Morphological Analysis and Evolution of Buried Tunnel Valleys in Northeast Alberta, Canada

Quaternary Science Reviews 65 (2013) 53e72

Contents lists available at SciVerse ScienceDirect

Quaternary Science Reviews

journal homepage: www.elsevier.com/locate/quascirev

Morphological analysis and evolution of buried tunnel valleys in northeast Alberta, Canada

N. Atkinson*, L.D. Andriashek, S.R. Slattery

Alberta Geological Survey, Energy Resources Conservation Board, Twin Atria Building, 4th Floor, 4999-98 Ave. Edmonton, Alberta T6B 2X3, Canada article info abstract

Article history: Tunnel valleys are large elongated depressions eroded into unconsolidated sediments and bedrock. Received 9 August 2012 Tunnel valleys are believed to have been efficient drainage pathways for large volumes of subglacial Received in revised form meltwater, and reflect the interplay between groundwater flow and variations in the hydraulic con- 26 November 2012 ductivity of the substrate, and basal meltwater production and associated water pressure variations at Accepted 28 November 2012 the ice-bed interface. Tunnel valleys are therefore an important component of the subglacial hydrological Available online 12 February 2013 system. Three-dimensional modelling of geophysical and lithological data has revealed numerous buried Keywords: Tunnel valleys valleys eroded into the bedrock unconformity in northeast Alberta, many of which are interpreted to be Subglacial drainage tunnel valleys. Due to the very high data density used in this modelling, the morphology, orientation and Sedimentary architecture internal architecture of several of these tunnel valleys have been determined. The northeast Alberta buried tunnel valleys are similar to the open tunnel valleys described along the former margins of the southern Laurentide Ice Sheet. They have high depth to width ratios, with undulating, low gradient longitudinal profiles. Many valleys start and end abruptly, and occur as solitary, straight to slightly sinuous incisions, or form widespread anastomosing networks. Typically, these valleys are between 0.5 and 3 km wide and 10 and 30 m deep, although the depth of incision along some thalwegs exceeds 100 m. Several valleys extend for up to 60 km, but most are between 10 and 30 km long. Valley fills comprise a range of lithofacies, including stacked sequences of diamict, glaciofluvial sands and gravels and glaciolacustrine silts and clays. Displaced bedrock, presumably of glaciotectonic origin, also occurs within several anastomosing valleys. Several channel bodies are exposed along a number of valley sections suggesting progressive valley development through repeated cycles of sediment discharge. Cut-and-fill structures that are capped by fine-grained sequences of rippled sand and mud-rich drapes within these channel bodies suggest unstable flow regimes within the valley and the discharge of sediment-laden basal meltwater under flood-like conditions followed by wane flow events or periods of lower meltwater discharge, likely concomitant with localized modification by glacial ice. Basal meltwater is inferred to have been released as episodic jökulhlaups beneath the western Laurentide Ice Sheet, which at times re-used existing valley systems, which were spatially and temporally stable features, and at other times incised new valleys. Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved.

1. Introduction modern drainage (Visser, 1988; Ehlers and Linke, 1989; Eyles and McCabe, 1989; Moores, 1989; Wingﬁeld, 1990; Sugden et al., A range of formerly glaciated landscapes, in both the mid- and 1991; Brennard and Shaw, 1994; Piotrowski, 1994; Clayton et al., high-latitudes, contain distinctive overdeepened valleys that typi- 1999; Cutler et al., 2002; Russell et al., 2003; Glasser et al., 2004; cally exhibit undulating long proﬁles and trend oblique to the Denton and Sugden, 2005; Fisher et al., 2005; Kozlowski et al., 2005; Hooke et al., 2006; Kehew et al., 2012; Van der Vegt et al., 2012). These valleys can be up to 100 km long, 4 km wide and 400 m deep, and occur as integrated, anastomosing networks, or relatively straight, isolated segments (Ó Cofaigh, 1996). Collectively * Corresponding author. E-mail address: [email protected] (N. Atkinson). referred to as tunnel valleys, they are inferred to be former drainage

0277-3791/$ e see front matter Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.quascirev.2012.11.031 54 N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72 pathways for large volumes of subglacial meltwater (Piotrowski, unconformity. In this paper, we describe the distribution, mor- 1997). Their formation reflects the interplay between ground- phology and the internal architecture of these buried valleys, and water flow and variations in the hydraulic conductivity of the discuss their evolution and possible implications for the subglacial substrate, as well as basal meltwater production and associated hydrology of a previously uninvestigated region of the western water pressure variations at the ice-bed interface (Alley, 1989; Laurentide Ice Sheet. Boulton et al., 1995; Piotrowski, 1997). Tunnel valleys are therefore considered vital components of the subglacial hydrological system, 2. Study area which controls in part, ice sheet dynamics. Consequently, understanding the genesis of tunnel valleys is of crucial importance 2.1. Physiography for the reconstruction and understanding of former ice sheets (Boulton et al., 1995; Piotrowski, 1997; Eyles, 2006; Jørgensen and The study area occupies w18 000 km2 in the Northern Alberta Sandersen, 2006). Lowland region of the Interior Plains of Canada, north of the city of However, although there is a general consensus that tunnel Fort McMurray (Fig. 1; Pettapiece, 1986). It is bounded by latitudes valleys were eroded by large, channelized subglacial meltwater 56 300 and 57 300, and longitudes 110 and 112. The region is flows that were driven by the hydrostatic gradient of the overlying characterized by flat to gently undulating terrain. The morphology ice sheet, a number of issues remain (Ehlers and Linke, 1989; Ó of this terrain is largely due to the erosive effects of Paleogene Cofaigh, 1996; Huuse and Lykke-Andersen, 2000; Kehew et al., fluvial systems and glaciation during the Quaternary. The major 2012). These include: (1) the mode of meltwater drainage, and physiographic features include the broad Athabasca River Lowland whether entire tunnel valley systems formed synchronously, due to (220 m above sea level (m asl)) in the central part of the study area, the catastrophic discharge of subglacial meltwater (Brennard and which is flanked by the Birch Mountains (860 m asl) to the Shaw, 1994; Sharpe et al., 2004), or whether tunnel valleys are northwest, and Muskeg Mountain (650 m asl) to the east (Figs. 1 time-transgressive features, resulting from repeated, more con- and 2; Pettapiece, 1986). Less prominent features include the Fort tinuous meltwater discharges (Praeg, 2003; Jørgensen and Hills, and the locally high-relief of the Firebag Plains in the eastern Sandersen, 2006; Lonergan et al., 2006; Kristensen et al., 2008); edge of the study area, along the AlbertaeSaskatchewan border (2) the influence of direct glacial erosion on valley morphology (Fig. 2). Surface drainage radiates from these uplands and is cap- (Niewiarowski, 1995; Jørgensen and Sandersen, 2006); and (3) the tured by tributaries of the Athabasca and Clearwater rivers which extent to which the substrate permeability and associated varia- incise the Athabasca River Lowland. The central part of the lowland tions in groundwater flow influence tunnel valley evolution is characterized by wetlands that are poorly connected to the sur- (Piotrowski, 1994, 1997; Van Dijke and Veldkamp, 1996; Janszen face drainage system. et al., 2012). Continued research in subglacial processes and tunnel valley 2.2. Geology formation has benefited from an increase in the availability of subsurface geophysical data (Gabriel et al., 2003; Jørgensen et al., The study area is located in the northeast part of the Western 2003; Sandersen and Jørgensen, 2003; Jørgensen and Sandersen, Canada Sedimentary Basin (WCSB; Mossop and Shetsen, 1994). 2006; Janszen et al., 2012). These studies have highlighted the The subsurface distribution of bedrock units in the study area is importance of identifying tunnel valleys buried within the sub- constrained by three major unconformities. These are the pre- surface, since the principal architectural elements of their fills Devonian unconformity on the surface of the Precambrian reveal the nature of the flows that passed through them, providing Shield, the sub-Cretaceous unconformity on the surface of further insights into their genesis (Cutler et al., 2002; Russell et al., Palaeozoic rock units, and the Quaternary unconformity on the 2003). Moreover, sediments associated with infilled tunnel valleys surface of Cretaceous rocks, which constitutes the bedrock top- form aquifer systems which are becoming an increasingly impor- ography. The geological scope of this paper extends from the tant supply of potable groundwater to many cities in North America Quaternary unconformity, which represents the period of erosion and Europe (Heinz et al., 2003; Sandersen and Jørgensen, 2003; from the Late CretaceousePaleogene, to the onset of Quaternary Mehnert et al., 2004; MacCormack et al., 2005). When combined glaciation, and finally Holocene fluvial incision. Three major for- with increasingly sophisticated modelling and visualisation tech- mations that subcrop on the Quaternary unconformity are niques, this research has enabled buried valleys, and their asso- described in this paper; the lowermost McMurray Formation, the ciated fills to be described in greater detail, leading to a better Clearwater Formation, and the uppermost Grand Rapids Forma- understanding of their evolution within the subglacial hydrological tion (Fig. 3). system (Boulton and Hindmarsh, 1987; Boulton et al., 1995, The McMurray Formation is mainly comprised of fluvial and 2007a,b; Piotrowski, 1997; Piotrowski et al., 1999; Praeg, 2003; estuarine sands (Langenberg et al., 2002; Hein, 2006). In much of Jørgensen and Sandersen, 2006; Lonergan et al., 2006). However, the study area, the lowermost sediments of the McMurray For- despite the increased utilization of commercially available 2-D and mation are water-bearing. Elsewhere, these sediments are satu- 3-D seismic data, Kristensen et al. (2008) and Kehew et al. (2012) rated with bitumen of moderate to high grade (Hein, 2006). The remarked that lithostratigraphic descriptions of the infill se- lower Clearwater Formation comprises shaley glauconitic sand- quences of buried tunnel valleys remain relatively scarce. Such stone which lies conformably on the McMurray Formation, and descriptions may augment seismic stratigraphic interpretations, interfingers with the overlying clastic sedimentary sequence of the and provide further insight into the geological characteristics of Grand Rapids Formation (Carrigy and Kramers, 1975; Kramers and tunnel valleys and their modes of formation. Prost, 1986). The province of Alberta has experienced rapid expansion of oil The study area occupies part of the Athabasca Oil Sands Area, and gas exploration in recent years, resulting in a significant which comprises a w21.7 billion m3 deposit of oil-rich bitumen increase in the availability of subsurface geologic data. Andriashek (Fig. 1; Alberta Energy and Utilities Board, 2007). As a result of the and Atkinson (2007) utilized new geophysical and lithological data, rapid expansion of new and existing surface-mine and in-situ and presented a three-dimensional model of the pre-Quaternary extraction operations, the Athabasca Oil Sands Area is now one of bedrock topography in northeast Alberta, focussing particular the most industrially active regions in North America. Borehole attention on buried tunnel valleys eroded into the Quaternary logging associated with regional-scale exploration operations has N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72 55

Fig. 1. Digital Elevation Model (DEM) of Alberta, showing study area (boxed) and place names used in the text. Major oil sand deposits are shaded grey.

generated an unprecedented amount of subsurface geological in- glacial origin but of unknown extent, eroded into the bedrock formation which reveals great complexity in the Quaternary sedi- surface (McPherson and Kathol, 1977; Horne and Seve, 1991; ments overlying the bedrock unconformity. Foremost is the Andriashek, 2000, 2003; Andriashek and Meeks, 2000; Andriashek recognition of sand and gravel ﬁlled valley systems, of presumed and Atkinson, 2007). 56 N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72

Fig. 2. Physiography of northeast Alberta, showing place names used in the text.

3. Data sources and modelling borehole locations. However, no algorithm can produce a realistic model where data are absent, especially across areas of relatively Andriashek and Atkinson (2007) constructed three-dimensional high relief. Moreover, because kriging algorithms compute cell maps of the bedrock topography and the Quaternary stratigraphic values by interpolation methods, whereby the x, y, z values in sequence using data from a number of sources (Figs. 4 and 5A and each grid cell depend in part on the values in adjacent cells, B). Prior to this project, Alberta Geological Survey data holdings for discontinuous units are difficult to model. Such problems have the study area totalled w2500 borehole logs and surface outcrop been addressed by digitizing hand-drawn contours, and intro- descriptions. An additional 35, 000 new borehole records were ducing bounding grids. provided by industries located within the study area (Fig. 4). These In areas of poor data control, or where kriging algorithms un- records provide lithological descriptions of the well cuttings, as derestimate the subsurface topography, such as along valley thal- well as down-borehole geophysical data (gamma-ray and re- wegs, synthetic data such as three-dimensional polylines were sistivity) and have aided in determining the elevation of formation used to manually contour the surface of each geological unit. tops within the region. The distribution and density of the data are Polyline nodes provide additional x, y, z data, which were added to highly varied, with most of the data clustered around major the modelling dataset to ensure that the geological model reflects surface-mine operations (Fig. 4). the conceptual understanding of the subsurface. Bounding grids Digital maps of the subsurface bedrock topography were cre- comprise hand-digitized polygons that outline the subcrop of each ated using VIEWLOGÔ modelling software by three-dimensional formation, and segregate the study area into discrete kriging do- interpolation of the x, y, z co-ordinates of the geological forma- mains. These ensure that the geological surfaces are not interpo- tion picks. The geological surfaces picked in this study were the lated across grids where they are known not to occur. top of the McMurray and Clearwater formations, as well as bed- The resulting 3-D geological model was tested and refined by rock units overlying the Clearwater Formation, which mainly field mapping and the systematic inspection of geological cross- comprise the Grand Rapids Formation. These formations subcrop sections. Aerial surveys were completed to locate key strati- on the Quaternary unconformity, and collectively define the graphic sections exposed throughout the study area. Subsequent contact between the bedrock and the overlying Quaternary sed- fieldwork was conducted on these sections, which were described iments (Fig. 3). Given the abundance of closely spaced data, using standard stratigraphic techniques. These stratigraphic logs a100m3 grid cell size was used to krige the elevation of each of were used to verify the geological model, as well as provide further these surfaces. This technique uses linear least squares estimation data to refine each modelled surface. Further testing was per- algorithms to interpolate the x, y, z co-ordinates of each geological formed by constructing cross-sections across the geological model surface in every grid cell, including those far removed from to ensure newly completed bedrock surfaces honoured known N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72 57

Fig. 3. Oblique, hill-shaded 3-D model (with 30 vertical exaggeration) of the bedrock unconformity of northeast Alberta, showing the subcrop of bedrock geological units beneath the Quaternary succession. stratigraphic relationships and respective picks. Inconsistencies or 4. Buried valleys errors in the geometry of each surface were evaluated, and when necessary, the reliability of the picks constraining that surface was 4.1. Occurrence checked in the modelling database. In the case of unreliable data, logs were either re-interpreted, or removed from the modelling As a result of high data density within the study area, the dataset. morphology and orientation of a number of valleys eroded into the

Fig. 4. Distribution of borehole data and mapped subsurface valley aquifers in northeast Alberta, including named tunnel valleys referred to in text (shaded grey). 58 N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72

Fig. 5. A. Oblique, hill-shaded 3-D model (with 20 vertical exaggeration) of the bedrock unconformity of northeast Alberta. B. Oblique, hill-shaded 3-D model (with 20 vertical exaggeration) of sediment thickness in northeast Alberta.

Quaternary unconformity have been determined (Figs. 3, 4 and 5A). orientation of these valleys parallels streamlined bedforms eroded Abutting the west slope of Muskeg Mountain, on the Athabasca into till and bedrock in the study area (Bayrock, 1971; Bayrock and River Lowland, the Kearl Valley comprises a widespread network of Reimchen, 1974). These regional fluting and drumlin tracts dem- anastomosing valleys incised into the McMurray Formation. On the onstrate that the Athabasca River Lowland was the focus for con- south flank of Muskeg Mountain, the Lewis and Clarke valleys form vergent ice flowing west-northwestward along the Clearwater a rectilinear series of incisions within the Clearwater Formation. River valley, and southwest across the Firebag Plains and the The Willow and Birch valleys form part of a suite of isolated, adjacent flanks of Muskeg and Birch mountains. straight to slightly sinuous sub-parallel valleys that occur on the western Athabasca River Lowland. These valleys are eroded into the 4.2. Morphology Clearwater Formation, and extend discontinuously to the west, across the adjacent uplands (Figs. 4 and 5A). Elsewhere on the The Birch and Clarke valleys reach respective depths of 80 and western Athabasca River Lowland, the North and South Spruce 50 m below the Quaternary unconformity, and are the most deeply valleys form a rectilinear network of incisions into the McMurray incised buried valleys in the study area. Moreover, the interfluves of and Clearwater formations. the Clarke Valley are incised by subsidiary valleys, up to 30 m deep The distribution of these valleys occurs predominantly within that extend parallel to the main valley. The remaining buried val- two major orientations. The Clarke and Lewis valleys, as well as leys are mostly between 10 and 30 m deep. Most valleys are be- tributaries of the Kearl Valley exhibit a northwest orientation, and tween 0.5 and 1.5 km wide, although in places, the width of the the Willow, Birch, Spruce and Kearl valleys trend southwest. The Birch and Clarke valleys increases to w3 km. These data show that N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72 59 the majority of the buried valleys have high depth to width ratios, sands and structureless, normally graded coarse-grained sands are as demonstrated by the steep-sided, V-shaped cross-sections, also present within the assemblage. Thin beds of laminated silt and particularly the Willow, Clarke, Birch and Spruce valleys. Never- massive matrix supported gravel, containing local and erratic theless, some valleys, notably branches of the Kearl, have more U- boulders and cobbles are also present within the assemblage but shaped cross-sections. The Birch and Kearl valleys are the longest in are considered rare. At each section, sand-rich lithofacies are the study area, extending for 46 and 60 km, respectively. Shorter overlain by moderately clast-rich, matrix supported sandy diamict valleys extend for 10e30 km, while branches of the Kearl Valley are (LA2005-03), which constitutes a second lithofacies assemblage typically <6 km long. Valley longitudinal profiles are irregular, with within the Lewis Valley. Clasts comprise rounded cobbles and numerous examples of localized overdeepening, typically boulders, and are dominated by locally derived sandstone, although associated with valley bends and constrictions or where occasional quartzite and Shield erratics also occur. Although indi- tributaries join the main valley. Longitudinal gradients for the vidual beds of this diamict are typically massive, some contain main valleys are generally low, ranging from <0.5 to 4 m/km, stratification, where deformed interbeds of sand, commonly with whereas gradients for branches of the Clarke and Kearl valleys are concave-up lower contacts occur close to the basal contact with the greater, reaching up to 10 m/km. underlying sand-rich lithofacies, and lenses of massive, matrix supported gravel occur towards the top of the diamict sequence. 4.3. Valley fills Larger intraclasts of massive sand are also present within this lithofacies assemblage and are also likely derived from underlying 4.3.1. Kearl Valley sand-rich lithofacies. Towards the top of some diamict beds, the Geophysical and lithologs indicate that the Kearl Valley is boundary with the overlying sand is characterised by discontinuous infilled by up to 60 m of Quaternary sediment overlying the layers of pebbles, often one clast thick. The third lithofacies McMurray Formation (Figs. 5A, B and 6). These sediments comprise assemblage identified within the Lewis Valley comprises a clast- an alternating sequence (up to w50 m thick) of coarse-grained poor, matrix supported silty diamict, occasionally interbedded sediment, inferred to be sand and gravel, overlain by a laterally with sandy diamict. Rare clasts comprise rounded pebbles and extensive, <10 m thick silt-rich diamict containing discontinuous cobbles, of a similar lithology to those described in the sandy dia- sand interbeds. In the southern part of the main valley, sands and mict. The contact with underlying assemblages is convoluted to gravels are interbedded with discontinuous units of undifferenti- deformed. ated fine-grained glaciogenic sediment and silty diamict. The architectural elements of the upper 15 m of this infill sequence have 4.3.3. Clarke Valley been examined at a section exposed along the Firebag River Geophysical and lithologs indicate that the Clarke Valley is (LA2005-08) and three major lithofacies assemblages, comprising infilled by 25e125 m of Quaternary sediment, comprising a south- alternating, stacked units were identified (Fig. 7). The major lith- eastward thickening sequence of sand and gravel (up to 110 m ofacies assemblage comprises up to 50% of the logged section, and thick) overlying the Clearwater Formation (Figs. 5A, B and 9). In consists of massive to horizontally bedded medium- to coarse- many places, the top of this sand and gravel sequence is close to the grained sands. Up to 2 m thick beds of massive to laminated silt modern land surface, and is mantled by 2e10 m of sandy diamict and massive matrix supported gravel comprise minor lithofacies with discontinuous sand interbeds. However, in the central part of assemblages within this infill sequence. the valley, the sands and gravels have been incised by an oblique The sand and gravel sequence in the southern part of the Kearl channel that is filled with up to 80 m of with silty diamict and Valley has been incised by a w50 m deep channel that trends undifferentiated fine-grained glaciogenic sediment. This sand and obliquely to the main valley and is filled with diamict. This sand and gravel sequence extends w5 km laterally across the adjacent in- gravel sequence extends w2.5 km laterally across the adjacent interfluves of the southern parts of the Clarke Valley, but appears to terfluves of the northern parts of the Kearl Valley, but to the south, be increasingly restricted to the confines of the northern and cen- appears to be increasingly restricted to the confines of the main tral parts of the valley. valley and its tributaries. Borehole lithologs also show numerous The logs show a high degree of lithological variability within the blocks (4e5 m thick) of glacially displaced Clearwater Formation sand and gravel sequence, ranging from uniform sand, to sand shale in the valley-fill sequence, particularly along the contact be- containing thick beds of fine-grained sediment, as well as dis- tween the sands and gravels and the diamict. continuous interbeds of gravel. In the eastern part of the valley, the sand and gravel sequence contains a number of displaced blocks of 4.3.2. Lewis Valley undifferentiated bedrock and Clearwater Formation shale. In the Geophysical log data, supplemented with borehole lithologs western part of the valley, borehole lithologs show that basal demonstrate that this valley is infilled by 30e80 m of Quaternary sediment consists of up to 4 m of gravel, overlain by 5e40 m of sand sediment, consisting of a laterally restricted sequence of sands and and silty sand. gravels (up to 50 m thick) containing discontinuous interbeds of a sand-rich diamict (Fig. 8). The sand and gravel sequence is over- 4.3.4. Willow Valley lain by a laterally extensive, 50 m thick silty diamict interbedded Geophysical and lithologs indicate that the Willow Valley is with discontinuous units of undifferentiated fine-grained sediment infilled by 25e60 m of Quaternary sediment, comprising a north- and discontinuous interbeds of coarse grained stratified sediments eastward thickening sequence of sand and gravel (up to 45 m thick) (Figs. 5A, B and 8). The architectural elements of the upper 30 m of floored on the Clearwater and McMurray formations in the western this infill sequence has been examined at three sections (LA2005- and eastern parts of the valley respectively (Figs. 5A, B and 10). The 01 to LA2005-03) exposed along the Steepbank and North Steep- top of this sand and gravel sequence is close to the modern land bank rivers (Figs. 2 and 7) and three major lithofacies assemblages, surface, and is mantled by 2e30 m of silty diamict. The comprising alternating, stacked units were identified. The first architectural elements of the upper 20 m of the infill sequence lithofacies assemblage comprises up to 50% of the logged sections, has been examined at a section (LA2005-05) exposed along the and consists of sheet or lens-shaped bodies of ripple-cross lami- Dover River (Figs. 2 and 10) and three major lithofacies assem- nated fine- to medium-grained sands, horizontal, and trough cross- blages comprising alternating, stacked units were identified bedded medium- to coarse-grained sands. Deformed units of silty (Figs. 5A, B, and 7). The first lithofacies assemblage comprises up to Fig. 6. Physiography of the Kearl Valley, showing valley cross- and longitudinal sections, and location of LA2005-08. Litholog and cross-section legend as in Fig. 8. .Akno ta./Qaenr cec eiw 5(03 53 (2013) 65 Reviews Science Quaternary / al. et Atkinson N. e 72 61

Fig. 7. Stratigraphic logs from the study area. Lithofacies code (Table 1) is that of Benn and Evans (1998), modiﬁed from Eyles et al. (1983). 62 N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72

Table 1 Lithofacies coding scheme of Benn and Evans (1998), modiﬁed from Eyles et al. (1983).

Diamict Boulders Gravels Sands Silts and clays Dmm Matrix-supported, BL Boulder lag Gms Matrix-supported, Sh Horizontally bedded or Fl Fine lamination with massive or pavement massive low angle cross-lamination minor fine sand Dms Matrix-supported, Bcm Clast-supported, Gmi Clast-supported, Sp Planar cross-bedded Fm Massive stratified massive massive (imbricated) _ _ _ (c) Evidence of current Gp Planar cross-bedded Sl Horizontal and draped _ _ _ (d) With dropstones reworking lamination _ _ _ (s) Sheared Sm Massive Sd Deformed bedding Su Fine to coarse with broad shallow scours and cross-stratification St Trough cross-bedded Sr Ripple cross-laminated Scr Climbing ripples

60% of the logged section, and consists of ripple-cross laminated 3 cm, and comprise sand beds ranging from 3 to 0.5 cm thick, and fine- to medium-grained sands, horizontal and trough cross- silt laminae <1 cm thick. At the base of the section, these sands are bedded medium-grained sands. Deformed beds of sand (0.3e2m inversely graded, and coarsen upwards into trough cross-bedded thick) and laminated silt (up to 0.3 m thick) occur towards the sand, whereas further up the section, they are normally graded, base of this assemblage. fining upwards into rippled silty sand. This sand-rich lithofacies alternates with a bolder-rich lith- The second lithofacies comprises w15% of the section and ofacies, which constitutes a second lithofacies assemblage within consists of a massive sandy diamict. Clasts are rounded, and are the Willow Valley. This assemblage comprises a thin (w3 cm), dominated by carbonates and quartz sandstone cobbles, with oc- discontinuous boulder lag resting on bedrock along the valley floor, casional Shield erratics. This unit does not appear to extend later- and a 2.5 m thick interbed of massive, matrix supported gravel ally across the entire section. The base of the slope is mantled by containing poorly sorted, sub-rounded granules to boulders. This colluvium, which contains numerous well-rounded boulders. gravel interbed is laterally continuous across the section, and has However, it remains uncertain if these originated from the sands, or a lenticular geometry with erosional boundaries. Lithologically, the from the overlying diamict. gravel is dominated by local derived sandstone, although occa- The upper sequence of the South Spruce Valley infill has been sional quartzite and Shield erratics also occur. exposed by recent oilsands mining excavations, and contains at Gamma logs suggest that the top of sand and gravel lithofacies is least 10 m of Quaternary sediments, comprising multistory sheets partially unconstrained by channel margins, and extends w1km of gravel interbedded with sand, overlain by diamict and glaciola- laterally across the adjacent bedrock surface. The third lithofacies custrine sediment (Fig. 13). Gravel-rich lithofacies comprise poorly assemblage identified within the Willow Valley comprises a mas- sorted, inversely graded, clast supported cobbles and boulders sive, matrix supported diamict, which is capped by glaciolacustrine within a sand matrix and occupy laterally discontinuous sheets (up sediment. This lithofacies assemblage is similar to that described in to 2 m thick) with concave-up erosional bases. This lithofacies the North Spruce Channel, 10 km to the south. typically grades into a sand-rich lithofacies comprising sheet-like bodies (20 cme1 m thick) containing horizontal beds of well- 4.3.5. Birch Valley sorted medium to fine sand, interbedded with planar cross-beds Geophysical and lithologs indicate that the Birch Valley is of sand and fine gravel, capped by thin sheets of laminated fine infilled with up to 90 m of Quaternary sediment consisting of sand. In places, this sand-rich lithofacies is truncated by a south- a laterally restricted sequence of sands and gravels (up to 75 m westward dipping sequence of alternating planar cross-bedded thick) overlying the Clearwater Formation (Figs. 5A, B and 11). The gravels, often one clast thick, and poorly sorted sand and fine sand and gravel sequence is interbedded with silt and clay, as well gravel. Long axes of imbricated clasts (Gmi) at the top of the as diamict (Andriashek, 2000), and is overlain by a laterally sequence in the South Spruce Valley show dominant flow direction extensive, up to 50 m thick silty diamict containing discontinuous to west-southwest, opposite to not only the gradient of the present- sand interbeds. These sediments thin abruptly in the eastern part of day landscape (north), but also to the northerly slope of the valley the valley, as it enters the Athabasca Lowland. thalweg (Fig. 13c).

4.3.6. Spruce Valley 5. Discussion Gamma and lithologs demonstrate that the North and South Spruce valleys are infilled by up to 50 m of Quaternary sediment, High-quality and closely spaced borehole data have been used to consisting of a laterally restricted sequence of stacked, alternating model the 3-D distribution and morphology of buried valleys ero- sands and gravels with discontinuous beds of silty diamict, overlain ded into the Quaternary unconformity in northeast Alberta by diamict of variable grain size and glaciolacustrine sediments (Andriashek and Atkinson, 2007). A number of valleys originate (Figs. 5A, B and 12). The architectural elements of the upper 15 m of from drainage divides on the bedrock surface, and are therefore the North Spruce Valley infill sequence has been examined at two interpreted as pre-glacial, subaerial valleys. However, the valleys sections (LA2005-04 to LA2005-04a) exposed along the MacKay described in this paper exhibit undulating longitudinal profiles, in River (Figs. 2 and 7) and two major lithofacies assemblages were some cases extending oblique to the bedrock slope, lack any sig- identified. The first lithofacies comprises w75% of the logged sec- nificant catchment areas, often display abrupt initiations or ter- tion, and consists of an alternating sequence of horizontally bedded minations, and have high depth to width ratios. Five distinct sand/silt couplets, and trough cross-bedded medium to coarse- macroform-scale architectural elements and associated lithofacies grained sands. Individual couplets vary in thickness, from <1to (c.f. Miall, 1985) have been identified within the infill sequence of Fig. 8. Physiography of the Lewis Valley, showing valley cross- and longitudinal sections, and location of LA2005-01, -02 and -03. 64 .Akno ta./Qaenr cec eiw 5(03 53 (2013) 65 Reviews Science Quaternary / al. et Atkinson N. e 72

Fig. 9. Physiography of the Clarke Valley, showing valley cross- and longitudinal-sections. Litholog and cross-section legend as in Fig. 8. N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72 65

Fig. 10. Physiography of the Willow Valley, showing valley cross- and longitudinal sections, and location of LA2005-05. Litholog and cross-section legend asinFig. 8. these valleys; 1) channels; 2) gravel bars and bedforms; 3) sandy associated with the collapse of confined sheet flows into distribu- bedforms; 4) lateral accretion deposits; 5) overbank deposits. These tary channel systems. Such variations would have reduced shear elements largely comprise stacked, alternating lithofacies within stress along the bar flank, resulting in lateral accretion and bar multistory bodies (Gibling, 2006). A sixth architectural element migration at an oblique angle to the main flow (Allen, 1983; Miall, comprises diamict (c.f. Boyce and Eyles, 2000), which occurs as 1985). The presence of lateral accretion deposits within the lith- discontinuous interbeds within the infill sequence, as well as lat- ofacies assemblage therefore suggests that the morphology and erally extensive sheets that mantle these sequences and extend infill history of these tunnel valleys may be in part due to cut-bank across the adjacent bedrock interfluves. Collectively, these mor- erosion concomitant with bar migration (Miall, 1985; Gibling, phological and sedimentological characteristics are considered 2006; Van Dijk et al., 2009). Lenticular gravel interbeds suggest diagnostic of tunnel valley evolution within the subglacial hydro- that during flow regime fluctuations, bar complexes were dissected logical system. by minor channels or chutes, which where subsequently infilled. These gravel bars, bedforms and channel elements are inter- 5.1. Valley infill bedded with sandy bedforms which comprise up to 60% of the logged sections described in this paper. Sheet or lens-shaped bodies Gravel-rich lithofacies (Gms, Gmi, Gp, Bcm, BL) occur within of horizontally bedded and massive sand record ongoing deposi- most of buried tunnel valleys in the study area and comprise a suite tion during upper-flow regimes, although the gradational contact of macroforms that include tabular sheets, downstream and lateral between these sand bodies and the underlying gravel bedforms accretion deposits and channelized lenses. Collectively, these document sedimentation within progressively lower energy flow macroforms constitute part of composite-compound bar complexes regimes, culminating with the deposition of wane flow lithofacies. deposited within the proximal reaches of the tunnel valley system. These include trough cross-bedded sand, rippled sand and lami- Gravel sheets are interpreted as having been deposited by high- nated fine sand, which record the progression from dune migration, energy, confined sheet flows (Van Dijk et al., 2009) associated deposition within sand-rich traction currents and suspension set- with abrupt outbursts of subglacial meltwater reservoirs along tling (Eyles and McCabe, 1989; Heinz et al., 2003; Russell et al., basal tunnels (Bell, 2008). 2003). Collectively, these sandy bedforms record the progradation The deposition of massive imbricated gravel within tabular of a downstream accretionary element (longitudinal bar) within sheets indicates that sediment was initially transported as bedload the composite-compound bar complex. under high-energy conditions (Harms et al., 1982; Hein, 1982; The architecture of these bar complexes documents the super- Heinz et al., 2003). During upper-flow regimes, the addition of imposition of morphologically and texturally distinct elements due clasts to the leading edges of the gravel sheets would have accel- to the downstream migration of a succession of accretionary bed- erated progradation. Planar cross-bedded gravels are interpreted as forms across former bar positions. The stacking of these elements lateral accretion deposits resulting from variable flow directions within multistory bodies, which switch from erosion-dominated 66 .Akno ta./Qaenr cec eiw 5(03 53 (2013) 65 Reviews Science Quaternary / al. et Atkinson N. e 72

Fig. 11. Physiography of the Birch Valley, showing valley cross- and longitudinal sections. Litholog and cross-section legend as in Fig. 8. N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72 67

Fig. 12. Physiography of the Spruce Valley, showing valley cross- and longitudinal sections, and location of LA2005-04, -04a. Litholog and cross-section legend as in Fig. 8. 68 N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72

sheet flows and channelized flows suggests that the loci of aggra- dation associated with each jökulhlaup was spatially and temporally variable, due in part to allogenic variations such as the glaciostatic pressure gradient and the volume of the subglacial reservoir and its location relative to the tunnel valley system. Autogenic controls such as changes in the gradient and cross- sectional area of the tunnel valley may also have contributed to the behaviour of each jökulhlaup vis-à-vis the location and timing of the transition from sheet flow to channelized flow and the effect this had on valley evolution (c.f. Whipple et al., 1998; Hampson et al., 1999; Van Dijk et al., 2009). The sandy and silty diamict facies are both interpreted as till which, where massive, and containing deformed interbeds sheared from underlying sand-rich lithofacies was likely deposited within the amalgamation zone of a subglacial deforming layer. Laterally discontinuous beds of till within sand- and gravel-rich valley-fill sequences suggests that ice recoupled with the bed and accreted till sheets during jökulhlaup quiescence. These till sheets were then incised and covered by sand and gravel during subsequent jökulhlaups. The occurrence of undeformed intraclasts of sand and Clearwater shale indicate that the overriding ice was effective at plucking or thrusting rafts of pre-existing material and emplacing it within the valleys (Moran et al., 1980; Evans and Campbell, 1992). Intertill lenses of massive gravel within the laterally extensive sequences of till represent the infills of canals that developed at the ice-bed interface and record relatively low-energy subglacial meltwater flow and reworking/winnowing of the underlying till (Walder and Fowler, 1994; Benn and Evans, 1996).

5.2. Tunnel valley evolution in northeast Alberta

The distribution of gravel-rich lithofacies overlying bedrock at the base of the infill sequence in a number of valleys suggests that jökulhlaups were characterized by high-energy confined sheet flows that scoured the valley floors. This relationship between scoured bedrock and upper-flow regime macroforms suggests that the same kind of discharges that deposited the sediment was responsible for the incision of the valleys (c.f. Lonergan et al., 2006). However, variations in the lateral extent of such deposits present the possibility that at least in some cases, the initial stages of tunnel valley evolution may have resulted from different modes of subglacial discharge. For example, laterally extensive, sheet-like sand and gravel bodies beyond the margins of the Clarke and Willow valleys may have been deposited by broad-front jökulhlaups that subsequently incised the bed, channelizing the flow and sealing subglacial meltwater drainage within the resulting valleys as the ice recoupled with the ice-bed interface. Alternatively, these laterally extensive sand and gravels may have been deposited as overbank deposits within local separations along the ice-bed interface during later stages of tunnel valley evolution. Such separations could occur if jökulhlaup volumes exceeded the bankfull discharge of the pre- established valleys, or if roof collapse down-flow impeded drainage, so that increased water pressure in the valleys would reverse Fig. 13. (A) The upper 10 m of the South Spruce Valley infill. (B) Detailed view of the the hydraulic gradient and initiate decoupling along the ice-bed South Spruce Valley section, showing poorly sorted, matrix supported, gravel, conformably overlain by planar and trough cross-bedded sand and fine gravel, overlain by interface adjacent to the valley (Walder and Fowler, 1994; diamict and glaciolacustrine sediment. Sledge hammer (0.6 m long) is circled for scale. Boulton et al., 1996; Ng, 2000; Kehew et al., 2012). The system (C) Imbricated clasts within the poorly sorted, matrix supported, gravel. Sledge ham- would be stabilized when water from the ice-bed interface col- mer circled for scale. Lithofacies coding scheme is presented in Table 1. lapsed back into channelized drainage (Kehew et al., 2012). Nev- ertheless, the laterally constrained distribution of upper-flow gravel sheets, to succession-dominated downstream and lateral regime macroforms within the remaining valley-fill sequences accretionary deposits indicate that tunnel valleys were reused described in this paper indicates that in most cases, incipient tun- during successive jökulhlaups. This suggests that subglacial melt- nel valleys were eroded by jökulhlaups resulting from the drainage water flowed along the ice-bed interface at the same location on of subglacial reservoirs through a network of basal tunnels (Bell, a number of occasions. This periodic alternation between confined 2008). N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72 69

Once established, the spatially and temporally variable distri- the margin of the western Laurentide Ice Sheet. Moreover, evidence bution of stacked, alternating erosion- to succession-dominated for the advection of till into valley-fill sequences during jökulhlaup multistory bodies, interbedded with till further suggests that tun- quiescence demonstrates that the region was overridden by nel valleys evolved due to continued incision, headward erosion actively flowing, warm-based ice. Collectively, these observations and coeval back-filling during a succession of jökulhlaups (Praeg, suggest that the region occupied the thawed-bed zone at the time 2003; Kristensen et al., 2007; Janszen et al., 2012). Back-filling of tunnel valley evolution (c.f. Cutler et al., 2000). likely initiated with bar deposition, causing the flow to bifurcate If tunnel valleys were initially eroded into a more permeable and erode the valley walls, further decelerating flow and promoting substrate, such as the McMurray Formation, then Darcian flow may the development of bedforms (c.f. Hampson et al., 1999; Van Dijk have discharged groundwater towards the low-pressure valleys et al., 2009). Under such conditions, flows persisted over a long (Hackbarth and Natasa, 1979; Piotrowski, 1997; Huuse and Lykke- period relative to a single jökulhlaup, which likely eroded a smaller Andersen, 2000; Cutler et al., 2002; Hooke et al., 2006; Janszen cross-sectional area than the total area of the tunnel valley. The et al., 2012). If the groundwater drainage regime was sustained variable stratigraphic position of jökulhlaup deposits within these for long enough, then the resulting tunnel valley network may have tunnel valleys indicates that headward erosion and back-filling become sufficiently integrated to draw down groundwater pres- occurred during regionally asynchronous discharges. Such evi- sures over a large enough area of the McMurray Formation for it to dence either suggests that subglacially impounded water was dis- act as a sink for groundwater flow, and thus become self-sustaining tributed in multiple basins to the northeast of the study area, or (Röthlisberger, 1972; Boulton et al., 2007a). These perennial valleys different valleys operated at different times. would remain focal points for enhanced subglacial discharge once The occurrence of till interbeds within the valley-fill sequences a jökulhlaup was triggered (Piotrowski, 1997; Cutler et al., 2002; indicates that as water pressure dropped during the waning stages Hooke et al., 2006). Therefore, we suggest that although some of each jökulhlaup, the ice closed to partly fill the valley, advecting branches of the Kearl Valley may have originated as subaerial val- till along ice-bed interface, likely concomitant with localized gla- leys on the western slope of Muskeg Mountain, the long-term ge- cial modification (c.f. Kehew et al., 2012). Furthermore, as the ometry of this valley network evolved as the subglacial meltwater subglacial drainage system continued to evolve, the headward system switched periodically from a stable, distributed mode, to erosion of tunnel valleys into their feeder basins may have reduced more a more dynamic, discrete condition (Andriashek and the storage capacity of subglacial reservoirs, thereby explaining Atkinson, 2007; Boulton et al., 2007b). This regime may have why upper-flow regime macroforms are less common in the upper been augmented by the proximity of the Kearl Valley to the Birch parts of the tunnel valley-fills, which become increasingly domi- and Muskeg mountains, which likely concentrated groundwater nated by laterally continuous sheets of till. Collectively, this discharge towards the Athabasca River Lowland during stable lithostratigraphic evidence suggests that a succession of jökulh- modes of the subglacial hydrological system. Moreover, the abun- laups occurred while the retreating margin of the western Lau- dance of glacially displaced bedrock within the sedimentary infill of rentide Ice Sheet occupied relatively stable positions an unknown the Kearl Valley may also reflect the relationship between valley distance to the south and west, beyond the extent of the study evolution and substrate permeability. Since porewater pressures in area. the surroundings of the Kearl Valley would likely balance ice These lithofacies-based interpretations support episodic out- pressures, the shear strength of the substrate would decrease, burst and back-filling models developed for buried tunnel valleys thereby creating unstable glacier flow and facilitating the prefer- based on seismic data in the North Sea basin and onshore regions of ential erosion of the valley walls (c.f. Van Dijke and Veldkamp, Denmark. These models proposed that channelized subglacial 1996; Piotrowski, 1997; Boulton et al., 2007a). Such erosion may meltwater drainage systems transported material incised and/or explain why branches of the Kearl Valley exhibit more U-shaped reworked from further up-valley and deposited it along the valley cross-sections than in other valleys in the study area. In contrast, if in a subglacial conveyer-belt system (Praeg, 2003; Lonergan et al., Darcian flow through the Clearwater Formation was insufficient to 2006; Kristensen et al., 2007, 2008). discharge all of the water produced by basal melting, then the accumulation of excess meltwater may have increased basal water 5.3. Effect of the substrate pressure, initiating local separations along the ice-bed interface and enhancing discharge along channels melted upwards into the ice, Recent studies in Denmark and Poland have suggested that re- and downward into the substrate (Walder and Fowler, 1994; lationships may exist between the location of tunnel valleys and the Boulton et al.,1996; Ng, 2000). Consequently, the straight to slightly hydrogeological properties of the substrate (Hermanowski, 2010; sinuous, isolated tunnel valleys distributed across the Clearwater Sandersen and Jørgensen, 2012). In this paper, we note that the Formation may represent erosion by an ephemeral system that was location of tunnel valleys on the subcrop map of northeast Alberta dominated by discrete drainage of episodic jökulhlaups, but was appears to support the Danish and Polish studies. However, we otherwise unable to capture sufficient quantities of groundwater to further suggest that there may also be a relationship between develop into an integrated network. tunnel valley morphology and substrate lithology (Fig. 3). Anastomosing valley networks, notably the Kearl Valley are 5.4. Generations and ages eroded into the McMurray Formation, whereas the straight to slightly sinuous valleys occur primarily within the Clearwater Based on the wide variety of preferred orientations and cross- Formation. Piotrowski (1997) demonstrated that the ability of the cutting relationships, successive generations of tunnel valleys substrate to discharge meltwater from the glacier bed played have been identified in Denmark and the North Sea basin (Praeg, a significant role in controlling tunnel valley formation. Moreover, 2003; Jørgensen and Sandersen, 2006; Lonergan et al., 2006; Grasby and Chen (2005) proposed that there may be a basin-scale Kristensen et al., 2007). These are considered to have evolved relationship between subglacial recharge and the hydrodynamics during successive ice advance-and-retreat cycles, potentially of the WCSB and landscape evolution in northeast Alberta. spanning multiple glaciations. Similarly, the distribution of buried Although little is known about the distribution of permafrost in the tunnel valleys in northeast Alberta occurs predominantly within region at the time of tunnel valley development, the absence of two major orientations. The Clarke and Lewis valleys, as well as downstream outwash fans demonstrates that they evolved inside tributaries of the Kearl Valley exhibit a northwest orientation, and 70 N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72 the Willow, Birch, Spruce and Kearl valleys trend southwest. Quaternary glaciations, and have been attributed to repeated However, due to the lack of clear cross-cutting relationships else- jökulhlaups along pre-existing subglacial erosion pathways (Praeg, where in the study area, it remains uncertain whether other 2003; Jørgensen and Sandersen, 2006; Lonergan et al., 2006). northwest-orientated valleys relate to an earlier phase of valley Repeated outbursts of subglacial reservoirs have been invoked to erosion relative to those trending southwest-northeast. Never- explain time-transgressive tunnel valley evolution elsewhere along theless, even if northwest orientated tunnel valleys preceded the southern Laurentide and Fennoscandian ice sheets (Piotrowski, southwest trending ones, the occurrence of confluence scours 1997; Cutler et al., 2002; Praeg, 2003; Hooke et al., 2006; Jørgensen where the Kearl Valley intersects its tributaries suggests that at and Sandersen, 2006; Lonergan et al., 2006; Janszen et al., 2012). least in this network, northwest and southwest trending valleys Although the chronostratigraphy of tunnel valleys in northeast were active at the same time. Instead, these variations in tunnel Alberta remains uncertain, the occurrence of valley segments that valley orientation may relate to the distance from the ice margin cross-cut axially and contain overdeepened thalwegs, together that the incipient valleys formed. In the submarginal settings that with the variable stratigraphic position of upper-flow regime these tunnel valleys are interpreted to have evolved, such varia- macroforms similarly record a time-transgressive sequence of tions may relate in part to a reorganization of the subglacial hy- jökulhlaups, potentially occurring in a series of hydraulic cycles drological system in response to major shifts in ice flow trajectory. beneath the western Laurentide Ice Sheet. In contrast to previous The Clarke and Lewis valleys extend subparallel to west-northwest time-transgressive reconstructions which invoke successive ice oriented flutings that cross the southern flank of Muskeg Mountain, advance-and-retreat cycles, the absence of any deposits attribut- whereas the other valleys described in this paper parallel a large able to older proglacial or nonglacial conditions indicates that fluting tract that extends southwestwards from Lake Athabasca and tunnel valleys northeast Alberta were eroded and infilled during across the Firebag Plains and the adjacent flanks of Muskeg and a single glacial cycle. The lithofacies-based interpretations pre- Birch mountains (Bayrock, 1971; Bayrock and Reimchen, 1974). A sented in this paper support a composite model of tunnel valley similar relationship has been described in the North Sea basin, evolution, whereby valleys formed over time by steady-state pro- where tunnel valleys formed in ice marginal settings exhibit cesses, punctuated by time-transgressive, episodic discharges from a convergent pattern in response to a hydrostatic gradient directed subglacial reservoirs (Huuse and Lykke-Andersen, 2000; Praeg, towards the ice sheet margin (Praeg, 2003) compared with those 2003; Jørgensen and Sandersen, 2006; Lonergan et al., 2006). The formed a greater distance from the ice margin, which are smaller waning stages of each jökulhlaup were characterised by the and display more varied orientations, likely due to the influence of advection of till along ice-bed interface, likely concomitant with local hydrostatic pressures (Lonergan et al., 2006). localized modification by glacial ice. Although these events in northeast Alberta lack the chrono- Jørgensen and Sandersen (2006) proposed that the re-use of logical control needed to place them into a Quaternary stratigraphy, Danish tunnel valleys by repeated jökulhlaups resulted in the the spatially and temporally variable distribution of jökulhlaup evolution of anastomosing systems. We observe a similar network deposits interbedded with till, together with the absence of any of tunnel valleys in northeast Alberta, but note that they are incised deposits attributable to older proglacial or nonglacial conditions primarily within McMurray sandstone, compared with the straight indicates that these tunnel valleys were eroded and infilled during to slightly sinuous, isolated valleys, which are more common upon a single advance-and-retreat cycle associated with Late Wiscon- Clearwater shale. Therefore, we suggest that the association be- sinan glaciation. This cycle culminated with first appearance of tween valley morphology and substrate lithology reflects the proglacial deposits at the top of number valley fill sequences, which interplay between meltwater production, basal water pressure and comprise early Holocene glaciolacustrine sediments deposited the hydraulic conductivity of the substrate (c.f. Piotrowski, 1997), within Glacial Lake McMurray, an ice-dammed lake that sub- rather than solely a temporal evolution of the tunnel valley system. merged low-lying areas of northeast Alberta during the regional We propose that tunnel valleys in northeast Alberta valleys evolved retreat of the western Laurentide Ice Sheet (Fisher et al., 2009). as the subglacial hydrological system switched periodically from Collectively, these interpretations support the time- discrete to distributed modes, in response to cyclic variations in transgressive subglacial drainage and back-fill model developed basal meltwater production and storage, in part associated with the for tunnel valley evolution along the margins of the Fennoscandian differential permeability of the substrate. We suggest that in areas Ice Sheet (Huuse and Lykke-Andersen, 2000; Praeg, 2003; where the hydraulic transmissivity of the substrate was sufficient Jørgensen and Sandersen, 2006; Lonergan et al., 2006; Kristensen to drain basal meltwater by Darcian flow, tunnel valleys acted as et al., 2007). However, in our reconstruction, we propose that sinks of groundwater flow, evolving into a widespread, self- rather than spanning multiple glaciations, tunnel valleys in sustaining, integrated network. However, if Darcian flow was northeast Alberta evolved due to a combination of steady-state insufficient to discharge basal meltwater, either due to an increase subglacial drainage processes, punctuated by time-transgressive, in meltwater production, or a decrease in hydraulic conductivity of episodic jökulhlaups during a single cycle of Laurentide glaciation the substrate, subglacial reservoirs may have accumulated at the (c.f. Cutler et al., 2002; Hooke et al., 2006). No evidence has been ice-bed interface. Once the drainage of reservoirs was triggered, the found in this study that supports near-synchronous erosion of pre-established valley aquifers formed low pressure drainage tunnel valleys by catastrophic bankfull discharges. Rather, the val- pathways, which focussed high velocity discharges during succes- ley fills described in this paper document jökulhlaups which were sive jökulhlaups (c.f. Piotrowski, 1997; Cutler et al., 2002). However, of low to moderate magnitude and/or high velocity, which at times if the jökulhlaup exceeded the capacity of the pre-established reused existing valleys, while at others, eroded new valleys (c.f. valley system, water pressure may have initiated local separations Lonergan et al., 2006). along the ice-bed interface, enhancing discharge along subglacial conduits, thereby forming new tunnel valleys which would evac- 6. Conclusions uate the excess meltwater (Walder and Fowler, 1994; Boulton et al., 1996; Ng, 2000). Evidence for regionally asynchronous drainages This paper documents the distribution, morphology and lith- through these tunnel valleys either suggests that basal meltwater ofacies architecture of buried tunnel valleys in northeast Alberta. was impounded in multiple basins which experienced episodic The evolution of comparable tunnel valleys in the North Sea basin emptying and re-filling, or that different valley systems were active and adjacent coast of Denmark evidently spanned multiple at different times. N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72 71

This paper emphasises that the genesis, morphology and dis- Clayton, L., Attig, J.W., Mickelson, D.M., 1999. Tunnel Channels Formed in Wisconsin tribution of buried tunnel valleys in northeast Alberta relate in part During the Last Glaciation. Geological Society of America (Special Paper) 337, pp. 69e82. to the time-transgressive linkage between substrate permeability Cutler, P.M., MacAyeal, D.R., Mickelson, D.M., Parizek, B.R., Colgan, P.M., 2000. and the subglacial hydraulic system. This work accords with A numerical investigation of ice-lobe e permafrost interaction around the a broad body of research that has demonstrated the close associa- southern Laurentide Ice Sheet. Journal of Glaciology 46, 311e325. Cutler, P.M., Colgan, P.M., Mickelson, D.M., 2002. Sedimentologic evidence for out- tion between ice sheet dynamics and basal meltwater production burst floods from the Laurentide Ice Sheet margin in Wisconsin, USA: impli- and associated water pressure variations at the ice-bed interface cations for tunnel-channel formation. Quaternary International 90, 23e40. (Alley, 1989; Boulton et al., 1995, 1996, 2007a,b; Piotrowski, 1997). Denton, G.H., Sugden, D.E., 2005. Meltwater features that suggest Miocene ice-sheet overriding of the Transantarctic Mountains in Victoria Land, Antarctica. Geo- Moreover, we suggest that the conclusions reached in this paper grafiska Annaler 87A (1), 67e85. need to be placed in the broader context of subglacial landsystem Ehlers, J., Linke, G., 1989. The origin of deep buried channels of Elsterian age in evolution elsewhere in the province, since linkages between the Northwest Germany. Journal of Quaternary Science 4, 255e265. Evans, D.J.A., Campbell, I.A., 1992. Glacial and postglacial stratigraphy of Dinosaur subglacial hydrology of the western Laurentide Ice Sheet and the Provincial Park and surrounding plains, southern Alberta, Canada. Quaternary hydrodynamics of the WCSB may have been more significant than Science Reviews 11, 535e555. previously recognized (c.f. Marshall et al., 1996; Hildes et al., 2004; Eyles, N., 2006. The role of meltwater in glacial processes. Sedimentology 190, e Grasby and Chen, 2005). 257 268. Eyles, N., McCabe, A.M., 1989. Glaciomarine facies within subglacial tunnel valleys: the sedimentary record of glacioisostatic downwarping on the Irish Sea Basin. Sedimentology 36, 431e448. Acknowledgements Eyles, N., Eyles, C.H., Miall, A.D., 1983. Lithofacies types and vertical profile models: an alternative approach to the description and environmental interpretation of e We are grateful for the internal reviews of Fran Hein and Matt glacial diamict and diamictite sequences. Sedimentology 30, 395 410. Fisher, T.G., Jol, H.M., Boudreau, A.M., 2005. Saginaw Lobe tunnel channels (Lau- Grobe (ERCB/AGS). The formal reviews of Colm Ó Cofaigh (Univer- rentide Ice Sheet) and their significance in south-central Michigan, USA. Qua- sity of Durham) and an anonymous reviewer helped clarify the ternary Science Reviews 24, 2375e2391. manuscript, and are much appreciated. Fisher, T.G., Waterson, N., Lowell, T.V., Hajdas, I., 2009. Deglaciation ages and meltwater routing in the Fort McMurray region, northeastern Alberta and north- western Saskatchewan, Canada. Quaternary Science Reviews 28, 1608e1624. Gabriel, G., Kirsch, R., Siemon, B., Wiederhold, H., 2003. Geophysical investigation of References buried Pleistocene subglacial valleys in northern Germany. Journal of Applied Geophysics 53, 159e180. Alberta Energy and Utilities Board, 2007. ST98-2007: Alberta’s Energy Reserves Gibling, M.R., 2006. Width and thickness of fluvial channel bodies and valley fills in 2006 and Supply/Demand Outlook 2007e2016. the geological record: a literature compilation and classification. Journal of Allen, J.R.L., 1983. Studies in fluviatile sediments: bars, bar-complexes and sand- Sedimentary Research 76, 731e770. stone sheets (low sinuosity braided streams) in the Brownstones (L. Devonian), Glasser, N.F., Etienne, J.L., Hambrey, M.J., Davis, J.R., Waters, R.A., Wilby, P.R., 2004. Welsh Borders. Sedimentary Geology 33, 237e293. Glacial meltwater erosion and sedimentation as evidence of multiple glacia- Alley, R.B., 1989. Water-pressure coupling of sliding and bed deformation: I. Water tions in west Wales. Boreas 33, 224e237. system. Journal of Glaciology 35, 108e118. Grasby, S.E., Chen, Z., 2005. Subglacial recharge into the Western Canada Sedi- Andriashek, L.D., 2000. Quaternary Stratigraphy of the Buried Birch and Willow mentary Basin e impact of Pleistocene glaciation of basin hydrodynamics. Bedrock Channels, NE Alberta. Earth Sciences Report 2000-15, Alberta Geo- Geological Society of America Bulletin 117, 500e514. logical Survey. Alberta Energy Utilities Board, 61 pp. Hackbarth, D.A., Natasa, N., 1979. The hydrogeology of the Athabasca oil sands area, Andriashek, L.D., 2003. Quaternary Geological Setting of the Athabasca Oil Sands (in Alberta. Alberta Research Council Bulletin 38, 39 p. Situ) Area, Northeast Alberta. Earth Sciences Report 2002-03, Alberta Geo- Hampson, G.J., Davies, S.J., Elliot, T., Flint, S.S., Stollhofen, H., 1999. Incised valley fill logical Survey. Alberta Energy Utilities Board, 286 pp. sandstone bodies in Upper Carboniferous fluvio-deltaic strata: recognition and Andriashek, L.D., Atkinson, N., 2007. Buried Channels and Glacial-drift Aquifers in reservoir characterization of Southern North Sea analogues. In: Fleet, A.J., the Fort McMurray Region, Northeast Alberta. Earth Sciences Report 2007-01, Boldy, S.A.R. (Eds.), Petroleum Geology of Northwest Europe. The Geological Alberta Geological Survey. Alberta Energy Utilities Board, 160 pp. Society, Proceedings of the 5th Conference, vol. 2. Andriashek, L.D., Meeks, J., 2000. Bedrock Topography, Drift Thickness and Buried Harms, J.C., Southard, J.B., Walker, R.G., 1982. Structures and Sequences in Clastic Channels, Northeast Alberta. Geonote 2000-01, Alberta Geological Survey. Rocks. Society of Economic Paleontologists and Mineralogists, Short Course Alberta Energy Utilities Board, 5 pp. Notes 9, 111 pp. Bayrock, L.A., 1971. Surficial Geology of Bitumont (NTS 74E). Energy Resources Hein, F.J., 1982. Depositional mechanisms of deep-sea coarse clastic sediments, Conservation Board. ERCB/AGS Map 140, Scale 1:250 000. Cap Enragée Formation, Quebec. Canadian Journal of Earth Sciences 19, Bayrock, L.A., Reimchen, T.H.F., 1974. Surficial Geology of Waterways (NTS 74D). 267e287. Energy Resources Conservation Board. ERCB/AGS Map 148, Scale 1:250 000. Hein, F.J., 2006. Subsurface geology and facies characterization of the Athabasca Bell, R.E., 2008. The role of subglacial meltwater in ice-sheet mass balance. Nature Wabiskaw-McMurray succession in the Lewis and Firebag-Sunrise areas, Geoscience 1, 297e303. northeastern Alberta. In: Gilboy, C.F., Whittaker, S.G. (Eds.), Saskatchewan and Benn, D.I., Evans, D.J.A., 1996. The recognition and interpretation of subglacially- Northern Plains Oil and Gas Symposium. Saskatchewan Geological Society deformed materials. Quaternary Science Reviews 15, 23e52. Special Publication 19, pp. 238e248. Benn, D.I., Evans, D.J.A., 1998. Glaciers and Glaciation. Arnold, London. Heinz, J., Kleineidam, S., Teutsch, G., Aigner, T., 2003. Heterogeneity patterns of Boulton, G.S., Hindmarsh, R.C.A., 1987. Sediment deformation beneath glaciers: Quaternary glaciofluvial gravel bodies (SW-Germany): application to hydroge- rheology and geological consequences. Journal of Geophysical Research 92, ology. Sedimentary Geology 158, 1e23. 9059e9082. Hermanowski, P., 2010. Palaeoglaciology of the Weichselian Odra ice lobe, NE Boulton, G.S., Caban, P.E., Van Gijssel, K., 1995. Groundwater flow beneath ice Germany and NW Poland. Landform Analysis 14, 12e24. sheets: part I e large scale patterns. Quaternary Science Reviews 14, 545e562. Hildes, D.H.D., Clarke, G.K.C., Flowers, G.E., Marshall, S.J., 2004. Subglacial erosion Boulton, G.S., Caban, P.E., Van Gijssel, K., Leijinse, A., Punkari, M., van Weert, F.H.A., and englacial sediment transport modelled for North American ice sheets. 1996. The impact of glaciation on the groundwater regime of Northwest Quaternary Science Reviews 23, 409e430. Europe. Global and Planetary Change 12, 397e413. Hooke, R., Le, B., Jennings, C.E., 2006. On the formation of tunnel valleys of the Boulton, G.S., Lunn, R., Vidstrand, P., Zatsepin, S., 2007a. Subglacial drainage by southern Laurentide Ice Sheet. Quaternary Science Reviews 25, 1364e1372. groundwater-channel coupling, and the origin of esker systems: part I e gla- Horne, E., Seve, G., 1991. Pleistocene ‘Buried Valley’ outwash channels e East Bank, ciological observations. Quaternary Science Reviews 26, 1067e1090. Athabasca River in the Fort McMurray Area. Canadian Institute of Mining Fifth Boulton, G.S., Lunn, R., Vidstrand, P., Zatsepin, S., 2007b. Subglacial drainage by District Five Meeting. Fort McMurray 76, 77-1 to 77-15. groundwater-channel coupling, and the origin of esker systems: part II e theory Huuse, M., Lykke-Andersen, H., 2000. Overdeepended Quaternary valleys in the and simulation of a modern system. Quaternary Science Reviews 26, 1091e1105. eastern Danish North Sea: morphology and origin. Quaternary Science Reviews Boyce, J.I., Eyles, N., 2000. Architectural element analysis applied to glacial deposits: 25, 1339e1363. internal geometry of a late Pleistocene till sheet, Ontario, Canada. Geological Janszen, A., Spaak, M., Moscariello, A., 2012. Effects of the substratum on the for- Society of America Bulletin 112, 98e118. mation of glacial tunnel valleys: an example from the Middle Pleistocene of the Brennard, T.A., Shaw, J., 1994. Tunnel channels and associated landforms, south- southern North Sea Basin. Boreas 41, 629e643. central Ontario: their implication for ice-sheet hydrology. Canadian Journal of Jørgensen, F., Sandersen, P.B.E., 2006. Buried and open tunnel valleys in Denmark e Earth Sciences 31, 505e522. erosion beneath multiple ice sheets. Quaternary Science Reviews 25,1339e1363. Carrigy, M.A., Kramers, J.W., 1975. Guide to the Athabasca Oil Sands Area. In: Alberta Jørgensen, F., Sandersen, P.B.E., Auken, E., 2003. Imaging buried Quaternary valleys Research Council Information Series, vol. 65. Alberta Geological Survey, Alberta using the transient electromagnetic method. Journal of Allied Geophysics 53 Energy Utilities Board. (4), 199e213. 72 N. Atkinson et al. / Quaternary Science Reviews 65 (2013) 53e72

Kehew, A.E., Piotrowski, J.A., Jørgensen, F., 2012. Tunnel valleys: concepts and Ó Cofaigh, C., 1996. Tunnel valley genesis. Progress in Physical Geography 20, 1e19. controversies e a review. Earth-Science Reviews 113, 33e58. Pettapiece, W.W., 1986. Physiographic Subdivision of Alberta. Agriculture Canada, Kozlowski, A.L., Kehew, A.E., Bird, B.C., 2005. Outburst flood origin of central Kala- Scale 1:500 000. mazoo River Valley. Michigan, USA. Quaternary Science Reviews 24, 2354e2374. Piotrowski, J.A., 1994. Tunnel-valley formation in Northwest Germany e geology, Kramers, J.W., Prost, A.W., 1986. Oil Sands Resources of the Grand Rapids Formation mechanisms of formation and subglacial bed conditions for the Bornhoved in the Wabasca Deposit. Report to Alberta Oil Sands Technology Research Au- tunnel valley. Sedimentary Geology 89, 107e141. thority under ARC/AOSTRA Oil Sands Geology Agreement 1588. Piotrowski, J.A., 1997. Subglacial hydrology in North-Western Germany during the Kristensen, T.B., Huuse, M., Piotrowski, J.A., Claausen, O.R., 2007. A morphometric last glaciation: groundwater flow, tunnel valleys and hydrological cycles. Qua- analysis of tunnel valleys in the eastern North Sea based on 3D seismic data. ternary Science Reviews 16, 169e185. Journal of Quaternary Science 22, 801e815. Piotrowski, J.A., Geletneky, J., Vater, R., 1999. Soft-bedded subglacial meltwater Kristensen, T.B., Piotrowski, J.A., Huuse, M., Clausen, O.R., Hamberg, L., 2008. Time- channel from the Welzow-Süd open-cast lignite mine, Lower Lusatia, eastern transgressive tunnel valley formation indicated by infill sediment structure, Germany. Boreas 28, 363e374. North Sea e the role of glaciohydraulic supercooling. Earth Surface Processes Praeg, D., 2003. Seismic imaging of Mid-Pleistocene tunnel-valleys in the North Sea and Landforms 33, 546e553. Basin e high resolution from low frequencies. Journal of Applied Geophysics 53, Langenberg, C.W., Hein, F.J., Lawton, D., Cunningham, J., 2002. Seismic modeling of 273e298. fluvial-estuarine deposits in the Athabasca oil sands using ray-tracing tech- Röthlisberger, H., 1972. Water pressure in intra- and subglacial channels. Journal of niques, Steepbank River area, northeastern Alberta. Bulletin of Canadian Pe- Glaciology 11, 177e203. troleum Geology 50, 178e204. Russell, H.A.J., Arnott, R.W.C., Sharpe, D.R., 2003. Evidence for rapid sedimentation Lonergan, L., Maidment, S.C.R., Collier, J.S., 2006. Pleistocene subglacial tunnel in a tunnel channel, Oak Ridges Moraine, southern Ontario, Canada. Sedi- valleys in the central North Sea basin: 3-D morphology and evolution. Journal of mentary Geology 160, 33e55. Quaternary Science 21, 891e903. Sandersen, P.B.E., Jørgensen, F., 2003. Buried Quaternary valleys in western Den- MacCormack, K.E., Maclachlan, J.C., Eyles, C.H., 2005. Viewing the subsurface in mark e occurrence and inferred implications for groundwater resources and three dimensions: initial results of modeling the Quaternary sedimentary infill vulnerability. Journal of Applied Geophysics 53, 229e248. of the Dundas Valley, Hamilton, Ontario. Geosphere 1, 23e31. Sandersen, P.B.E., Jørgensen, F., 2012. Substratum Control on Tunnel-valley Forma- Marshall, S.J., Clarke, G.K.C., Dyke, A.S., Fisher, D.A., 1996. Geologic and topographic tion in Denmark. In: Geological Society of London, Special Publication, vol. 368. controls on fast flow in the Laurentide and Cordilleran Ice Sheets. Journal of Sharpe, D., Pugin, A., Pullan, S., Shaw, J., 2004. Regional unconformities and the Geophysical Research 101, 17827e17839. sedimentary architecture of the Oak Ridges Moraine area, southern Ontario. McPherson, R.A., Kathol, C.P., 1977. Surficial Geology of Potential Mining Areas in the Canadian Journal of Earth Sciences 41, 183e198. Athabasca Oil Sands Region. Quaternary Geosciences Ltd, Report to Alberta Sugden, D.E., Denton, G.H., Marchant, D.R., 1991. Subglacial meltwater channel Research Council, 177 pp. systems and ice sheet overriding, Asgard Range, Antarctica. Geografiska Mehnert, E., Hackley, K.C., Larson, T.H., Panno, S.V., Pugin, A., Wehrmann, H.A., Annaler 73A (1), 109e121. Holm, T.R., Roadcap, G.S., Wilson, S.D., 2004. The Mahomet Aquifer: Recent Van der Vegt, P., Janszen, A., Moscariello, A., 2012. Tunnel valleys: current knowl- Advances in Our Knowledge. In: Illinois State Geological Survey Open File, Se- edge and perspectives. In: Huuse, M., Redfern, J., Dixon, R.J., Craig, J., Moscar- ries 2004-16. iello, Le Heron, D.P. (Eds.), Glaciogenic Reservoirs and Hydrocarbon Systems. Miall, A.D., 1985. Architectural-element analysis: a new method of facies analysis Geological Society of London Special Publication, vol. 368. applied to fluvial deposits. Earth Science Reviews 22, 261e308. Van Dijk, M., Postma, G., Kleinhans, M.G., 2009. Autocyclic behaviour of fan deltas: Moores, H.D., 1989. On the formation of the tunnel valleys of the Superior Lobe, an analogue experimental study. Sedimentology 56, 1569e1589. central Minnesota. Quaternary Research 32, 24e35. Van Dijke, J.J., Veldkamp, A., 1996. Climate-controlled glacial erosion in the un- Moran, S.R., Clayton, L., Hooke, R.Le B., Fenton, M.M., Andriashek, L.D., 1980. Glacier- consolidated sediments of northwest Europe, based on a genetic model for bed landforms of the prairie region of North America. Journal of Glaciology 25, tunnel valley formation. Earth Surface Processes and Landforms 21, 327e340. 457e476. Visser, J.N.J., 1988. A Permo-Carboniferous tunnel valley system east of Barkly West, Mossop, G.D., Shetsen, I., 1994. Geological Atlas of the Western Canada Sedimentary northern Cape Province. South African Journal of Geology 91, 350e357. Basin. Canadian Society of Petroleum Geologists and Alberta Research Council, Walder, J.S., Fowler, A., 1994. Channelized subglacial drainage over a deformable Calgary, Alberta. bed. Journal of Glaciology 40, 3e15. Ng, F.S.L., 2000. Canals under sediment-based ice sheets. Annals of Glaciology 30, Whipple, K.X., Parker, G., Paola, C., Mohrig, D., 1998. Channel dynamics, sediment 146e152. transport, and the slope of alluvial fans: experimental study. Journal of Geology Niewiarowski, W., 1995. Diagnostic features of subglacial channels of glacial and 106, 677e693. glaciofluvial origin, exampled by channels of the Che1mno-Dobrzyn and the Wingfield, R., 1990. The origin of major incisions within the Pleistocene deposits of eastern Gniezno Lakelands. Questiones Geographicae 4, 225e231. the North Sea. Marine Geology 91, 31e52.