A time-lapse movie technique has been developed to study processes taking place near the interface when ice under high normal stress slides over a rough slab for shear stresses and sliding velocities comparable to those of real glaciers.

The “sledge rig” described by Budd and others (1979) was set up with one of the strips of ice (160 mm long and 23 mm wide) close to the edge of the pebble slab on which it rested. The normal stress was applied by loading the rig with lead to 5 bars. The shear stress was applied by a direct load on a cable over a pulley while the rig rested on the slab on a strong table. Shear stresses in the range 1 to 2 bars were used giving velocities from about 50 to 500 m a–1. Beyond 2 bars it was found that acceleration would set in, so this was avoided.

A time-lapse movie camera was set up beside the rig to focus first on the whole configuration and then on the ice and interface as a close-up. A flash light was set up behind the rig to shine through the ice strip. A scale rule was placed along the slab under the ice strip. A dial-gauge micrometer was also used to monitor displacement.

A time-lapse frequency of about 1 frame per 8 s was found to give a suitable speed. An automatic control triggered the camera and flash and also recorded the number of frames. A number of different markings were tried on the ice to indicate the relative ice deformation. Small graphite particle tracers were found the most successful.

A number of runs have been filmed showing the ice sliding speed increasing with increasing shear stress. The tests were carried out in a cold room at a temperature close to 0°C.

Copies of the film can be made available at cost.

The authors were greatly assisted by Mr David Gardner of the Department of Educational Technology of Melbourne University by the provision of the filming equipment and general support for the operation.

If sediments being transported by glaciers in subglacial, englacial, and supraglacial situations can be differentiated on the basis of their textural characteristics, some potential exists for interpreting the relative importance of subglacial and sub-aerial periglacial processes in the excavation of glacial valleys, by examining the sediments in abandoned or active terminal moraines. Caution should be used with compound glacial valleys where subglacially eroded debris may become englacial as ice streams merge, be subsequently transported passively, and in terminal moraine deposits appear to represent sub-aerially derived sediments.

This paper reviews the results of ten years study of the only four subglacial sites which are permanently accessible thanks to the work of hydro-electrical companies. All the sites occur beneath temperate ice. The first part is devoted to the study of the rock-ice interface as a glaciological phenomenon. The dynamic conditions for separation of the ice from the rock bed are considered. This cavitation phenomenon occurs when tan α > Vi/Hi. “Regressive cavitation” explains the existence up-stream of large permanent cavities and of a series of small cavities which, although they are not permanent, are fundamental because they control sub-glacial water drainage. The second part analyses the sliding movement of ice on a rock bed. Indeed the deformation of the cavities depends mainly on the variations in the velocity of the glacier. The sliding velocity measured at the interface accounts for 60 to 80% of the surface movement of the glacier. 80 to 90% of the surface velocity is attained a few metres above the glacier-bed interface. The third part describes the characteristics of subglacial drainage which are necessary to understand the nature of the ice-rock interface. The fourth part is devoted to the precise description of the different types of interface as they appeared in the subglacial sites.

This paper has been accepted for publication in full in a future issue of the Journal of Glaciology.

Glacial erosion wears the bedrock grain by grain, that is it breaks free a multitude of little fragments. It also acts upon the bedrock to cause trains of concentric fractures and quarried surfaces. Since this type of erosion is selective in places of variable resistance, it brings about dug-out forms like grooved joints, or salient forms like veins carved out from up-stream down, it produces small-scale features like grooves or chattermark trains or, on a larger scale, acts upon rocky knobs to produce roches moutonnées; the latter, in spite of their dimensions, are subject to an action that can still affect every point simultaneously. Our objective is thus to present hierarchically the glacial forms left on the bedrock, along with the terminology that designates them. In Table I we give both English and French terminology, but it is not our purpose to make definite proposals for the few new terms in English.

The glacier bed, where basal sliding occurs, was reached by cable-tool drilling and sand-pump bailing in seven bore holes in Blue Glacier, Olympic National Park, Washington. Basal sliding velocities measured by bore-hole photography and confirmed by inclinometry are unexpectedly low, ranging from 0.3 to 3.0 cm/day and averaging 1.0 cm/day. This is much less than about half the surface velocity of 15 cm/day, which was the sliding-rate expected from earlier deformation measurements in bore holes made by thermal drilling alone.

It is difficult to deduce sliding properties from the numerical modelling of ordinary glaciers because the flow law of ice is still not known well enough to clearly differentiate sliding from internal deformation of the ice. For glaciers undergoing high-speed surges it appears that the majority of the total speed is due to sliding. Furthermore the average basal shear stress of the ice mass is lowered during the surge. This suggests that surging glaciers can be modelled by incorporating a sliding friction law which has the effective friction coefficient decreasing for high velocities. A relation of this type has been found for ice sliding on granite at −0.5°C by Barnes and others (1971) and has also been obtained for rough slabs with ice at the pressure-melting point by Budd and others (1979).

A simple two-dimensional model was developed by Budd and McInnes (1974) and Budd (1975), which was found to exhibit the typical periodic surge-like characteristics of real ice masses. Since the sliding-stress relation for the low velocities and stresses was not known, and was not so important for the surges, it was decided to use the condition of gross equilibrium (i.e. that the ice mass as a whole does not accelerate) together with a single-parameter relation for the way in which the friction decreases with stress and velocity to prescribe the basal shear-stress distribution. The low-stress-velocity relation can thus be obtained as a result.

This two-dimensional model has now been parameterized to take account of the three-dimensional aspects of real ice masses. A number of ice masses have since been closely matched by the model including three well-known surging ice masses: Lednik Medvezhiy, Variegated Glacier, and Bruarjökull. Since the flow properties of ice are so poorly known—especially for longitudinal stress and strain-rates—the model has been run with two unknown parameters: one a flow-law parameter (η) and the other a sliding parameter (ø). The model is run over a wide range of these two parameters to see if a good match can be made to the real ice masses and if so what the values of the parameters η and ø are for best fit.

The matching of the three above ice masses gave very similar values for each of the two parameters η and ø, the value of η being within the range of values expected for the flow properties of temperate ice as determined by laboratory experiments. Using the same values of η and ø it is found that the ordinary glaciers modelled so far do not develop surging but that they could do if the value of ø were increased or if the mass-balance input were sufficiently increased.

For Lednik Medvezhiy a detailed analysis of the friction coefficient with velocity was carried out and it was found that the values required for best fit showed a very close agreement to the sliding friction curve of Barnes and others (1971) at −0.5°C. It is concluded that this type of sliding relation can account for the major features of glacier surge phenomena.

Finally it is apparent that the numerical modelling technique can be used very effectively to test any large-scale bulk sliding relation by the analysis of real surges of ice masses and in addition can provide further insight into the sliding relation in association with other stresses in the ice mass.

In the heat-pump effect developed by Robin (1976), pressure-melting within the basal ice mass, as distinct from processes at the ice–bedrock interface, can be responsible for the formation of excess water in zones of high-pressure ice up-stream of obstacles. This water is squeezed out of the ice by the pressure and can refreeze at the interface to form regelation ice down-stream.

The peculiar chemical characteristics of regelation ice at the base of Glacier de Tsijiore Nouve in the Swiss Alps can only be explained by refreezing of this squeezed water. Melting at the interface up-stream of bed protuberances and refreezing of these melt waters downstream, as implied in weertman’s theory, can be excluded in this case.

This paper is published in full in Nature, Vol. 273, No. 5662, 1978, p. 454–56,

It is well known that regelation may occur by pressure-melting in front of a wire and refreezing at the rear. The velocity of the wire has been observed to have values ranging from 10–5 to 10–1 mm/s. However, there have always been large discrepancies between experiments and any theory based on this mechanism, and, when moving at a comparable velocity, hard balls slid on an ice surface leave grooves made by plastic deformation. So, we conducted experiments to test whether regelation phenomena might be explained by plastic deformation of ice around the wire.

Two major types of terrain that formed at or near the bed of Pleistocene continental ice sheets are widespread throughout the prairie region of Canada and the United States. These are (1) glacial-thrust blocks and source depressions and (2) streamlined terrain.

Glacial-thrust terrain formed where the glacier was frozen to the substrate and where elevated pore-water pressure decreased the shear strength of the substrate to a value less than that applied by the glacier. The marginal zone of ice sheets consisted of a frozen-bed zone, no more than 2 to 3 km wide in places, within which glacial-thrust blocks are large and angular. Up-glacier from this zone the thrust blocks are generally smaller and smoothed. Streamlined terrain begins 2 to 3 km behind known ice-margin positions and extends tens of kilometres up-glacier. Streamlined terrain formed in two ways: (1) erosion of the substrate as a consequence of basal sliding in the sub-marginal thawed-bed zone and (2) erosional smoothing accompanied by emplacement of till in the lee of thrust blocks where they were deposited and subsequently exposed to thawed-bed conditions as a result of further advance of the glacier.

This paper has been accepted for publication in full in a future issue of the Journal of Glaciology.

The shape of the former Laurentide ice sheet in what is now northern Ohio and southern Ontario is calculated and the subglacial hydrological potential field computed. Subglacial water flow is found to be concentrated over high points in the bed, such as what is now Kelleys Island in Lake Erie. Some 1010 litres of subglacial melt water per year passed over Kelleys Island.

It is argued that the “glacial” grooves on Kelleys Island are subglacial melt-water channels. Assuming that the subglacial melt water attained saturation with respect to CaCO3, in passing over the limestone of Kelleys Island, the material formerly occupying the grooves could have been dissolved in 102 years. This is a much shorter time than glacier occupancy and the assumption of chemical equilibrium is not critical.

The features on Kelleys Island are fluted valleys about 6 m across and each flute is about 0.1 m in width. Each flute represents a subglacial channel but only one or two of these channels operated at any one time. Blockages, perhaps caused by basal debris, caused the rivers to make minor course changes and in many instances it is possible to determine the order of channel occupancy. The channels were striated after abandonment by water when moving basal debris or debris-laden ice occupied the channels. From the Gauckler-Manning formula and the potential gradient obtained from the ice-sheet model the river velocity is calculated to be about 4 m s–1, in rough agreement with the water flux calculated earlier.

The grooves support the concept that subglacial water flows in “Nye” channels, and raise the suggestion that subglacial erosion by solution is of widespread importance.

Most numerical models of present ice-sheet dynamics predict basal thermal conditions for an assumed geothermal heat flux and measured ice thickness, surface temperature, and snow precipitation. These models are not ideally suited for reconstructing former ice sheets because what is known for present ice sheets is unknown for former ones, and vice versa. In particular, geothermal heat fluxes are immeasurable at an ice-sheet bed but can be measured after the ice sheet is gone, and the thermal conditions predicted at an ice-sheet bed can be inferred from the glacial-geological–topographic record after the ice sheet is gone. The Maine CLIMAP ice-sheet reconstruction model uses these inferred basal thermal conditions to compute ice thicknesses from basal shear stresses.

Basal shear stress is assumed to reflect the degree of ice–bed coupling which, in turn, is assumed to reflect the amount and distribution of basal water under the ice sheet. Under the ice-sheet interior, basal water exists in a thin film of constant thickness covering the low places on the bed. This film expands for a melting bed and contracts for a freezing bed. Along the ice-sheet margin, basal water exists in narrow channels of varying thickness corresponding to troughs on the bed. These water channels become deeper for a melting bed and shallower for a freezing bed. In areas covered by the Laurentide and Scandinavian ice sheets, myriads of interconnected lakes in regions of greatest postglacial rebound are interpreted as evidence suggesting the interior basal water distribution, whereas eskers pointed toward terminal moraines and troughs across continental shelves are interpreted as evidence suggesting the basal water distribution toward the margins. Continental-shelf troughs were assumed to correspond to former ice streams, by analogy with observations in Greenland and Antarctica.

Three modes of glacial erosion are considered to be responsible for the lakes, eskers, troughs, and associated topography. Quarrying is by a freeze-thaw mechanism which occurs where the melting-point isotherm intersects bedrock, so it is important only for freezing or melting beds because high places on the bed are frozen, low places are melted, and minor basal temperature fluctuations shift the isotherm separating them. Crushing results when rocks at the ice-bed interface are ground against each other and the bed by glacial sliding, so it occurs where the bed is melted and is most important when the entire bed is melted. Abrasion of bedrock occurs when rock cutting tools imbedded in the ice at the ice–rock interface are moved across the interface by glacial sliding, so it is also most important when the entire bed is melted. If basal melting continued after the entire bed is melted, abrasion-rates drop because the basal water layer thickens and drowns bedrock projections otherwise subjected to abrasion. Basal freezing reduces both crushing and abrasion-rates by coating quarried rocks with a sheath of relatively soft ice and transporting them upward from the ice–rock interface.

An initially flat subglacial topography will develop depressions where glacial erosion is greatest and deposition is least, and ridges where the opposite conditions prevail. We interpret the central depressions represented today by Hudson Bay and the Gulf of Bothnia as caused by erosion on a melting bed under the Laurentide and Scandinavian ice sheets, respectively. The arc of lakes, gulfs, and shallow seas surrounding these depressions are interpreted as resulting from a freezing bed under the former ice sheets. The present watershed separating the depressions from the arcs marks the approximate former basal equilibrium line where the bed was melted. The Canadian and Baltic continental shields beyond these arcs are blanketed by material eroded from within the arcs, and represent areas having a frozen bed where evidence for abrasion is missing and a second zone having a melting bed where evidence for abrasion is present. This basic pattern was assumed to be imprinted on the bed during the steady-state period of maximum ice-sheet extent, and maintained in varying degrees during growth and shrinkage of these ice sheets.

A technique for drilling deep holes with a hot-water jet has been developed in recent years at our institute (Iken and others, [1977]). The holes have served to investigate the water pressure at the bed of various Swiss glaciers since 1973. Drainage occurred naturally in rare cases when the drill reached the bed, but more often it was necessary to use explosives first, probably because the drill was stopped short of the bottom of the glacier by rock inclusions in the ice. In order to record piezometric water pressure over an extended period of time it was necessary that water was draining fairly continuously into the hole, otherwise the water level dropped eventually to a great depth when the weather turned cold, whereupon the holes closed off. By suddenly shutting off the water supply to a hole and observing the lowering of the water level with time some information on the channel characteristics has been obtained. In many cases there was little change of level, indicating that such a hole gives almost the true pressure head of the subglacial drainage system.

Our efforts have so far been concentrated on flat tongues of fair size over 100 m thick, the distance from the uppermost hole to the terminus ranging from about 1 to 4.5 km. The main characteristic of the water pressure is a very large diurnal variation of the order of 100 m and more. The mean pressure generally rises and falls in times of high and low water supply, respectively, but re-adjusts to approximately the original level within a few days. Mean levels are higher early in the melt season than later, and the amplitude of the diurnal variation has a tendency to increase with time, but also shows strong short-term modulations depending on the water supply. From observations on moulins and a hole which had remained connected to the bed from the previous year it seems likely that at the beginning of the melt season the water pressure at the bed may become as large as or larger than the ice pressure.

On Gornergletscher a record of water pressures has been obtained during the drainage of Gornersee, an ice-dammed lake at the confluence of the two main branches of the glacier. Levels stayed high day and night in the piezometer holes, which were located off to the side of where the main drainage channel was suspected to pass through. The surface drainage pattern was affected over a large area of the glacier. These two observations indicate that during the drainage of the lake some sort of sheet flow must have occurred. The surface velocity of the glacier roughly doubled during that time, but no lifting-up of the ice was observed within the accuracy of the survey. On one occasion on Glacier de Breney the main channel must have been blocked temporarily between an upper area, where the water level was rising simultaneously in three piezometer holes, and an area further down-glacier, where the holes were not affected.

The results of systematic movement studies carried out by means of an automatic camera on the Unteraargletscher since 1969 (Flotron, 1973) are discussed together with more recent findings from theodolite measurements made at shorter intervals and over a longer section of the glacier.

In addition to the typical spring/early-summer maximum of velocity known from other glaciers, an upward movement of up to 0.6 m has been recorded at the beginning of the melt season. It was followed, after various fluctuations of the vertical velocity, by a similar but slower downward movement which continued at an almost constant rate for about three months. The uplift was not confined to the section covered by the camera but occurred nearly simultaneously in profiles located 1 km below and 2 km above. The times of maximum upward velocity (increases of up to 140 mm/d) coincided approximately with periods of large horizontal velocity and occurred after increases of melt-rate.

The following explanations for the variations of vertical velocity are considered: (1) Changes of longitudinal strain-rate. (2) Changes of the sliding velocity in a channel of variable width and with a bed slope deviating from horizontal. (3) Changes of volume due to opening or closing of crevasses. (4) Swelling or contraction of veins at the grain edges. (5) Growth (and closure) of cavities in the interior of the glacier. (6) Changes of large-scale water storage at the bed.

Although all of the mechanisms (1)–(5) have some effect on the vertical ice movement, they cannot account for the observed variations of vertical velocity. We therefore conclude that large-scale water storage at the bed is the main cause of the uplift. Apparently the storage system is efficiently connected with the main subglacial drainage channels only during times of very high water pressure in the channels.

The findings are of some interest to the concepts of glacier sliding: As mentioned above the maxima of horizontal velocity—and thus of the sliding velocity—have not been measured at the time when the storage had attained a maximum, but at the time of maximum vertical velocity, which we assume to be the time of most rapid growth of cavities at the bed. This behaviour of the sliding velocity agrees with that predicted by a simple finite-element model of the basal ice on a wavy bed with water-filled cavities. In particular, the model shows that the sliding velocity is larger during the process of cavity growth than at the final stage when the cavities have grown to the size which is stable for the applied water pressure.

A hydraulic mechanism is postulated as the source of certain low-frequency (1-5 Hz) seismic signals that are recorded in the vicinity of temperate glaciers. To illustrate that the mechanism is a plausible seismic source, the equations of motion and continuity for transient flow in a subglacial conduit are developed. The solution to these equations is obtained for the case of the flow of water being terminated in a glacial conduit. Using reasonable values of physical parameters taken from the literature, the displacement of the conduit walls is determined as a function of time and distance along the tunnel. The spectral character of the proposed source and the period of time over which it operates is compared with seismic records of typical type II signals. The agreement in terms of these parameters is good.

To further examine the validity of the proposed seismic source a preliminary energy calculation is made which demonstrates that the proposed mechanism is energetically reasonable. Evidence of hydraulically generated seismic waves from a dam site is introduced to support the theory of a hydraulic source.

Based on the evidence presented, it is suggested that the flow of water from glaciers is modulated by an internal flow regime that is intermittent in time and space.

The substance of this paper is included in a report by the present authors entitled “Hydraulic transients: a seismic source in volcanoes and glaciers”, published in Science, Vol. 203, No. 4381, 1979, p. 654–56.

The efficiency of drainage systems within the glacier Austre Okstindbreen apparently increases as the summer progresses. As the transient equilibrium line moves up-glacier and the main system extends headwards, several systems may join together. Much of the detail of the glacier river hydrograph depends on the development of the internal drainage systems, although its general form reflects climatic conditions. Streams flowing from both Austre Okstindbreen and the adjacent, much smaller, steeper glacier, Charles Rabots Bre respond quickly to rain, but the sustained high summer discharge of the Austre Okstindbreen river, even during periods without rainfall and ablation, indicates that much water is stored within the glacier. The volume of water leaving Austre Okstindbreen in summer partly reflects the amount of snow accumulated during the preceding winter. Surface structures and internal drainage systems convey water rapidly towards the margins of both glaciers, but their rivers do not display regular diurnal variations of level until part way into August; at glacier rivers further south, such fluctuations occur throughout the summer. Each year, in late July or early August, a lake dammed by Austre Okstindbreen drains through the glacier. Drainage may be triggered by a decline of water pressure in conduits near the lake.

The origin of drumlins has long been one of the most intriguing, yet unsolved problems in glacial geology. It seems that until the mechanism responsible for drumlin formation is fully understood some of the key glaciological problems related to the glacier bed will remain obscure.

A strong relationship between the shapes of glacier excavation and some specific structures of the bedrock seems to exist in all the cases observed by the authors. Important examples of this kind include present and Quaternary glaciers in the Iberian Peninsula, the Alps, and the Peruvian Andes. A preliminary mathematical model has been developed, a brief description of which is given in this paper. Within the framework of this model the above interdependence can be interpreted. In this way the excavation mechanism that occurs in stationary glaciers can be explained without requiring the use of more sophisticated theories implying quite complex morphological phenomena. Suggestions for further studies are also made.

The physical characteristics of the ice and debris of the Matanuska Glacier, Alaska (lat. 61° 47’, long. 147° 45’) and their consistent variation in a vertical stratigraphic column of the glacier indicate multiple origins for both. The two zones of ice recognized in the glacier are the upper englacial zone, which comprises the bulk of the glacier (est. 300 m thick), and the thin (3 to 15 m thick) basal zone, which contains an estimated 97% of the sediment transported in the glacier. The physical characteristics of the ice and the quantity and mode of distribution of debris in it define two ice facies in each zone. The englacial zone consists mainly of the debris-poor (0.002 volume per cent) diffused facies. White coarse-bubbly ice and blue coarse-clear ice (2 to 10 cm diameter) comprise most of this facies. The bubble-rich banded facies (debris bands) contains abundant debris (25 to 50 volume per cent), varies in lateral and vertical extent, and occurs randomly in the diffused facies.

Studies of radar reflection characteristics can indicate much about the bed of an ice sheet and the associated ice motion. This contribution analyzes the relative significance to the ice flow of rough ice-rock, rough ice-water-rock, and flat ice-water-rock interfaces, and establishes criteria for distinguishing among these interface types under ice sheets using radar equipment of the type developed by the Technical University of Denmark.