SEDIMENTARY EVIDENCE FAVOURING THE FORMATION OF ROGEN LANDSCAPES BY OUTBURST FLOODS

John Shaw
Department of Earth and Atmospheric Sciences
University of Alberta
Edmonton, Alberta, Canada, T6G 2E3
E-mail: John.Shaw@ualberta.ca

ABSTRACT

Rogen landscapes are defined by the morphology of the ridges and troughs and their locations in areas of former glaciation. They are transitional with hummocky and drumlin landscapes in associations of subglacial bedforms. This paper concentrates on the sediments within Rogen ridges and approaches their interpretation by way of the fundamental mechanics of high Reynolds Number flow beneath a wavy ice bed. The morphology, sedimentary characteristics and landscape associations of Rogen terrain are interpreted to record outburst flood from beneath the mid-latitude Pleistocene ice sheets. Analogous sediments to those in Rogen ridges are found around violently eruptive volcanoes and in areas inundated by outburst floods in modern environments. These sediments are deposited from highly turbulent, hyperconcentrated flows. Flood volumes of the order 106 km3 have implications for ice-sheet/climate/ocean interactions that are not recognized in present models. These implications are presented as speculations assuming the reality of outburst floods of huge magnitudes and short duration.

INTRODUCTION

Rogen "moraines", like drumlins, are familiar to the point where they are easily and confidently recognized - see Hättestrand (1997) for an excellent review of the characteristics and the history of ideas on Rogen moraines. Such familiarity commonly leads to a loose definition of Rogen landscapes. The resulting problems are compounded when definitions imply genesis.

A simple definition of the Rogen landscape is adopted here, using the term landscape to include geomorphology and sedimentology. Thus, the Rogen landscape is defined as: Fields of ridges and troughs arranged transverse to flow (Figs. 1 and 2). Some ridges may be fluted or drumlinized, and some troughs may contain large, circular depressions. Ridges may contain beds truncated at the land surface or beds that are conformable with the ridge surface. It is common to find angular boulders of the underlying bedrock in the troughs and on ridge surfaces. The Rogen landscape is found in areas of former glaciation.

Qualifications regarding distance from ice divides or the thermal regime of the ice during formation of the Rogen fields would introduce inference and unduly complicate the definition. The Rogen landscape has analogs in deserts, on the seabed of continental shelves, and in ripples eroded upwards into river ice (Fig. 1).

Since the surface rocks and unlithified sediments of Rogen landscapes are highly variable (see Hättestrand 1997, his Table 1), a universal depositional mechanism for Rogen formation is ruled out. From theory and observation Rogen ridges may form in different ways: (i) by erosion of the adjacent troughs (e.g., ridges with truncated beds (Munro and Shaw 1997) and giant ripples cut in bedrock (Cox 1998)); (ii) by deposition (e.g., ridges containing sedimentary beds conformable with the surface morphology (e.g., Fisher and Shaw 1992)); or (iii) by a combination of erosion and deposition (Fig. 1). Rogen ridges composed of angular blocks of locally derived bedrock in a finer grained matrix are the main topic of this paper (Fig. 3).

Many Rogen ridges contain folds and thrust planes that are generally assumed to have resulted from direct glacial action (Shaw 1979; Hättestrand 1997). Since bed shear stresses and lift (Gerhardt and Gross 1972) generated by high velocity fluids may also produce large folds and thrusts, it is best not to assume a glacial origin for these tectonic features.

SUBGLACIAL MELTWATER PROCESSES: FUNDAMENTALS

Yalin (1972) illustrated waveforms in fluids and explained bed formation in terms of varying rates of deposition and/or erosion with distance (Fig. 4). Similarly, Allen (1982) described erosional and depositional bedforms caused by preferential erosion around obstacles and/or deposition or reduced erosion to the lee of the obstacles (Fig. 5). The principles illustrated in Figures 4 and 5 are fundamental to the understanding of Rogen landscapes - the creation of bedforms is simply the result of rates of change of bed height over distance (Fig. 4). Such bedforms may result from the action of different agents (e.g., wind, flowing water) and different processes (e.g., abrasion by tools in flowing ice or sand blasting by saltating grains).

The meltwater hypothesis presented here for Rogen terrain assumes there were waves, similar to those beneath river ice (Fig. 1), carved into the underside of former ice sheets by broad meltwater flows (Fig. 6). Flow depicted in Figure 6 is a generalization of extremely complex, turbulent and turbid flows. As well, the locations of erosion and deposition are derived from this generalization which, by considering only time-averaged characteristics, is deceptively simple. It is also qualitative - a complete physical and mathematical analysis of such flow requires an understanding of: (i) the non-linear flow of ice under high pressure gradients; (ii) non-uniform, unsteady, and highly turbulent, meltwater flow with a hyperconcentrated suspension load; and (iii) non-uniform rates of erosion, sediment transport, and deposition which are non-linear functions of bed shear stress. While such problems cannot be solved at present, we can safely assume that constrictions are zones of erosion and expansions are zones of deposition. Furthermore, the Bernoulli principle dictates that zones with high fluid velocities experience low pressures (Fig. 6). Application of the Bernoulli equation, using previous estimates for discharge per unit width, flow depth and overburden pressure (Shaw 1996), gives lift capable of plucking slabs of bedrock 1.5 m thick.

Local low pressures associated with vertical vortices or kolks (Matthes 1947) would generate much larger lift forces, capable of plucking thicker slabs and creating folds and diapirs in soft beds. Bed shear stress exerted by flowing water, which increases with v2 (where v is velocity) (Gerhardt and Gross 1972), may then stretch diapirs in flame structures several tens of metres long.

SEDIMENTARY ROGEN RIDGES

Description

Greater erosion at troughs than at ridges is implicit where underlying beds are truncated by extensive, landscape unconformities. Troughs commonly contain large boulders or exposed bedrock, and ridges are composed of remnant materials (till, fluvial sediments, organic deposits or bedrock) and large boulders transported from the trough immediately upflow are found perched on ridge surfaces (Fig. 3). The perched boulders are very angular with sharp edges.

Tectonic features, thrust and normal faults, folds and overfolded diapirs are common in the remnant materials of erosional ridges (Shaw 1979); Munro and Shaw 1997). By contrast, many Rogen ridges with conformable architecture (internal structure corresponding to external form) contain angular blocks of local bedrock (Fig. 7a), graded sand with minor deformation around large clasts (Fig. 7b) and graded beds in granules, sand or silt (Fig. 7c). Angular slabs of bedrock, which may be as much as one metre thick and five metres long, are usually imbricate with upstream dip (Fig. 7d) (Shaw 1979).

Interpretation

Ridges with truncated strata of earlier geological formations are evidently remnants around which depressions were eroded into the surrounding bed. Those with conformable architecture and sorted sediments are more likely depositional. Angular blocks in a fine grained matrix suggest erosion by hydraulic plucking, transportation in hyperconcentrated flow and rapid deposition from suspension (Maizels 1989). High flow velocities are required to generate the lift forces for hydraulic plucking (Matthes 1974) and high shear stresses generated by fluid flow combined with high lifting forces best explain the diapirism and flame structures. Fast deposition is expected at expansions where flow decelerates quickly (Fisher and Shaw 1992) (Fig. 6). Thus, bedrock eroded from an up-flow trough is deposited in the ridge immediately down flow (Fig. 6).

Angular bedrock blocks arranged like currents in a cake in a finer, polymodal matrix are seen in other depositional environments: (i) overlying landscape unconformities (erosional surfaces) and along probable outburst flood paths (Fig. 8); (ii) around eruptive volcanoes (Fig. 9), and (iii) along well documented outburst flood paths (Fig. 10). Violently turbulent, hyperconcentrated flows characterize these environments and, although the details of transport and depositional processes are beyond our understanding, the resulting sediments tell of rock slabs torn from the bed, enormous concentration of suspended sediment, and extremely fast deposition. As well, the tectonic structures are explained in this hypothesis by the enormous lift and drag forces associated with flows of high Reynolds Number.

CONCLUSIONS

It is clear that Rogen landscapes result from coherent flow patterns at the regional scale (100's km, Fig. 2). Ridges record flows that were confined to valleys or diverged over plateaus and plains (Fig. 2). Since such ridges are observed at great distances from ice divides (Fig. 1c), their distribution is not, as is commonly believed (see review by Hättestrand 1997), restricted to the inner zones of former ice sheets. On the other hand, Rogen landscapes are best developed near proposed divides of the Laurentide and Fennoscandian ice sheets. Scoured bedrock, extending down flow from the Rogen landscapes, implies erosion, perhaps corresponding to flow acceleration where isostatic effects caused the land surface to rise towards the ice bed and reduced the flow cross-sectional area.

Subglacial landscapes indicate a continuum between fluting, drumlins, Rogen, and hummocky terrain (Shaw 1996; Munro and Shaw 1997) and these subglacial landforms were probably formed by regional-scale flows. The bedding and architecture of these ridges and the postulated flow conditions presented here strongly support a meltwater origin for Rogen landscapes.

Large, angular blocks of local bedrock and fine-grained suspension deposits are explained well by hydraulic plucking at flow constrictions and rapid sedimentation in areas of flow expansion (Fig. 6). Thus, meltwater processes best explain field observations of form, pattern and internal composition. Other mechanisms such as a combination of polythermal beds and flow extension where warm-based ice passes upflow into cold-based ice (Hättestrand 1997) do not account for the full range of Rogen landscape characteristics. A reconstruction of subglacial bed formation by outburst floods, including Rogen landscapes forming beneath waves eroded into the overlying ice portrays a vivid picture of the imagined scene (Fig. 11). Figure 11 also illustrates the surface material in various areas of the reconstruction.

IMPLICATIONS

  1. If drumlin, Rogen and hummocky landscapes record floods on the scale of the former ice sheets, then the ice sheets themselves must have been partly floating on water sheets.
  2. Volumes of water required to sustain such floods would have been on the order 106 km3 equivalent to a rise of several metres in sea level over a matter of weeks (Shaw 1996).
  3. Hyperconcentrated, freshwater turbidity currents of this magnitude would have extended beneath much of the North and South Atlantic Oceans. Deposition would have resulted in cold, freshwater plumes rising through the ocean water column causing abrupt changes in sea surface temperature and salinity. Such changes, had they occurred, hold significant implications for climate modelling (cf. Peng et al. 1995).
  4. Rapid sea level rise is also expected to have destabilized ice sheets extending over continental shelves.


As far as I know, there has been no attempt to model events of the magnitude of the outburst floods postulated by the meltwater hypothesis for subglacial landscapes. Perhaps, these speculations will prompt some brave numerical analyst to give it a try.

ACKNOWLEDGEMENTS

Mandy Munro-Stasiuk, Alanna Vernon, Michael Fisher, and Randy Pakan provided much needed help with the final preparation of this paper. Many people, particularly Bruce Rains, have supported me when I needed it most. I would like to thank NSERC Canada for financial support.

REFERENCES

Allen, J.R.L. 1982. Sedimentary Structures. Elsevier, Amsterdam, 663 pp.

Branney, M.C. and Kokelaar P. 1997. Giant bed from sustained catastrophic density current flowing over topography: Acatlan ignimbrite, Mexico. Geology, 25: 115-118.

Cox, D.E. 1998. Transverse drift ridges - giant current ripples? www.sentex.net/~tcc/moraine.html

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

Gerhardt, P.M. and Gross, R.J. 1972. Fundamentals of Fluid Mechanics. Addison Wesley, Reading, 856 pp.

Hättestrand C. 1997. Ribbed moraine in Sweden B distribution pattern and paleoglaciological implications. Sedimentary Geology, 111, 41-56.

Maizels, J. 1989. Sedimentology, paleoflow dynamics and flood history of jökulhlaup deposits: paleohydrology of Holocene sediment sequences in southern Iceland sandur deposits. Journal of Sedimentary Petrology, 59, 204-223.

Matthes, G.H. 1947. Macroturbulence in natural stream flow. EOS American Geophysical Union Transactions, 28, 255-262.

Munro, M. and Shaw, J. 1997. Erosional origin of hummocky terrain in south- central Alberta, Canada. Geology, 45(11), 1027-1030.

Peng, S.L. , Mysak, L.A., Ritchie H., Derome, J., and Dugas, B. 1995. The difference between early and mid-winter atmospheric responses to sea surface temperature anomalies in the northwest Atlantic. Journal of Climate, 8(2), 137-157.

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

Shaw, J. 1994. A qualitative view of sub-ice-sheet landscape evolution. Progress in Physical Geography, 18, 159-184.

Yalin, M.S. 1972. Mechanics of Sediment Transport, Pergamon, Oxford, 290 pp.

List of Figures

Figure 1.
(a) Ripples on the underside of river ice exposed by surface melt(courtesy of G. Ashton).
(b) Rogen landscape, Boyd Lake, N.W.T., Canada

(c) Hummocky terrain including Rogen landscapes, Stettler, Alberta, Canada. Note that the original air photograph is enhanced by infilling depressions with black ink, as though they are filled by water.

Figure 2.
Rogen landscape, Avalon Peninsula, Newfoundland. The map represents all ridges observed on 1:50 000 scale air photographs. The coherent, regional flow pattern (inset) was constructed by assuming that flowlines were orthogonal to ridge crests (revised from Fisher and Shaw 1992).
Figure 3.
Sediment exposed in Rogen ridge, Anåkröken, central Sweden distal is to the left in the photographs.

(a) Panoramic view of the cross-section (small car for scale), showing major erosional surfaces, boulder lag, and forest bedding.

(b) Lower part of the exposure. Crudely bedded, angular boulders derived from the local shale bedrock in a multimodal matrix (gravel to silt). The boulders are imbricate with major planes dipping proximally (Shaw 1979).

(c) Upper part of the exposure; the sediment is finer and better stratified than that below. Sand and silt in graded and cross laminae are common in this part of the section (see (d)).

(d) Interbedded poorly sorted gravel to silt with angular clasts and graded, fine sand and silt.

Figure 4.
Bedforms originating from imposed surface waves (after Yalin 1972). t is time, x is distance from the start of the erodible bed, qs is sediment transport rate, dqs/dx is rate of change of qs with respect to distance, l is water surface wavelength, L is bed wave length, and D is bed amplitude. Note the direct relationship between zones of erosion and deposition and dqs/dx.
Figure 5.
Bedforms resulting from flow around bluff obstacles (after Allen 1982). This figure provides a framework for testing hypotheses on bedform genesis related to obstacles.
Figure 6.
Flow in ice and water related to waveforms eroded in the ice roof. The Bernoulli principle dictates the locations of high and low pressure and water velocities, which ultimately control rates of ice drawdown and of bed erosion, sediment transport, and deposition.
Figure 7.
Sediment in Rogen ridges.
(a) Poorly sorted boulders to silt with angular, locally derived clasts, Lillholmsjö, Jämtland, Sweden. Clean sand beneath boulder (arrowed).

(b) Graded, slightly deformed sand with deformation structures around a boulder, Metagamie, Quebec. Graded sand and silt laminae (arrowed).

(c) Parallel laminated silt and fine sand overlying poorly sorted gravel to silt with angular clasts. Note that some large clasts "float" in parallel bedded silt and fine sand (arrowed), Anåkröken, Jämtland, Sweden.

(d) Imbricate, angular boulders of local bedrock, Brigus Junction, Avalon Peninsula, Newfoundland. Scale bar 20 cm.

Figure 8.
(a) Pyroclastic flow (after Branney and Kokelaar, 1997).

(b) Pyroclastic base flow deposits. Poorly sorted, crude bedding and very angular clasts, New Mexico (courtesy of George Morris).

Figure 9.
(a) Poorly sorted gravel to silt with angular clasts of local bedrock. Possible flood deposits Aberystwyth, Wales.

(b) Poorly sorted boulders to silt with large angular blocks and slabs of local bedrock. The upper face of the large slab (right of figure) is striated and sculpted. Flow from right to left. These beds lie on a landscape unconformity defining the surface of drumlins. Pigeon Point, Clew Bay, Ireland.

Figure 10.
(a) Dry Falls, Washington Scablands, the rock face of the falls is about 120 m high.

(b) Flood deposits of the Washington Scablands. Poorly sorted, gravel to fine sand with angular clasts of locally derived basalt interbedded with horizontally stratified medium to fine sand. The conformable bedding indicates rapid sedimentation.

Figure 11.
Reconstructed flow conditions during the formation of subglacial bedforms by outburst floods. The accompanying photographs illustrate details of the landscape depicted in the reconstruction.

(a) Sediment in a depositional Rogen ridge, Anåkröken, Jämtland, Sweden.

(b) Erosional drumlin, the Ovens, Nova Scotia. Note: boulders similar in size to those on the modern beach must have been transported by the drumlin forming process.

(c) Erosional forms (s-forms) in crystalline bedrock, French River, Georgian Bay, Ontario.


© 1998 John Shaw

11 December 1998