Glacial-thrust terrain

Glacial-thrust terrain consists of hills and ridges composed of material that was translocated, more or less intact, by glacial plucking or thrusting and of elongate or sub-circular depressions that were the original locations of the thrust masses. Glacial-thrust terrain ranges from isolated single hills or depressions as little as 100 m on a side to composite thrust masses covering more than 100 km². These features have been recognized throughout North Dakota (Fig. 1: 19–30), Footnote ^* Saskatchewan (Fig. 1: 5, 9, 11–18), and Alberta (Fig. 1: 1–10).

Fig. 1.

Map showing distribution of glacial-thrust terrain in the prairie region of North America.

Glacial thrust masses, like tectonic thrust masses, have a variety of gradational manifestations (Reference Moran and GoldthwaitMoran, [^c1971]). The simplest form consists of a hill or hills lying down-glacier from a depression of similar size and shape (Reference BluemleBluemle, 1970; Reference Clayton and MoranClayton and Moran, [^c1974], p. 104) Fig. 2. In many places the thrust block has been broken into a series of imbricate slabs that appear as parallel ridges on the hill. In other places the hills consist of a series of imbricate thrust sheets or overturned folds that occur in arcuate groups, which are concave up-glacier (Fig. 3.).

Fig. 2.

Glacial-thrust hill and source depression at Anamoose, North Dakota (photograph: BA _1–1 BB₋₄, U.S. Department of Agriculture).

Fig. 3.

Antelope Hills, arcuate glacial-thrust ridges in north-central North Dakota composed of imbricate slabs of Tertiary continental sediment. The hatchered line outlines the source depression (photograph: 340–6909, 11–8–52, Acro Service Corp.).

The related depression is normally partly filled with supraglacial mud-flow sediment in the form of hummocky subsidence terrain or lacustrine silt and clay. In some places the source depressions are as much as 2 km up-glacier from the hill. In other places no depression is present, and in at least some places there is evidence that the thrust block was a hill on the pre-glacial surface and was simply moved to a new location. In other places, the existence of glacial thrust features is known only from study of the sub-surface (Reference StalkerStalker, 1975, Reference Stalker1976; Reference Christiansen, Whitaker and LeggetChristiansen and Whitaker, [^c1976]; Reference ScottSauer, 1977).

The material comprising thrust sheets and folds is highly variable. Older glacial sediment is perhaps the most common constituent. It is of least interest, however, because it lacks the bedding needed to reconstruct the structure of the mass. Many thrust masses are composed of pre-existing lake or river sediment of Quaternary age. The most interesting are those known to be composed of Cenozoic or Mesozoic marine or continental sediment (Reference SlaterSlater, 1927; Reference KupschKupsch, 1962). For example, Sibley Buttes in North Dakota (Fig. 1: 28) consists of a ridge of Cretaceous marine sandstone about 2 km wide, with dozens of parallel sandstone hogbacks.

The internal structure of most thrust masses is known only from shallow road cuts and, in such cases, information is not available to distinguish between overturned folds and thrust sheets. Bedding that can be seen on air photographs on the down-glacier side of some indivi-dual thrust ridges on Sibley Buttes suggests that the ridges are the edges of thrust sheets dipping up-glacier. In contrast, Reference ByersByers (1959) and Reference KupschKupsch (1962) have concluded that many thrust masses in Saskatchewan and Alberta are overturned anticlines or thrust-faulted overturned anticlines. The depth of folding and thrusting is generally unknown but Reference KupschKupsch (1962) suggested that the vertical extent of glacier-thrust deformation in Saskatchewan is between 60 and 180 m. Reference Moran and GoldthwaitMoran ([^c1971], Reference Moranunpublished) reported glacial deformation to a depth of about 200 m beneath Thunder Hill in eastern Saskatchewan.

Streamlined terrain

Glacial terrain is characterized by a wide variety of streamlined landforms that were formed by processes operating at or near the bed of vanished glaciers (Reference Clayton and MoranClayton and Moran, [^c1974], Reference Clayton and Moran1977). These features range in size from scratches or striations less than 1 mm deep and several millimetres long to smoothed hills at least 200 m high and 20 km or more long. These features are characteristically smoothed and elongated in the direction of ice flow, as deduced from other evidence. Length-to-width ratios vary from as little as 2 to 1 to more than 250 to 1. Of the wide variety of terms applied to these features, “drumlin” and “fluting” are probably the most generally used.

In the western prairie areas of the United States and Canada, glacially streamlined forms generally are low and have large length-to-width ratios. Ridges in McHenry County, North Dakota (Fig. 4: 41), for example, are generally 1–5 km long, 45–60 m wide, and 2–5 m high. One especially well-developed ridge, Hogback Ridge, is over 20 km long, nearly 100 m wide at the base, and decreases in height from 15 to 2 m along its trend. Similar dimensions characterize streamlined forms throughout North Dakota, Saskatchewan, and Alberta. In some areas in the prairies, shorter, higher features, which are more typical of classical drumlins, have been reported (Reference KupschKupsch, 1955, p. 329) (Fig. 4: 32).

Fig. 4.

Map showing distribution of glacially streamlined terrain in the prairie region of North America.

Although streamlined terrain does not cover extensive areas in the prairies, its occurrence is widespread. Only 2% of the glaciated part of North Dakota, for instance, is occupied by such terrain (Reference Clayton and MoranClayton and Moran, 1977; Reference ClaytonClayton, in press) but streamlined features are described from 11 of 40 glaciated counties distributed throughout the state (Fig. 4: 37–48). Small areas of drumlins or flutes have been reported for Manitoba (Fig. 4: 35, 36), Saskatchewan (Fig. 4: 24–34, 49, 50), and Alberta (Fig. 4: 1–23).

Streamlined features in the prairies are variously composed of pebble-loam, gravel, sand, clay, silt, shale, siltstone, and sandstone (Fig. 4.). Observations by the authors as well as others (e.g. personal communication from Reference Mollard and MollardJ. D. Mollard, 1978) indicate that outcrops sufficiently deep to reveal the interior of streamlined features almost invariably expose granular material or deformed pre-existing rock or sediment.

A variety of structural features has been described within the ridges of streamlined terrain. Systematic sets of joints have been reported from drumlins in Saskatchewan by Reference KupschKupsch (1955, p. 330–36). Two types of folds, broad open anticlinal folds and complexly overturned anticlines that in places form diapiric intrusions into overlying materials, have also been observed. In McHenry County, North Dakota, a road cut exposes nearly the entire thickness of a low streamlined ridge composed of a sequence of interceded sand and silt capped by about 1 m of pebble-loam. These materials are folded into a broad anticline, the limbs of which parallel the surface of the ridge itself. Reference Whittecar and MickelsonWhittecar and Mickelson (1977) reported classic dikes and overturned folds within drumlins near Milwaukee, Wisconsin. The folds, which were as much as 30 m high, have axes that are parallel or oblique to the long axis of the drumlins and in some places are faulted and displaced beyond reconstruction (Reference Whittecar and MickelsonWhittecar and Mickelson, 1977, p. 216). The sediment in cores of many streamlined forms is broken by thrust faults (Reference Whittecar and MickelsonWhittecar and Mickelson, 1977). In North Dakota, the shale cores of drumlins, where they have been exposed, are generally brecciated.

A unique type of streamlined feature, which was termed a “murdlin” by Reference StalkerStalker (1973), was originally recognized in a small area north-east of Calgary, Alberta. Other examples have been found in north-central North Dakota. A murdlin consists of two parallel streamlined ridges that are seprated by a trough, which extends below the level of the surrounding plain. The down-glacier end of the trough is formed by a mound of debris that tails off on either side to form the lateral ridges. Murdlins as much as 2–3 km long occur in fields with other streamlined features. The appearance of a murdlin can best be visualized by dragging a finger through a substance such as cake frosting or peanut butter.

Evidence of thrusting associated with streamlined terrain is widespread throughout the prairies. In most areas of streamlined terrain, depressions lie immediately up-glacier from the streamlined ridges. In some places a depression is associated with each ridge, whereas in other places a large depression lies up-glacier from a broad area of fluted terrain. Where the internal structure of ridges is exposed, faulting, brecciation, and folding commonly indicate differential movement within the core of the ridge in the direction of ice flow (Reference Moran and WhitakerMoran and Whitaker, 1969; Reference Moran and GoldthwaitMoran, [^c1971], Reference Moranunpublished; Reference ClaytonClayton, in press; Reference HobbsHobbs, unpublished; Reference Fenton and AndriashekFenton and Andriashek, in preparation). North of Calgary, Alberta, near the town of Balzac, a field of small thrust blocks has been extensively streamlined (Moran, in preparation) (Fig. 4: 10). Nearly all of the larger streamlined ridges lie immediately down-glacier of a depression. Displacements between individual thrust blocks indicate the presence of numerous strike-slip faults or fault zones along which differential movement has occurred. The streamlined thrust terrain consists largely of sandstone and siltstone draped by a thin veneer of lodgement till.

Basal temperature

Most thermal models of the continental glacier that covered North America during the Wisconsin an suggest an outer thawed-bed zone several hundred kilometers wide around much of the periphery (Reference HookeHooke, 1977; Reference SugdenSugden, 1977). However, some models suggest the possibility of a narrow frozen-bed zone a few kilometres to a few tens of kilometres wide at the outermost limit of the ice sheet (Reference Clayton and MoranClayton and Moran, [^c1974], Reference Clayton and Moran1977; Reference HookeHooke, 1977; Reference SugdenSugden, 1977).

Up-glacier from this narrow frozen-bed zone, the models suggest that the bed of the glacier was at the pressure-melting point. It is generally accepted that glacial movement involved a significant component of basal sliding in areas of such thawed-bed conditions (Reference WeertmanWeertman, 1957, Reference Weertman1964; Reference Clayton and MoranClayton and Moran, [^c1974]; Reference SugdenSugden, 1977). Where frozen-bed conditions exist, glacial movement occurs primarily by internal deformation of the ice, but shear failure in the substrate may also occur if stresses exceed the strength of the material.

Because of the importance of the thermal regime to the mechanism of formation of subglacial landforms, we have made some additional calculations to evaluate the glaciological conditions under which it might have existed. The computer program described by Reference HookeHooke (1977) was modified to include the depression of the melting point as a result of pressure at the bed. The program gives the steady-state temperature distribution along a flow line from the divide to a margin (Fig. 6.). The boundary conditions and various constants used in the calculations are those given by Reference HookeHooke (1977, p. 6, 11–12), with the exception that the gradient required to remove the geothermal heat varies along the flow line, following Reference SugdenSugden (1977). The non-steady-state conditions under which glacial-thrust terrain is believed to have formed probably do not deviate sufficiently from a steady state to invalidate the general conclusions.

Fig. 6.

Calculated temperature distribution in a glacier with a maximum accumulation rale of 20 cm/year and mean annual temperature at the margin of −7.5°C. Basal temperature and melt rate (cross hatched) are shown at bottom.

The procedure used in the program for estimating the ice-sheet profile avoids arbitrary assumptions regarding the profile or basal shear stress. Instead, the primary independent variables that determine the profile—the accumulation pattern and the viscosity parameters—are estimated. A limitation of the procedure, however, is that the dependence of the viscosity parameters on temperature and fabric is not included. The viscosity parameters used are based on measurements on the Barnes Ice Cap (Reference HookeHooke, 1973 [b]) and on study of the profile of the Greenland ice sheet (Reference LliboutryLliboutry, 1968). These values are probably reasonable for situations in which basal temperatures are in the neighbourhood of 0 to −5°C and a strong preferred orientation of crystals has developed. As our primary concern in the present context is with the outermost few kilometres of a 1300 km flow line, these uncertainties in the profile will not introduce appreciable error into the conclusions.

Two important assumptions in the column model equations (Reference Budd, Budd, Jenssen and RadokBudd and others, 1971) that were used for the temperature calculations should be discussed. The first is that the effect of longitudinal advection of heat can be approximated by assuming that the warming rate at depth is proportional to the surface warming rate as ice moves to lower latitudes and elevations. Reference Budd, Budd, Jenssen and RadokBudd and others (1971) assumed a constant of proportionality of 1.0 but Reference HookeHooke (1977) found that this produced physically impossible results near the equilibrium line. Hooke arbitrarily used a value of 0.5 for most of his calculations and that value is used here. The effects of this assumption are appreciable in the lower part of the accumulation zone, where use of the lower constant of proportionality results in calculated basal temperatures that are as much as 10 deg warmer (Reference HookeHooke, 1977, fig. 4d; note that the labels on the two curves in this figure are switched). However, within a few kilometres of the margin this effect is probably negligible.

The second important assumption is that the magnitude of the vertical velocity decreases linearly with depth. This assumption results in calculated basal temperatures in the accumulation area that are probably a few degrees too cold (Reference Philberth and FedererPhilberth and Federer, 1971; Reference Whittecar and MickelsonHooke, 1977, fig. 3). The magnitude of the error in the ablation area is unknown but it is probably not nearly as large.

An additional assumption of the present model is that virtually all melt water produced by basal melting drains away from the bed into the substrate rather than flowing down-glacier along the bed and refreezing to the base in colder areas. Such refreezing could raise the tem-perature at the base of the glacier to the melting point in places where the present model calculates temperatures that are below freezing.

The approach used in the present modelling is to seek combinations of accumulation pattern and mean annual marginal temperature that result in sub-freezing temperatures in a 2 km-wide zone along the margin. The accumulation rate is assumed to be greatest 65 km up-glacier from the equilibrium line. Farther up-glacier the rate decreases linearly from this maximum value to one-quarter of it at the divide. Down-glacier the accumulation rate decreases linearly from the equilibrium line to the margin and is adjusted to give a balanced mass budget.

Figure 7 gives curves of maximum accumulation rate versus mean annual temperature at the margin for three different equilibrium-line positions—130, 185, and 260 km from the margin, respectively. A point on one of these curves represents a particular combination of maximum accumulation rate, mean annual temperature at the margin, and equilibrium-line position that will result in sub-freezing basal temperatures within 2 km of the margin and melting-point temperatures for some distance up-glacier from this 2 km position (Fig. 6). The net ablation rate at a point 2 km from the margin is given numerically at various points along the curves. This is the rate required for a balanced mass budget according to the assumed accumulation/ablation pattern described above. At the top of the figure the net ablation that might be expected for a given mean annual temperature is shown. These figures are based on measurements on the Barnes Ice Cap (Reference HookeHooke, 1973[a], Reference Hooke1976) and near SondrØstramfjord, Greenland (Reference WeidickWeidick, 1968, p. 25), and arc very approximate as they ignore many factors that have a substantial influence on ablation such as radiation balance, length of the melt season, and June snow depth. However, the pertinent observation is that if the mean annual temperature at the margin were appreciately below −10°C, for example, ablation rates would probably be too low to result in a balanced mass budget. In other words the glacier would be advancing.

Fig. 7.

Numerical model predicts basal temperatures that are below freezing within 2 km of margin for combination of accumulation rate, marginal temperature, and equilibrium-line position shown by solid curves. Dashed curves show mean annual air temperature at equilibrium line; ice temperatures there will be somewhat warmer (Fig. 6) due to the effect of percolating melt water (Hooke, 1976). Ticks are deliberately omitted from ablation-rate scale at top due to uncertainty of figures.

The mean annual temperature at the equilibrium line is shown by dashed lines crossing the curves in Figure 7. This temperature places a further constraint on allowable combinations of accumulation rate and marginal temperature, as summer conditions at the equilibrium line must be mild enough, on the average, to melt all of the snow-pack but none of the underlying ice. Mean annual equilibrium-line temperatures appreciably below −15°C probably would not satisfy this condition. This appears to place a fairly severe restriction on the maximum accumulation rate, a restriction that we are reluctant to believe. Possibly the lapse-rate used in the calculations, 0.01 deg/m, is too high but such a rate is certainly consistent with present-day lapse-rates in Antarctica (Reference Budd, Budd, Jenssen and RadokBudd and others, 1971, p. 73) and Greenland (Reference Mock and WeeksMock and Weeks, 1966, p. 29 and 31). The only other factor used that would have an appreciable effect on the resulting equilibrium-line temperature is the position of the equilibrium line itself. But moving the equilibrium line to lower elevations, closer to the margin, reduces the size of the ablation area, thus requiring an increase in the ablation rate. This in turn requires higher upward vertical velocities and thus results in higher basal temperatures (Reference HookeHooke, 1977, fig. 3A). However, if flow lines from the divide actually diverged more rapidly than radii of a circle (as is assumed in the model), this effect would be less serious.

In summary, despite the assumptions that enter into these calculations, three valid conclusions can probably be drawn from them. First, a frozen toe roughly 2 km wide could have existed at the margin of the Laurentide ice sheet in North Dakota. Thus this is a viable explanation for the observed thrust features. Farther north in Alberta, mean annual temperatures presumably were somewhat lower, resulting in a wider frozen toe, and this is consistent with an observed northward increase in the width of the marginal zone of thrust features. Secondly, for such a frozen toe to have existed in North Dakota, the mean annual temperature at the margin probably would have had to have been at least a couple of degrees less than −5°C but not much less than −10°C. This is in agreement with pollen data (Reference Wright and MahaneyWright, 1976). Thirdly, accumulation rates were probably not very large; maximum rates substantially in excess of 0.3-0.4 m/year seem unlikely. This may be the first quantitative support for the common assumption that accumulation rates on the Laurentide ice sheet were comparable to those on the Greenland and Antarctic ice sheets today, and that these present-day ice sheets may thus be reasonable analogues for the Laurentide. Models tested using other viscosity and accumulation patterns do not give any reason to question the generality of these conclusions.

Basal shear

It is reasonable to suggest that as ice advanced over an area, the area experienced first frozen-bed conditions, and then, with further advance, thawed-bed conditions. Under frozen-bed conditions, a vertical profile through the glacier and into the bed would normally consist of four zones: (1) debris-free ice, (2) debris-laden ice, (3) sediment or rock below the pressure-melting point and probably containing interstitial ice, and (4) sediment or rock that is free of ice but that is water saturated in some places. The debris-laden ice can contain interbedded layers of clean ice.

The penetration of continuous ice down into the substrate results in a firm bond between the glacier and the substrate, which permits transmission of shear stress from the moving ice to the bed. Glacial ice deforms readily at stresses of about 100 kN/m² (1 bar) and has a fundamental strength (the stress required to initiate creep) appreciably less than this. The shear strength of all but the weakest of unconsolidated sediments is much larger. For example, glacial till in Montrail County, North Dakota, has shear strength values that range from 135 to 540 kN/m² with a mean value of about 300 kN/m² (Reference ClaytonClayton, 1972). Thus, where glacial ice is frozen to substrate material that has a shear strength significantly greater than the fundamental strength of ice, movement will normally occur by plastic flow within the ice.

The fact that glacial-thrust features do not occur everywhere but rather occur in only certain areas suggests that the shear strengths of sediment and rock are generally greater than the highest shear stresses that could be applied by the glacier. It is evident, however, that in those areas where glacial-thrust structures are found the shear strength of the substrate was exceeded.

Such failure could result from the substrate material having a low initial strength, possibly because of advantageously oriented planes of weakness, such as bedding planes or fractures, or from weakening of the substrate by a decrease in cohesion or effective normal stress.

Some unconsolidated sediment, especially clayey glacial lake sediment, has a very low shear strength. Sediment of glacial Lake Agassiz, for example, has shear-strength values that range from less than 25 kN/m² to more than 65 kN/m² (Reference MoranMoran, 1972; Reference ArndtArndt, 1977). Glacially-thrust lacustrine sediment is a common constituent of some ice-marginal disturbed complexes in eastern North Dakota (Bluemle, 1975). It is entirely possible that, in these cases, the inherent weakness of the substrate material itself permitted formation of the thrust structures without other factors being involved.

In many thrust structures in the prairies, semi-consolidated to thoroughly consolidated bedrock, which has quite high shear strength, has been deformed. Although failure of such materials may have resulted from pre-existing planes of weakness within the rock mass, analysis suggests that this is unlikely. Consider a glacier too m thick with a surface slope such that basal shear stress is 100 kN/m² (1 bar). At a depth of 10 m in a substrate with a density of 2.041 Mg/m³, the shear stress will be about 100 kN/m² and the normal stress will be about 1080 kN/m². We use the shear-strength relation presented by Reference Terzaghi and PeckTerzaghi and Peck ([^c1967]),

where s is the shear strength, c is cohesion, N the normal stress, h is the pressure head, γ is the unit weight of water, and ϕ is the angle of internal friction. Taking the case of a dry cohesionless body of rock and using a friction term, tan ϕ = 0.5, a very high value for rock, gives a shear strength s of 540 kN/m², about 4.5 times the applied shear stress. Thus, the fundamental strength of the ice is so much lower than the shear strength of the substrate that ice-surface slope and thickness never build up to the point of causing failure in the substrate.

However, the shear strength of the substrate can be substantially reduced if the pressure head is large. In the example above, if the pressure head is 100 m above the failure plane, 10 m below the surface of the glacier, the shear strength is reduced to 50 kN/m². The role of pore-water pressure in controlling shear strength has been well known in soil mechanics for many years (Reference Terzaghi and PeckTerzaghi and Peck, [^c1967]). Its role in tectonic faulting was discussed by Reference KupschHubbert and Rubey (1959) and its role in glacial thrusting was discussed by Reference Moran and GoldthwaitMoran ([^c1971], Reference Moranunpublished), Reference Clayton and MoranClayton and Moran ([^c1974], 1977), Reference Christiansen, Whitaker and LeggetChristiansen and Whitaker ([^c1976]), Reference BluemleBluemle (1970), and others.

The necessity for the reduction of shearing resistance in most bed materials in order for thrusting to occur, combined with the common coincidence of glacial-thrust features with buried aquifers, strongly suggests that elevated pore-water pressures played a major role in the genesis of glacial-thrust terrain.

Origin of subglacial landforms

By combining the preceding thermal model and mechanical principles, we can construct the following model of the genesis of landforms beneath an advancing Pleistocene glacier in the prairie region. Where the basal water was free to drain from the substrate through a conduit such as a buried valley fill, it moved under a gradient of decreasing potential toward the margin of the glacier. Where the aquifer was sealed at the margin as a result of a stratigraphic pinch-out, burial of the aquifer by glacial ice, or the development of thick permafrost in front of the glacier, the elevated pressure at some distance back under the glacier was transmitted, only slightly diminished, to the marginal zone. Where this elevated pressure was high enough, shear failure occurred in the substrate and blocks from the substrate moved along with the ice.

As the glacier continued its advance over the area, some of the blocks that were sheared from the bed were carried down-glacier. Other blocks remained near the base and were deposited immediately down-glacier of their source. Deposition occurred as a result of an encounter with still frozen-bed conditions, which allowed the block to again refreeze to the substrate, or as a result of a reduction in pressure head, which increased the shear resistance along the failure plane. The importance of compressive-flow conditions in the ice in lifting blocks above the bed has been stressed by Reference Christiansen, Whitaker and LeggetChristiansen and Whitaker ([^c1976]). A complete gradation exists from blocks that were moved as little as 100 m and remained essentially intact to blocks that were moved many kilometres and, in many places, were thoroughly fragmented.

Where the glacier reached its furthest advance near the site of thrusting, the blocks were deposited within the frozen-bed zone and have been preserved more or less unmodified. Where deposition of blocks occurred more than about 2 km up-glacier from the final limit of the glacial advance, however, the frozen-bed zone moved on past the blocks and the thawed-bed zone migrated over the blocks. As the contact between glacier and the substrate thawed, the bond between them decreased until basal sliding began.

The proximal edge of the frozen-bed zone was probably irregular in most places. Variations in thickness of ice within the substrate that resulted from original irregularities in depth of ice penetration as well as differences in the rate of melting became more pronounced as the thickness of the frozen zone decreased until the zone became discontinuous. The disintegration of the frozen-bed zone undoubtedly began with the complete thawing of isolated areas of this ground ice. Closer to the thawed-bed zone, coalescence of some of these thawed areas formed continuous extensions of the thawed-bed zone into the frozen-bed zone. Finally, still farther up-glacier, isolated patches of frozen substrate surrounded by thawed-bed conditions marked the last remnants of originally thick or slowly melting ground ice.

Within this transition zone a variety of mixed and composite landforms was generated. Numerous, small thrust blocks were formed where frozen-bed conditions continued. Streamlined thrust blocks were developed where sliding conditions replaced frozen-bed conditions. Differential shear transverse to glacier movement, possibly as a result of frozen-bed conditions adjacent to thawed-bed and sliding conditions produced strike-slip faulting within thrust blocks. Midlines formed where a small remnant of frozen substrate broke loose and was dragged down-glacier plowing up the highly saturated, soft thawed substrate on either side.

Formation of landforms within the thawed-bed zone appears to occur by two principal processes: (1) erosion of the substrate and (2) emplacement of thrust masses followed by erosion and depositional smoothing. In both processes the streamlined form is the result of abrasive erosion caused by sliding of the over-riding glacier.

Erosion of pre-existing substrate material is the most likely explanation for streamlined forms that contain fluvial or lacustrine sediment with undisturbed or slightly disturbed bedding or stratigraphic sequences consisting of several distinctly different layers of glacial sediment separated by fluvial sediment.

Most streamlined forms, however, appear to have formed by a composite process of erosion and deposition. In most streamlined terrain, the presence of evident source depressions up-glacier from the streamlined ridges, combined, in some places, with exposures that reveal structural deformation, strongly suggests that glacial thrusting in the frozen-bed zone played an integral part in their formation. Evidently, the glacial-thrust blocks provided obstructions around which the ice was required to deform. As the ice converged again around the thrust-block obstruction, debris in transit in the basal zone of the glacier was deposited in the form of a lee-side tail. Because of the converging and rising flow created in the basal zone of the ice by the obstruction, pebbles were deposited with the “herringbone” fabric that has been reported from these streamlined features (Reference Gravenor and MenIeyGravenor and Menley, 1958; Reference Shaw, Suggate and CresswellShaw, 1975; Reference EvensonEvenson, [^c1971]). Continued erosion of the generally unconsolidated block at the head of the feature and deposition of till in its lee led to the streamlined form of the feature.

A second mechanism by which debris can be deposited in the lee of thrust blocks is suggested by the evidence from eastern Wisconsin presented by Reference Christiansen, Whitaker and LeggetWhittecar and Mickelson (1977). Here a significant part of some drumlins consists of deformed substrate material that has been injected upward into the drumlins. Although evidence for this type of process is scarce in streamlined landforms in the prairies, it may be important in some places. A zone of low pressure is created in the lee of obstructions on the glacier bed. Where the substrate is thawed, saturated, and highly plastic, substrate material can flow into this low-pressure zone in response to the load of the adjacent glacier. Continued movement of the glacier streamlines the form.