid elements in the ridge field, or fluting of the ridges. This is a rather common feature of ribbed moraines in general (Fig. 11). In most cases the drumlinisation is at a right angle to the ridges and can thus be associated with an ice-flow direction similar to (and probably at a time of formation close to) the ribbed moraine formation. However, this is not always the case. There are several examples of drumlinisation at oblique angles (Wastenson,

52

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

Fig. 11. Aerial photograph of ribbed moraine in Ha¨rjedalen, westcentral Sweden. The upper left part of the ribbed moraine include drumlinoid elements (Blattnick moraine) and superimposed fluting (towards the west). Top north. Published with permission of the National Land Survey of Sweden, 1997-09-19.

1983; Borgstro¨m, 1989) or even parallel to the ridges (Soyez, 1974). In these cases the drumlinisation is clearly a later feature, separated from the ribbed moraine formation. Examples of the opposite age relationship have been reported from Keewatin (Aylsworth and Shilts, 1989a), where drumlins are broken up into ribbed moraines. However, in general, this is a very rare feature. 3.2.2. Ribbed moraines and eskers There are many reports of eskers superimposed on ribbed moraines, either running across ridge crests, or running through meltwater cuts in the ridges. However, there is not one single observation where ribbed moraines are superimposed on eskers. This suggests that the formation of ribbed moraine precedes the presence of abundant meltwater at the base of the ice sheet. 3.3. Morphology 3.3.1. Detailed shape The morphology of individual ridges is generally consistent in well-developed ribbed moraines. They also tend to be of similar size throughout the field. Typical size values are: length, 300—1200 m; width, 150—300 m; and height, 10—30 m. The spacing between the ridges is also very regular. Taking away water bodies, mires, and sediments filling the inter-ridge basins, a longitudinal profile across a ribbed moraine field would show a rather regular asymmetrical wave form. In many cases the crest height is also very consistent. Bouchard (1989) exemplified this ‘accordant summit’ characteristic with a measured profile of the crest height across a ribbed moraine field in Quebec, showing a variability of less than 3 m over a distance of more than 1 km. Bouchard (1989) argued that this was the result of post-formational trunc-

ation, as a shear plane developed in the basal parts of the ice sheet, along a plane following the ridge crests (Fig. 2). Yet, post-formational planing cannot explain all examples of this ‘accordant summit’ characteristic, because it is not confined to ribbed moraines that are smoothed or drumlinised, but is common also in ribbed moraines with rather sharp-crested ridges. This regularity in ridge morphology is generally confined to longitudinal profiles, in the ice-flow direction. Transverse to ice flow (i.e. parallel to the ridges), the morphology is less regular. Ridges are often anastomosing and frequently show crescentic segments, separated by gaps, with spurs pointing down-ice. The anastomosing pattern appears to be a primary feature, and not the result of the drumlinisation process, because this pattern is also featured in ribbed moraines lacking signs of drumlinisation. The cross-section of individual ridges is commonly asymmetric. When compared to the direction of ice flow as recorded by e.g. striae and crag-and-tail drumlins, the ridges are commonly steeper on their distal side. The ridges occasionally have multiple sub-crests, or have flat crests, giving the ridges a tabular appearance in crossprofile. In ribbed moraine fields situated in valleys there is commonly a central gap in the moraine ridges along the valley axis (J. Lundqvist, 1969a; Borgstro¨m, 1989). The ridges bordering this gap tend to have spurs pointing in the direction of ice flow. 3.3.2. Detailed matching of ridges J. Lundqvist (1969a, b), Bouchard (1989), and Ha¨ttestrand (1997) observed that ribbed moraine ridges fit together like a jig-saw puzzle, giving the impression of a broken-up till sheet. The type locality of Rogen moraine, at Lake Rogen in west-central Sweden, is a good example where the detailed morphology of individual ridges fit into each other (Fig. 12). The morphology shown in Fig. 12, indicates a 35—60% extension of a preexisting till sheet, into well-developed ribbed moraine ridges. However, the extension is not uniform over the field. It is most pronounced in the central parts of the ribbed moraine field, while ridges along the lateral margin of the ribbed moraine field are closer to each other. Most ridges have been subjected to a simple parallel extension, but some, especially in the central parts of the field, also have a rotational component of movement (marked with an R in Fig. 12, B3). A slight rotational component is expected as the extension increases from zero at higher terrain, along the lateral edges, up to 60% in the lower central parts. Fig. 13 shows another example of detailed matching of ribbed moraine ridges from central Quebec, Canada. In this area the amount of extension between ridges also varies longitudinally. A succession can be seen, where the proximal ridges are still joined together, while the distal ridges have become completely separated from each other.

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

53

Fig. 12. Jig-saw puzzle-matching of ribbed moraine ridges at Lake Rogen in west-central Sweden. The morphology indicates that the predominant process in the formation of the ribbed moraine is by fracturing of a pre-existing till sheet. The direction of the lines in the shading of the ridges in (A2, A3) and (B2, B3) corresponds to the faint overprinted fluting direction. The areas of overlapping ridges in (A3) and (B3) are confined to ‘horns’ which are interpreted to be formed by post-ridge formational drumlinisation processes. The ridges marked with an R in (B3) are elements which are interpreted to have been slightly rotated during the fracturing process. Aerial photograph mosaic published with permission of the National Land Survey of Sweden, 1997-09-19.

54

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

Fig. 13. Ribbed moraine in northern Quebec. The morphology in this area is interpreted to show the evolution of the fracturing process of ribbed moraine ridges. Ridges A and B are still joined together in their central parts, but are separated towards their ends. Ridges C and D are just becoming two ridges, but are still in contact with each other. Ridges E and F are separated by a narrow basin, while ridges F and G are completely separated from each other. Note the weak fluting superimposed on the ridges. Canadian National Air Photo Library A23592-28, 29.

Ha¨ttestrand (1997) showed an example in the area south of Hassavare (Fig. 9), where the ribbed moraine ridges appear to be flat-crested ‘walls’ between fractures in the till sheet. Laterally, this ribbed moraine grades into an even till sheet with a surface at the same elevation as the ribbed moraine ridge crests. Ha¨ttestrand (1997) interpreted this morphology to represent incomplete fracturing of a pre-existing till sheet. 3.4. Internal properties 3.4.1. Composition The internal composition and structure of ribbed moraines have been discussed in many papers. A striking feature is the vast range in ridge composition. Widely different materials, from pure stratified sand to semiconsolidated lodgement till, have been observed (Cowan,

1968; J. Lundqvist, 1969a, 1989, 1997; Ha¨ttestrand, 1997, Table 1). This feature seems to be related to the fact that ribbed moraines commonly are composed of the same material as the surrounding terrain (Henderson, 1959; Cowan, 1968; J. Lundqvist, 1969a). As mentioned earlier, Ho¨gbom (1885, 1894), in the first papers describing ribbed moraines, noted this diversity in ridge composition. However, few authors describe the stratigraphy throughout the ridges. Frequently, pits 1 or 2 m deep have been used to draw conclusions on the overall composition of the ridges. Still, there are some studies that have described the interior of the ridges in detail. Most commonly, ribbed moraine ridges consist of till of different properties. J. Lundqvist (1969a, 1997) has described sections in ribbed moraine ridges from numerous locations in west-central Sweden. He noted that the most common material in ribbed moraines is ‘ordinary

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

basal till’, but that there are many exceptions and the till composition mostly reflects the regional variations of the till in general, from clayey to coarse textured till. Several sections showed massive basal till, without sorted material or sediment lenses, while other displayed more or less sorted sediment layers interbedded in the till. A specific till type, the Kalix till (Beskow, 1935), characterised by slightly disturbed water-deposited sediments with interbedded stones and boulders, has been found in ribbed moraines in northeastern Sweden (Hoppe, 1948, 1952; Fromm, 1965; J. Lundqvist, 1969a, 1997), and occasionally in other regions (J. Lundqvist 1969a; Bouchard, 1989). There are also many reports of subglacial melt-out till in ribbed moraine ridges. Shaw (1979) describes several large sections in ribbed moraine ridges in west-central Sweden. The internal composition of these ridges showed stratification with diamicton and frequent sorted sediment lenses, and Shaw interpreted this stratigraphy to reflect basal melt-out of till under permafrost conditions during the deglaciation. Similar conclusions were drawn by Bouchard (1980, 1989), who studied many sections of ribbed moraine in Quebec and described distinct layering of diamict and sorted sediments in the ridges. Some ribbed moraines seem to be almost entirely formed from sorted sediments. Ives (1956) described ridges in Quebec, composed of stratified medium to coarse sand. J. Lundqvist (1937) and Aylsworth and Shilts (1989a) reported glaciofluvial gravel and gravely sediments in ribbed moraine ridges in central Sweden and Keewatin, respectively. J. Lundqvist (1997) also described a ribbed moraine in west-central Sweden, consisting almost entirely of well-sorted glaciofluvial sediments. Poorly sorted sediments, interbedded with diamictons have been reported from ribbed moraines in Newfoundland (Fisher and Shaw, 1992). In some cases ribbed moraines contain shattered bedrock. Minell (1977) described sections of ribbed moraine ridges in central Sweden, showing folded beds of alternating till and shattered bedrock. In the same region, J. Lundqvist (1997) found ribbed moraine ridges consisting almost entirely of disintegrated shale. In Manitoba, Dredge et al. (1986) found ribbed moraine ridges consisting primarily of broken rock with little matrix. Aylsworth and Shilts (1989a) also reported ribbed moraines in Keewatin, consisting of disaggregated sandstone without fines. The compactness of the till in ribbed moraines ranges from loose ‘ablation till’ (J. Lundqvist, 1958; Shilts et al., 1987) to very compacted ‘basal till’ (Rasmusson and Tarras-Wahlberg, 1951; Hoppe, 1952; Lindqvist and Svensson, 1957, J. Lundqvist, 1969a). It should also be noted that multiple till beds of different properties are common (J. Lundqvist, 1997). For example, J. Lundqvist (1997) describes a section through a ribbed moraine ridge in western Sweden, that comprised two diamictons,

55

separated by a continuous layer of sorted meltwater sediments. In addition, the material composing the main part of the ridges is often capped by a thin surface layer of diamict sediments, such as flow tills, related to the deglaciation phase (J. Lundqvist, 1969a; Shaw, 1979). There are few investigations of sediment thickness in ribbed moraine areas, but J. Lundqvist (1969a, 1997) noted that the till sheet between individual ridges generally is rather thin. This is supported by seismic investigations by Wastenson (1983) in the Lake Rogen area, indicating that most surficial material is found in the ribbed moraine ridges and that the till sheet is thin or lacking between the ridges. 3.4.2. Structure The internal structure of ribbed moraines is almost as varied as its composition, but one common feature that often appears is the presence of inclined layers and various signs of glaciotectonic structures, such as folds and shear planes. Shaw (1979) described folding and stacking of till beds in ribbed moraine ridges in west-central Sweden. Similarly, in the same region Minell (1977) showed folding and thrusting of bedrock and till beds. In studies of ribbed moraine ridges in Quebec, Bouchard (1980) found till slabs thrusted on top of each other and inclined about 5° in the up-glacier direction. Pronounced signs of glaciotectonic processes were found by Fisher and Shaw (1992) in some ribbed moraines on Newfoundland. Sections revealed inclined layers (up to 40°) conforming with surface slopes, as well as truncated, folded, and brecciated beds, indicating a high degree of postdepositional reworking. However, most sections through ribbed moraine ridges in their study area showed no such signs of glaciotectonism. In addition to the studies mentioned here, several other studies have described inclined layering and glaciotectonic structures in ribbed moraine ridges (e.g. Hoppe, 1948, 1952; Cowan, 1968; Minell, 1979; Dredge et al., 1986; J. Lundqvist, 1997). Yet, it should be noted that there are also reports of ribbed moraine ridges composed of massive till, lacking laminae, thrusts, faults, or other signs of glaciotectonism (e.g. J. Lundqvist, 1997). Studies of the preferred fabric of pebbles and stones in the till show no systematic trend. For example, Shaw (1979) found no preferred orientation at all, but a strong up-glacier long axis dip, Bouchard (1989) and Fisher and Shaw (1992) found a strong clast orientation in the iceflow direction, while Hoppe (1952), showed consistent clast orientations parallel to the ribbed moraine ridges. G. Lundqvist (1948) and J. Lundqvist (1958, 1969a, 1997) describe several examples of preferred clast orientations both parallel to as well as transverse to the ice-flow direction. Still other ridges had a preferred clast orientation neither parallel nor transverse to the last recorded ice flow. There are also reports of clast orientations

56

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

varying within one single ribbed moraine ridge (e.g. J. Lundqvist, 1997), occasionally following the curvature of the ridge (Cowan, 1968; Aario, 1987). Wastenson (1983) showed that the preferred orientation of stones in the upper metre of till was aligned with the superimposed drumlinisation, i.e. transverse to the ribbed moraine ridges. At greater depth, a weaker stone orientation parallel with the ridge, was found. Wastenson‘s (1983) study demonstrates the need for deep sections when making stratigraphic studies in ribbed moraine ridges (cf. J. Lundqvist, 1969a). The study also indicates that the drumlinisation is a secondary process, not linked to the primary ridge construction.

4. Discussion Any theory of ribbed moraine formation must be compatible with, and be able to explain, the observations described above. In summary, ribbed moraines 1. occur only in core areas of former glaciation, 2. commonly occur close to and distal of frozen-bed areas, 3. are formed during deglaciation of ice sheets, 4. are most common on concave or flat surfaces in the terrain, 5. are commonly drumlinised, 6. are rarely found superimposed on aligned drumlins, 7. have eskers superimposed on them, not vice versa, 8. are commonly regular in ridge height and spacing, 9. may have flat, or sometimes multiple, crests of individual ridges, 10. are commonly asymmetric, with a steeper lee face, 11. often display a ‘jig-saw puzzle matching’ of the ridges, 12. are composed of a variety of materials, 13. commonly consist of material also found in its surroundings and 14. commonly display glaciotectonic structures. In addition to these observations, a theory of ribbed moraine formation must also be theoretically sound with respect to existing glaciological knowledge. Therefore, we disregard Aario’s (1977a, b) model of wavy and spiral ice flow at the base of the ice sheet, because such behaviour of basal glacier ice is considered to be highly unlikely (Thompson, 1979). Also, we reject the subglacial meltwater flood hypothesis of Fisher and Shaw (1992). The storage capacity of the ice sheet for subglacial water reservoirs proximal of the innermost ribbed moraines is minimal, as the ribbed moraines, at least in Fennoscandia, exist right up to the position of the last ice divide (e.g. Ha¨ttestrand, 1997). Neither are en- or supraglacial water reservoirs possible, as the central parts of the ice sheet had basal temperatures below the pressure melting point, and thus, must have been even colder higher up in the ice

mass. In addition, Fisher and Shaw’s formation theory (their Fig. 23) results in a cross-section of the ribbed moraine ridges where the steep side are facing the upglacier direction. This is in direct conflict with almost all other observations of ribbed moraines, where the distal side of the ridges is steeper than the proximal side (relative to the ice flow as recorded by e.g. striae). Some authors have also suggested that ribbed moraines are a polygenetic group of landforms (e.g., Kurimo, 1980; Fisher and Shaw, 1992; J. Lundqvist, 1997), but we consider it unlikely that these landforms, with such a well-defined distribution pattern (matching between different morphological varieties of ribbed moraine), and so many unique morphological characteristics could be formed by several, independent, processes. Because the overprint of drumlinisation and eskers clearly defies a frontal or dead-ice (slumping of supraglacial material into crevasses) explanation for the ribbed moraine formation, we will concentrate the discussion on the following two theories: (1) ¹he shear and stack theory: Glaciotectonic shearing and stacking of subglacial sediments due to compressive ice flow, followed by basal melt-out of englacial debris, entrained along subglacial shear planes. (2) ¹he fracturing theory: Fracturing of a pre-existing till sheet, as the result of extensional basal ice flow during a transition from frozen (non-sliding) to melting (sliding) basal conditions. It should be noted that although we treat the shear and stack theory as one coherent theory, we acknowledge that there are differences in the precise mechanisms and boundary conditions proposed (see earlier section on formation hypotheses). However, the key elements are generally the same, i.e. shearing and stacking of sediments during compressive flow. The shear and stack theory was developed mainly as the result of observations on glaciotectonic structures, the presence of basal melt-out till, and the detailed morphology of individual ridges (e.g., Shaw, 1979; Bouchard, 1989). Other morphological characteristics were used to a lesser extent, while the regional and global distribution of ribbed moraines and their relation to other landforms were rarely used at all. Thus, the shear and stack theory is primarily concerned with the subglacial conditions that prevailed during the time of sediment deposition, which, in this theory, is linked to the ridge formation. The fracturing theory is based primarily on morphological and distribution characteristics, and most of the stratigraphy is regarded to be inherited from a pre-existing drift sheet. This theory requires not only that specific subglacial conditions are met, but also a specific sequence of events leading to the ribbed moraine formation (Ha¨ttestrand, 1997). The characteristic distribution of the ribbed moraines, restricted to the core areas of glaciation (observation 1),

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

and its spatial agreement with the reconstructed extent of frozen bed areas under the Fennoscandian Ice Sheet (Fig. 6; Kleman et al., in press) and parts of the Laurentide Ice Sheet (Kleman et al., 1994), conforms well with the idea of ribbed moraine formation by fracturing. Ha¨ttestrand (1997) links the ribbed moraine distribution to the extent and gradual decrease of frozen bed areas underneath ice sheets. Zones that were fully warm-based, or fully cold-based, throughout the glaciation, did not supply the necessary conditions of (i) extensional basal ice flow, and (ii) a shear plane at the base of the drift sheet, separating non-deformable bedrock beneath from the brittle-deformable drift, frozen to the overlying glacier sole, (Fig. 3). Therefore, according to the fracturing theory, the characteristics in the spatial distribution of ribbed moraine (Figs 4 and 5) reflect the extent of frozen bed areas underneath ice sheets, with the exception of areas that never reached the pressure melting point before deglaciation (i.e. ribbed moraines could not form)., However, there may be areas where subsequent prolonged wet-based ice flow erased previously formed ribbed moraine ridges. The patchy distribution of ribbed moraines, and their increase in abundance towards frozen-bed zones (e.g. Fig. 7), may also be linked to later erosion. During deglaciation, the ice margin must have retreated faster than the migrating frozen-bed/thawed-bed boundary. Otherwise the deglaciation would not have been cold-based in the central parts of the ice sheets. Thus, the most distal ribbed moraines experienced longer periods of wet-based erosive ice flow than the inner ones, causing a secondary change in the distribution pattern. In addition, the patchy distribution and abundance variations of ribbed moraines may also reflect the patchiness of primary subglacial conditions. The patchy nature of the transition zone between cold- and warm-based areas, where frozen-bed patches become gradually larger and more continuous towards the ice sheet centre, has been demonstrated both in theory (e.g. Hughes, 1981) and in the glacial geological record (e.g. Kleman and Borgstro¨m, 1994). Ribbed moraine formation by fracturing may also explain why no ribbed moraines are found beneath the Antarctic and Greenland ice sheets or in areas recently deglaciated by smaller valley glaciers and ice caps. These environments lack the substantial spatial reorganisation in ice-flow directions and basal thermal regime that characterised the deglaciation phase of the Laurentide and Fennoscandian ice sheets. Thus, the specific subglacial conditions required in the fracturing theory, i.e. a change from a frozen to a thawed bed over large areas, are not met in these modern glacial environments. In our view, the distribution of ribbed moraines is not satisfactory accounted for in the shear and stack theory. Aylsworth and Shilts (1989a, b) explained the distribution as the result of the availability and, foremost, the

57

physical properties of the sediment load. Outside the area of ribbed moraine occurrence there is commonly areas dominated by exposed bedrock, and landforms in loose deposits are therefore not developed. This is true for most of their study area, Keewatin, but not in other areas of glaciation. The southern limit of ribbed moraine distribution in Labrador and Scandinavia does not correspond with a decrease in drift sheet thickness (Fulton, 1995; Ha¨ttestrand 1997). In addition, along the outer limits of glaciation there are generally thick glacial deposits and other morainic landforms such as drumlins and end moraines. However, ribbed moraines are completely lacking. It does not appear possible to correlate the ribbed moraine distribution to spatial variations in the physical properties of the drift sheet, as there is such a high variability in the interior composition of ribbed moraines (J. Lundqvist, 1969a, 1989, 1997). Sollid and S+rbel (1984, 1990, 1994) explained the ribbed moraine distribution as a consequence of an outward migration of a frozen core area of the ice sheet and they argued that ribbed moraines formed during this transition. This explains the proximity of ribbed moraines to cold-based parts of the ice sheet, but is in direct conflict with observation 3 above, i.e. that ribbed moraines formed close to the retreating ice margin just prior to deglaciation. The common position of ribbed moraines distal of cold-based patches (observation 2) is not a favourable position for compressive ice flow and shearing. A downice transition from frozen to melting subglacial conditions is rather expected to result in extensional flow, as the ice flow velocity increases due to basal sliding. Furthermore, ice that flows from a continuously frozen bed core area of an ice sheet, which has been cold-based since glaciation inception, is devoid of debris and cannot deposit any till, despite its warm based nature. Consequently, we regard it highly unlikely that an ice sheet can deposit large transverse ridges just distal to frozen bed areas. According to the fracturing theory, ribbed moraines should be formed only in the core areas of the ice sheets, because this is where cold based zones are expected (Schytt, 1974; Hooke, 1977; Huybrechts and T’siobbel, 1995; Heine and McTigue, 1996) and interpreted to have existed (Sollid and S+rbel, 1984; Lagerba¨ck, 1988a, b; Dyke, 1993; Ha¨ttestrand, 1997; Kleman et al., in press). Some formational hypotheses (e.g. Punkari, 1984; Bouchard and Salonen, 1989) require, or are based on, an inferred thermal distribution at the ice sheet base that includes a melting inner area grading into a frozen margin through an intermediate freezing zone. The glaciotectonism leading to the formation of the ribbed moraine is suggested to occur at the transition to frozen conditions as thrusting and shearing of unfrozen sediments. However, a frozen margin outside a thawed inner zone is only likely to result if permafrost conditions

58

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

prevail. Yet, almost all ribbed moraines are situated inside the Younger Dryas ice margin and we find it unlikely that the ice had a frozen margin during the rapid deglaciation in Pre-Boreal/Boreal time. In Sweden there are no indications at all of deglacial permafrost conditions inside the Younger Dryas moraines (J. Lundqvist, 1962, 1981). Bouchard (1980, 1989), Minell (1980), and Sollid and S+rbel (1984) suggest that topographic obstructions cause the compressive flow leading to the shearing and stacking of ribbed moraine ridges. These studies were conducted in a mountainous or undulating hilly terrain, and as described above in the section on local distribution, there is in some areas a tendency for ribbed moraines to be located mostly in topographic lows. However, in other areas, such as parts of Keewatin, the underlying terrain may be extremely flat and still holds abundant ribbed moraine (Aylsworth and Shilts, 1989b). Thus, topographic expressions do not seem to be a primary controlling factor in ribbed moraine formation, even though drift sheet thickness variations may induce more favourable conditions for ribbed moraine formation in lower parts of the terrain. The observations that drumlinisation of ribbed moraines is by far more common than ribbed moraines superimposed on drumlins (observation 5 and 6) do not agree well with ribbed moraine formation by shearing and stacking. If warm-based conditions precede ribbed moraine construction, many more examples of e.g. sheared and stacked drumlins would be expected, especially if the ribbed moraine construction was the last geomorphic action before deglaciation or before the ice became cold based (Sollid and S+rbel, 1984, 1990, 1994). The same argument holds true for eskers, as there is no known examples of ribbed moraines superimposed on eskers (observation 7). According to Ha¨ttestrand (1997), ribbed moraine construction is the first geomorphic action in a sequence of events, from preservation beneath cold-based ice to melting (and geomorphic active) conditions and finally deglaciation. This agrees well with observations 5, 6—7. The drumlinoid features of ribbed moraines have lead many authors to propose a genetic continuum, where ribbed moraine form one end member and drumlins the other (J. Lundqvist, 1969a, 1981, 1989; Aario, 1977a, b, 1987; Carl, 1978; Dredge et al., 1986; Menzies and Shilts, 1996). For example, Dredge et al. (1986) and Aario (1987) argue that ribbed moraines and drumlins form contemporaneously and Dredge et al. (1986) suggest that variations in the basal debris-load control whether one or the other landform is created. J. Lundqvist (1969a) also reported a spatial continuum where ribbed moraines found in basins are replaced by drumlinised terrain on higher, surrounding ground. However, transitional forms, in a morphological sense, need not represent transitional conditions during their formation. An alter-

native explanation is that they simply represent ribbed moraine ridges that were later drumlinised to various degrees (cf. Bouchard, 1989), similar to other geomorphological features being overprinted by drumlinisation/ fluting. Hence, the subglacial conditions during which the drumlinisation was formed may have been totally unrelated to the conditions associated with ribbed moraine formation. Therefore, we argue that the continuum concept only has merit for morphological descriptions, at most, and may lead to false conclusions if integrated in a genetic discussion. In the fracturing theory, all aspects of the apparent morphological continuum between ribbed moraines and drumlins can be explained by post-formative streamlining and drumlinisation of originally intact ribbed moraine. The detailed morphology of the ridges can to some degree be explained by shearing and stacking of subglacial sediments. For example, the multiple crests and the asymmetry of ribbed moraine ridges (observation 9) are in agreement with what is expected from ridges formed by shear planes. The longitudinal regularity (observation 8) and the transverse irregularity in ribbed moraine fields are less agreeable with the shear and stack theory. If the ribbed moraine ridges were formed by shearing processes, the regularity would be expected in the attained length of individual ridges, such as in shear or thrust moraines in front of present-day glaciers, and not in the height and distance between ridges. Shear moraines in front of present day glaciers are typically regular and continuous in their length, but have irregular spacing (Sugden and John, 1976). They are also generally arcuate with their concave side facing up-glacier. However, ribbed moraine ridges are quite the opposite. Ridge construction by extensional ice flow and fracturing can be compared with friction cracks on glacially abraded outcrops (Harris, 1943), created by tensional forces at the base of the ice sheet. Friction cracks are regular in spacing and are convex up-glacier, similar to crescentic ribbed moraine ridge segments. Of critical importance are the observations that the shape of individual ribbed moraine ridges match each other, like a jig-saw puzzle (observation 11). We argue that the matching of ridges shown from Sweden and Canada in Figs 12 and 13 cannot be explained by shearing and stacking. Especially, the fit of small ribbed moraine elements, without a distinct ridge shape, into larger ridges (e.g. Fig. 12: A3) is hard to explain by these processes. The shear planes would have to be very restricted in their width and length, but still carry enough sediments to form ridge elements up to 30 m high. In our opinion, the matching of ridges rather strongly indicates that fracturing is an important process in the formation of the ridges. Also other morphological characteristics may be explained by fracturing of a pre-existing drift sheet.

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

Ha¨ttestrand (1997) subdivided the population of ribbed moraines into four different classes, based on their morphology. These different ribbed moraines can to some extent be explained by the initial properties of the preexisting drift sheet, from which the ribbed moraine is formed, and by post-formational processes: (i) If the initial drift sheet has an even surface the result will be a ribbed moraine with ridges of similar shape and size, such as the classic Rogen moraine. We also argue that the ‘accordant summit’ characteristic, suggested by Bouchard (1989) to be the result of post-formational truncation, is better explained by a flat, pre-existing plain, broken up into ridges. (ii) If the initial drift sheet is hummocky, the result will be irregular ridges of varying height and length (cf. the ‘hummocky ribbed moraine’ of Ha¨ttestrand (1997)). (iii) If the drift sheet is thin, the resulting ridges will be low (but may still be as long) and closely spaced (Borgstro¨m, 1989; Bouchard, 1989). This type of morphology has been classified as ‘minor ribbed moraine’ by Ha¨ttestrand (1997). (iv) The Blattnick moraine (Markgren and Lassila, 1980), is characterised by low and broad ridges, like incomplete drumlins, and has been considered an intermediate form in a genetic continuum from Rogen moraine to drumlins. We argue that the Blattnick moraine is best characterised as strongly drumlinised ribbed moraines of the types (i) or (ii) above. As mentioned, the shear and stack theory is primarily based on stratigraphic studies. For example, Shaw (1979) and Bouchard (1989) based their theories on studies of subglacial melt-out till and inclined shear planes within the ridges (observation 14). They invoked topographic and ice thickness control to account for the compressive flow. Aylsworth and Shilts (1989a, b) found an intimate relationship between the occurrence of ribbed moraines and the presence of coarse-grained till and concluded that the physical nature of the sediments was a major cause for the compressive flow. However, shearing and stacking of sediments, followed by basal melt-out, cannot explain all of the great diversity in ridge stratigraphy. Evidence of subglacial melt-out of near-basal debris exist in some ridges, but the presence of lodgement till and glaciofluvial sands is more difficult to account for by shearing, stacking, and basal melt-out. Rather, the fact that ribbed moraines can be composed of almost any material (observation 12) and commonly of the same material as in the surrounding terrain (observation 13), indicates that the primary material deposition is separated, in time and process, from the formation of the ridges, i.e. ribbed moraines form from pre-existing drift sheets (cf. J. Lundqvist, 1997). According to Ha¨ttestrand (1997), the glaciotectonic structures are formed by deformation of the ridges while they are in transport at the base of the ice sheet, after initial fracturing, and during final deposition of the ridges. These post-formational processes, combined with subsequent drumlinisation, is also likely the explanation for the commonly found

59

asymmetry of the ridges (observation 10). Ribbed moraine formation by shearing and stacking does not conform with the observations that almost all drift in some ribbed moraines is confined to the ridges (Wastenson, 1983). Shearing and stacking of subglacial debris would cause an excess of material in the ribbed moraine field, i.e. a thickening of the till sheet (Fig. 2). The lack of till between the ridges agrees better with the idea of fracturing and extension of a pre-existing drift sheet. Detachment and fracturing of the drift sheet are likely to occur as soon as the phase change surface (separating underlying thawed material from overlying frozen material) rises to the bedrock/drift interface (Ha¨ttestrand, 1997). To sum up: In our opinion, the shear and stack theory is able to satisfactory explain observations 4, 9, 10, and 14. It does not contradict nor explain observations 1, 3, 5, 13, but is in direct conflict with observations 2, 6, 7, 8, 11, and 12. In contrast, the fracturing theory is compatible with, and satisfactorily explains, most of the observations stated above. We therefore argue that the formation of ribbed moraines is best explained by fracturing of a preexisting drift sheet at the transition from frozen to thawed basal conditions. Ribbed moraine formation according to the fracturing theory has several implications for the interpretation of the evolution of past ice sheets. The regional distribution of ribbed moraines generally outlines the distribution of areas that were cold based during the last glaciation, and that thawed during the deglaciation. However, in areas that were cold-based throughout the glaciation, the necessary conditions for ribbed moraine formation were never met. Therefore, other information on the extent of continuously frozen-bed conditions, such as preserved older morphological strata (e.g. Lagerba¨ck 1988a, b), have to be incorporated when reconstructing the full extent of frozen bed areas under past ice sheets.

5. Conclusions The distribution pattern, morphology, relation to other landforms, and stratigraphy of ribbed moraines indicate formation by fracturing of a pre-existing till sheet, in accordance with the fracturing theory. Key elements in this formation theory are extensional ice flow and a sandwich layering of competent materials (frozen sediments and bedrock) and incompetent materials (thawed sediments and glacier ice) with brittle and ductile deformation behaviour, respectively. Both of these key elements result from the inward retreat of a frozen-bed/ thawed-bed boundary of a deglaciating ice sheet. Thus, ribbed moraines are potential tools in reconstructing the extent of frozen-bed zones of past ice sheets.

60

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61

Acknowledgements We would like to thank Peter Jansson, Jan Lundqvist, Arjen Stroeven, and two anonymous referees for constructive comments on the manuscript. This project was funded through a grant to Johan Kleman from the Swedish Natural Science Research Council. References Aario, R. (1977a). Associations of flutings, drumlins, hummocks and transverse ridges. GeoJournal, 16, 65—72. Aario, R. (1977b). Classification and terminology of morainic landforms in Finland. Boreas, 6, 87—100. Aario, R. (1987). Drumlins of Kuusamo and Rogen-ridges of Ranua, northeast Finland. In J. Menzies, and J. Rose, (Eds.), Drumlin symposium, 87—101. Rotterdam: Balkema. Attig, J. W. (1985). Pleistocene geology of Vilas County, Wisconsin. ¼isconsin Geology and Natural History Survey, Informational Circular, Vol. 50, 32pp. Aylsworth, J. M., & Shilts, W. W. (1989a). Bedforms of the Keewatin Ice Sheet, Canada. Sedimentary Geology, 62, 407—428. Aylsworth, J. M., & Shilts, W. W. (1989b). Glacial features around the Keewatin ice divide: Districts of Mackenzie and Keewatin. Geological Survey of Canada Paper, 88—24, 21pp. Beskow, G. (1935). Praktiska och kvarta¨rgeologiska resultat av grusinventeringen i Norrbottens la¨n. Geologiska Fo¨ reningens i Stockholm Fo¨ rhandlingar, 57, 120—123. Borgstro¨m, I. (1989). Terra¨ngformerna och den glaciala utvecklingen i so¨dra fja¨llen. Dissertation, Department of Physical Geography, University of Stockholm. Bouchard, M. A. (1980). Late Quaternary geology of the Te´miscamie area, central Que´bec, Canada. Dissertation, Department of Geological Sciences, McGill University. Bouchard, M. A. (1986). Ge´ologie des depoˆts meubles de la re´gion de Te´miscamie, Territoire du Nouveau-Que´bec. Ministe´re de l‘‘Energie et des Ressources du Que´bec, Me´moire, MM 83—03, 90pp. Bouchard, M. A. (1989). Subglacial landforms and deposits in central and northern Quebec, Canada, with emphasis on Rogen moraines. Sedimentary Geology, 62, 293—308. Bouchard, M. A., Ignatius, H. G., & Ko¨nigsson, L.-K. (1989). Ribbed moraines of North America—A historical account. Geologi, 41, 75—79. Bouchard, M. A., & Salonen, V. P. (1989). Boulder transport in shield areas. In: R. Kujansuu, & M. Saarnisto (Eds.), Handbook of Glacial Indicator ¹racing, 87—107. Rotterdam: Balkema. Boulton, G. S. (1970a). On the deposition of subglacial and melt-out tills at the margins of certain Svalbard glaciers. Journal of Glaciology, 9, 231—245. Boulton, G. S. (1970b). On the origin and transport of englacial debris in Svalbard glaciers. Journal of Glaciology, 9, 213—229. Boulton, G. S. (1971). Till genesis and fabric in Svalbard, Spitsbergen. In: R. P. Goldtwaith (Ed.), ¹ill: A Symposium, 41—72. Columbus: Ohio State University Press. Boulton, G. S. (1972). Modern arctic glaciers as depositional models for former ice sheets. Quarterly Journal of the Geological Society of ¸ondon, 128, 361—393. Boulton, G. S. (1987). A theory of drumlin formation by subglacial sediment deformation. In: J. Menzies, & J. Rose (Eds.), Drumlin Symposium, 25—80. Rotterdam: Balkema. Carl, J. D. (1978). Ribbed moraine—drumlin transition belt, St. Lawrence Valley, New York. Geology, 6, 562—566. Cowan, W. R. (1968). Ribbed moraine: till-fabric analysis and origin. Canadian Journal of Earth Sciences, 5, 1145—1159.

Craig, B. G. (1965). Glacial Lake McConnel, and the surficial geology of parts of Slave River and Redstone River map-areas, District of Mackenzie. Geological Survey of Canada Bulletin, 122, 33pp. Dredge, L. A., Nixon, F. M., & Richardson, R. J. (1986). Quaternary geology and geomorphology of northwestern Manitoba. Geological Survey of Canada Memoir, 418, 38pp. Dyke, A. S. (1993). Landscapes of cold-centred late Wisconsinan ice caps, Arctic Canada. Progress in Physical Geography, 17, 223—247. Dyke, A. S., Morris, T. F., Green, D. E. C., & England, J. (1992). Quaternary geology of Prince of Wales Island, Arctic Canada. Geological Survey of Canada Memoir, 433, 142pp. Fisher, T. G., & Shaw, J. (1992). A depositional model for Rogen moraine, with examples from the Avalon Peninsula, Newfoundland. Canadian Journal of Earth Sciences, 29, 669—686. Fro¨din, G. (1913). Bidrag till va¨stra Ja¨mtlands senglaciala geologi. Sveriges Geologiska ºnderso¨ kning C, 246, 236pp. Fro¨din, G. (1925). Studien u¨ber die Eissheide in Zentralskandinavien. Bulletin of the Geological Institutions at the ºniversity of ºpsala, 19, 214 pp. Fro¨din, G. (1954). De sista skedena av Centralja¨mtlands glaciala historia. Geographica, 24, 1—251. Fromm, E. (1965). Beskrivning till jordartskartan o¨ver Norrbottens la¨n nedanfo¨r Lappmarksgra¨nsen. Sveriges Geologiska ºnderso¨ kning Ca, 39, 236pp. Fulton, R. J. (Ed.) (1995). Surficial materials of Canada, 1 : 5,000,000. Geological Survey of Canada, Map, 1880A. Granlund, E. (1943). Beskrivning till jordartskarta o¨ver Va¨sterbottens la¨n nedanfo¨r odlingsgra¨nsen. Sveriges Geologiska ºnderso¨ kning Ca, 26, 165pp. Harris, S. E. (1943). Friction cracks and the direction of glacial movement. Journal of Geology, 51, 244—258. Ha¨ttestrand, C. (1997). Ribbed moraines in Sweden—distribution pattern and paleoglaciological implications. Sedimentary Geology, 111, 41—56. Heine, J. T., & McTigue, D. F. (1996). A case for cold-based continental ice sheets—a transient thermal model. Journal of Glaciology, 42, 37—42. Henderson, E. P. (1959). A glacial study of Central Quebec-Labrador. Geological Survey of Canada Bulletin, 50, 94pp. Ho¨gbom, A. G. (1885). Glaciala och petrografiska iakttagelser i Jemtlands la¨n. Sveriges Geologiska ºnderso¨ kning C, 70, 39pp. Ho¨gbom, A. G. (1894). Geologisk beskrifning o¨fver Jemtlands la¨n. Sveriges Geologiska ºnderso¨ kning C, 140, 107pp. Ho¨gbom, A.G. (1920). Geologisk beskrivning o¨ver Ja¨mtlands la¨n 2nd edn. Sveriges Geologiska ºnderso¨ kning C, 140, 139pp. Holmsen, G. (1935). Nordre Femund. Beskrivelse til det geologiske rektangelkart. Norsk Geologisk ºnderso¨ kelse, 144, 55pp. Hooke, R., & Le, B. (1977). Basal temperatures in polar ice sheets: a qualitative review. Quaternary Research, 7, 1—13. Hoppe, G. (1948). Isrecessionen fra n Norrbottens kustland. Geographica, 20, 1—112. Hoppe, G. (1952). Hummocky moraine regions, with special reference to the interior of Norrbotten. Geografiska Annaler, 34A, 1—72. Hoppe, G. (1959). Glacial morphology and inland ice recession in northern Sweden. Geografiska Annaler, 41A, 193—212. Hoppe, G. (1968). Ta¨rnasjo¨-omra dets geomorfologi—En o¨versiktlig orientering med sa¨rskild ha¨nsyn till de glaciala och postglaciala formelementen. University of Stockholm, Department of Physical Geography, Research Report, 8, 17pp. Hughes, O. L. (1964). Surficial geology, Nichicun-Kaniapiskau maparea, Quebec. Geological Survey of Canada Bulletin, 106, 18pp. Hughes, T. J. (1981). Numerical reconstruction of paleo-ice sheets. In: G. H. Denton, & T. J. Hughes (Eds.), ¹he ¸ast Great Ice Sheets, 211—261. New York: Wiley-Interscience. Huybrechts, P., & T’siobbel, S. (1995). Thermomechanical modelling of northern hemisphere ice sheets with a two-level mass-balance parametrization. Annals of Glaciology, 21, 111—116.

C. Ha( ttestrand, J. Kleman / Quaternary Science Reviews 18 (1999) 43—61 Ives, J. D. (1956). Till patterns in Central Labrador. Canadian Geographer, 8, 25—33. Kleman, J., & Borgstro¨m, I. (1994). Glacial land forms indicative of a partly frozen bed. Journal of Glaciology, 40, 255—264. Kleman, J., Borgstro¨m, I., & Ha¨ttestrand, C. (1994). Evidence of a relict glacial landscape in Labrador-Ungava. Palaeogeography, Palaeoclimatology, Palaeoecology, 111, 217—228. Kleman, J., Ha¨ttestrand, C., Borgstro¨m, I., & Stroeven, A. (in press). Fennoscandian palaeoglaciology reconstructed using a glacial geological inversion model. Journal of Glaciology. Kujansuu, R. (1967). On the deglaciation of western Finnish Lapland. Bulletin de la Commission Ge´ ologique de Finlande, 232, 98pp. Kurimo, H. (1977). Pattern of dead-ice deglaciation forms in western Kemija¨rvi, Northern Finland. Fennia, 153, 43—56. Kurimo, H. (1980). Depositional deglaciation forms as indicators of different glacial and glaciomarginal environments. Boreas, 9, 179—191. Lagerba¨ck, R. (1988a). Periglacial phenomena in the wooded areas of north Sweden—relicts from the Ta¨rendo¨ Interstadial. Boreas, 17, 487—499. Lagerba¨ck, R. (1988b). The Veiki moraines in northern Sweden—widespread evidence of an Early Weichselian deglaciation. Boreas, 17, 469—486. Lagerba¨ck, R., & Robertsson, A. -M. (1988). Kettle holes—stratigraphic archives for Weichselian geology and Palaeo-environment in northernmost Sweden. Boreas, 17, 439—468. Lee, H. A. (1959). Surficial geology of the southern district of Keewatin and the Keewatin Ice Divide, Northwest Territories, Canada. Geological Survey of Canada Bulletin, 51, 42pp. Lindqvist, A> ., & Svensson, S. (1957). Glacialmorfologiska studier i Gysenomras det i nordva¨stra Ja¨mtland. Geographica, 31, 206—222. Lundqvist, G. (1935). Isavsma¨ltningen i Bergslagen. Geologiska Fo¨ reningens i Stockholm Fo¨ rhandlingar, 57, 287—301. Lundqvist, G. (1937). Sjo¨sediment fras n Rogenomras det i Ha¨rjedalen. Sveriges Geologiska ºnderso¨ kning C, 408, 80pp. Lundqvist, G. (1943). Norrlands jordarter. Sveriges Geologiska ºnderso¨ kning C, 457, 166pp. Lundqvist, G. (1948). Blockens orientering i olika jordarter. Sveriges Geologiska ºnderso¨ kning C, 497, 29pp. Lundqvist, G. (1951). Beskrivning till jordartskarta o¨ver Kopparbergs la¨n. Sveriges Geologiska ºnderso¨ kning Ca, 21, 213pp. Lundqvist, J. (1958). Beskrivning till jordartskarta o¨ver Va¨rmlands la¨n. Sveriges Geologiska ºnderso¨ kning Ca, 38, 229pp. Lundqvist, J. (1962). Patterned ground and related frost phenomena in Sweden. Sveriges Geologiska ºnderso¨ kning C, 583, 101pp. Lundqvist, J. (1969a). Problems of the so-called Rogen moraine. Sveriges Geologiska ºnderso¨ kning C, 648, 32pp. Lundqvist, J. (1969b). Beskrivning till jordartskarta o¨ver Ja¨mtlands la¨n. Sveriges Geologiska ºnderso¨ kning Ca, 45, 418pp. Lundqvist, J. (1981). Moraine morphology—Terminological remarks and regional aspects. Geografiska Annaler, 63A, 127—138. Lundqvist, J. (1989). Rogen (ribbed) moraine—identification and possible origin. Sedimentary Geology, 62, 281—292. Lundqvist, J. (1997). Rogen morain—an example of two-step formation of glacial landscapes. Sedimentary Geology, 111, 27—40. Mannerfelt, C. (1942). Nas gra norrla¨ndska kartanalyser. Globen, 3, 17—20. Mannerfelt, C. (1945). Nas gra glacialgeologiska formelement och deras vittnesbo¨rd om inlandsisens avsma¨ltningsmekanik i svensk och norsk fja¨llterra¨ng. Geografiska Annaler, 27A, 1—239. Markgren, M., & Lassila, M. (1980). Problems of moraine morphology: Rogen moraine and Blattnick moraine. Boreas, 9, 271—274. Menzies, J., & Shilts, W. W. (1996). Subglacial environments. In: J. Menzies (Ed.), Past Glacial Environments: Sediments, Forms and ¹echniques, 15—136. Oxford: Butterworth-Heineman.

61

Midnorden Project (in press). Map of glacial geomorphology and paleohydrology, 1:1 mill. Geological Surveys of Finland, Norway and Sweden. Minell, H. (1977). Transverse moraine ridges of basal origin in Ha¨rjedalen. Geologiska Fo¨ reningens i Stockholm Fo¨ rhandlingar. 99, 271—277. Minell, H. (1979). The genesis of tills in different moraine types and the deglaciation in a part of central Lappland. Sveriges Geologiska ºnderso¨ kning C, 754, 83pp. Minell, H. (1980). The distribution of local bedrock material in some moraine forms from the inner part of northern Sweden. Boreas, 9, 275—281. Nordkalott project (1986). Map of Quaternary Geology, sheet 2: Glacial geomorphology and paleohydrology, Northern Fennoscandia, 1:1 mill. Geological Surveys of Finland, Norway and Sweden. Prest, V. K. (1968). Nomenclature of moraines and ice-flow features as applied to the glacial map of Canada. Geological Survey of Canada Paper, 67—57. 32pp. Prest, V. K., Grant, D. R., & Rampton, V. N. (1968). Glacial map of Canada. Geological Survey of Canada, Map 1253A. Punkari, M. (1982). Glacial geomorphology and dynamics in the eastern part of the Baltic shield interpreted using Landsat imagery. ¹he Photogrametric Journal of Finland, 9, 77—93. Punkari, M. (1984). The relations between glacial dynamics and the tills in the eastern part of the Baltic shield. In: L. -K. Ko¨nigsson (Ed.), ¹en ½ears of Nordic ¹ill Research, Striae, 20, 49—54. Rasmusson, G., Tarras-Wahlberg, N. (1951). Terra¨ngformerna vid Saxvattnet. Svensk Geografisk A> rsbok, 27, 173—176. Schytt, V. (1974). Inland ice sheets—recent and Pleistocene. Geologiska Fo¨ reningens i Stockholm Fo¨ rhandlingar, 96, 299—309. Shaw, J. (1979). Genesis of the Sveg till and Rogen moraines of central Sweden: a model of basal melt out. Boreas, 8, 409—426. Shilts, W. W. (1977). Geochemistry of till in perennially frozen terrain of the Canadian shield—application to prospecting. Boreas, 6, 203—212. Shilts, W. W., Aylsworth, J. M., Kaszycki, C. A., & Klassen, R. A. (1987). Canadian shield. Geological Society of America Centennial Special »olume, 2, 119—161. Sollid, J. L., & S+rbel, L. (1984). Distribution and genesis of moraines in Central Norway. In: Ko¨nigsson, L. -K. (Ed.), ¹en ½ears of Nordic ¹ill Research, Striae, 20, 63—67. Sollid, J. L., & S+rbel, L. (1990). Studier av moreneformer i sentrale "st-Norge, særlig Rogenmora¨ner. Abstract. Norsk Geologiske Forening, Geonytt, 17, p. 105. Sollid, J. L., & S+rbel, L. (1994). Distribution of glacial landforms in southern Norway in relation to the thermal regime of the last continental ice sheet. Geografiska Annaler, 76A, 25—35. Sollid, J. L., & Torp, B. (1984). Glacialgeologisk kart over Norge, 1:1 000 000. Nasjonalatlas for Norge. Geografisk Institutt, University of Oslo. Soyez, D. (1974). Studien zur Geomorphologie und zum letztglacialen Eisru¨ckzug in den Gebirgen Su¨d-Lapplands/Schweden. Geografiska Annaler, 56A, 1—71. Sugden, D. E., & John, B. S. (1976). Glaciers and ¸andscape: A Geomorphological Approach. Edward Arnold, London. Tanner, V. (1944). Outlines of the Geography, life and Customs of Newfoundland and ¸abrador. Societas geographica Fenniae, Helsinki. Thompson, D. E. (1979). Stability of glaciers and ice sheets against flow perturbations. Journal of Glaciology, 24, 427—442. Thompson, W. B., & Borns, H. W. (Eds.), (1985). Surficial geological map of Maine, 1:500 000. Maine Geological Survey, Department of Conservation. Wastenson, L. (1983). Rogen-Myskelsjo¨omras det. In: I. Borgstro¨m, (Ed.), Geomorphological map 17C Funa¨ sdalen—Description and Assessment of areas of Geomorphological Importance, 25—31. Statens Naturvas rdsverk, PM 1709, Stockholm.