Evidence as to the potential roles of marine ice flows in the dramatic climatological changes which have occurred from the late Pleistocene to the present is reviewed, indicating the need for careful modeling studies to evaluate several crucial hypotheses. A scale analysis of the flow of a marine ice stream coupled to a freely floating ice shelf is presented, in two dimensions and ignoring thermodynamic effects. With these limitations, the most important control of the dynamics of the ice stream is associated with first-order buoyancy effects related to the density contrast ∆ρ/ρw between ice and sea-water. It is shown that longitudinal stretching, arising from large gradients in basal sliding velocity, dominates shearing deformation provided the aspect ratio ω2 « ∆ρ/ρw. The buoyancy control is established through the necessity of having continuously varying longitudinal strain-rates in the neighborhood of the grounding line.

The scale analysis is the basis for derivation of a simplified model of a fast-flowing ice stream coupled to a freely floating ice shelf. The distance in the ice stream up-stream from the grounding line over which the above dynamic regime extends is estimated and found to be relatively insensitive to the basal sliding velocity and to the rheological constant of ice. A further potentially important feed-back mechanism between ice stream and ice shelf is associated with buoyancy corrections to the longitudinal deviatoric stress field.

Apparent ice-surface topography is observed at several scales on Landsat multi-spectral scanner (MSS) imagery. Digitally enhanced MSS scenes from Antarctica and Nordaustlandet, Svalbard, are compared with ice-surface elevations from aircraft altimetry (relative accuracy 2–3 m) to show that this apparent topography is real. Apparent ice divides on Landsat images fit closely with divides on altimetric records. Ice-surface irregularities within drainage basins are also shown to be real. On Byrd Glacier, Antarctica, apparent “flow lines” coincide with ridges on altimetric records. Synoptic Landsat data, calibrated by information from aircraft altimetric flight lines, are used to classify the surface roughness of the ice caps on Nordaustlandet and 40% of the Antarctic ice sheet. On Nordaustlandet, the roughest ice is of amplitude 15–25 m and wavelength 3–4.5 km. Drainage basins with such rough surface characteristics may be associated with ice streams or possibly past surge activity. The most rough Antarctic terrain is up to 60 m in amplitude, with wavelengths of <10 km. The roughness of the Antarctic ice sheet increases with distance from ice divides, reflecting changes in the parameters affecting the transfer of basal stresses to the ice surface.

The energy budget of a snow or ice surface is determined by atmospheric variables like solar and atmospheric long-wave radiation, air temperature, and humidity; the transfer of energy from the free atmosphere to the surface depends on the stability of the atmospheric boundary layer, where vertical profiles of wind speed and temperature determine stability, and on surface conditions like surface temperature (and thus surface humidity), roughness, and albedo.

This paper investigates the conditions exactly at the onset or the end of melting using air temperature, humidity, and as the radiation term the sum of global and reflected short-wave plus downward long-wave radiation. For the turbulent exchange in the boundary layer, examples are computed with a transfer coefficient of 18.5 W m−2 K−1 which corresponds to the average over the ablation period on an Alpine glacier. Ways to estimate the transfer coefficient for various degrees of stability are indicated in the Appendix.

It appears from such calculations that snow may melt at air temperatures as low as –10°C and may stay frozen at +10°C.

Periods of dramatically accelerated motion, in which the flow velocity increases suddenly from about 55 cm/d to a peak of 100–300cm/d and then decreases gradually over the course of a day, occurred repeatedly during June and July 1978–81 in Variegated Glacier (Alaska), a surging-type glacier that surged in 1982–83. These “mini-surges” appear to be related mechanistically to the main surge. The flow-velocity peak propagates down-glacier as a wave at a speed of about 0.3 km/h, over a reach of about 6 km in length. It is accompanied by a propagating pressure wave in the basal water system of the glacier, in which, after a preliminary drop, the pressure rises rapidly to a level greater than the ice-overburden pressure at the glacier bed, and then drops gradually over a period of 1–2 d, usually reaching a new low for the summer. The peak velocity is accompanied by a peak of high seismic activity due to widespread fresh crevassing. It is also accompanied by a rapid uplift of the glacier surface, amounting to 6–11 cm, which then relaxes over a period of 1–2 d. Maximum uplift rate coincides with the peak in flow velocity; the peak in accumulated uplift lags behind the velocity peak by 2 h. The uplift is mainly due to basal cavitation driven by the high basal water pressure, although the strain wave associated with the mini-surge motion can also contribute. Basal cavitation is probably responsible for the pulse of high turbidity that appears in the terminal outflow stream in association with each mini-surge. In the down-glacier reach, where the mini-surge waves are attenuating, the observed strain wave corresponds to what is expected for the propagating pulse in flow velocity, but in the reach of maximum mini-surge motion the strain wave has a form quite different, possibly related to special features in the mini-surge initiation process from that point up-stream. The flow acceleration in the mini-surges is due to enhanced basal sliding caused by the high basal water pressure and the consequent reduction of bed friction. A preliminary velocity increase shortly before the pressure wave arrives is caused by the forward shove that the main accelerated mass exerts on the ice ahead of it, and the resulting preliminary basal cavitation causes the drop in water pressure shortly before the pressure wave arrives. The mini-surge wave propagation is controlled by the propagation of the water-pressure wave in the basal water-conduit system. The propagation characteristics result from a longitudinal gradient (up-glacier increase) in hydraulic conductivity of the basal water system in response to the up-glacier increase of the basal water pressure in the mini-surge wave. The mini-surge waves are initiated in a succession of areas situated generally progressively up-glacier during the course of the summer season. In these areas, presumably, melt water that has accumulated in subglacial (?) reservoirs is released suddenly into the basal water system immediately below, generating a pressure rise that propagates down-stream from there. Relationships of the mini-surges to the main surge are seen in the role of high basal water pressure in causing the rapid glacier motion in both phenomena, in the pulse-propagation features of both, and in the high outflow turbidity associated with both. The mini-surges of Variegated Glacier have a strong resemblance to movement and uplift events observed in Unteraargletscher and Findelengletscher, Switzerland. This bears on the question whether the mini-surges are a particular characteristic of surge-type glaciers prior to surge.

An opportunity to analyse the mechanism for the formation of a basal ice layer was provided by Glacier de Tsanfleuron in the Swiss Alps. This glacier has some lateral subglacial cavities in which a basal ice layer forms on the bedrock floor and is subsequently incorporated at the base of the glacier. Petrographic and crystallographic analyses of the different types of ice have provided a means of investigating the successive stages of the process. These analyses show major structural changes which provide field confirmation of the deformation mechanisms studied by several authors in laboratory experiments.

The characteristics of two full-depth slab avalanches that occurred in 1982 and 1984 at the same location on Meall Uaine in Glen Shee, Scotland, are described. Notable features include the relatively low altitude of the site (c. 670 m) compared to sites of other avalanches observed elsewhere in Scotland, the depth of snow involved (up to 4 m), a snow-pack similar to those recorded in the Cairngorm Mountains, and the entrainment, transport, and deposition of boulders and turf. The probability of the two avalanches occurring when they did was greater than 60%. It is suggested that further observations of full-depth slab avalanches might be obtained by identifying terrain similar to that of the site studied here.

Modes of formation and rates of collapse have been determined for kame terraces and a fan delta in the Adams Inlet area, Alaska. The hummocky and trenched surface of the kame terrace develops by mass-wasting processes caused by differential melting of buried ice. The dry channels in the collapsed part of the terrace are formed by melt water during stagnant-ice bursts. Measurements over a period of 17 years indicate that terrace back-wasting averages 4.3 m a−1. The fan delta is forming near sea-level by streams that derive part of their load from the kame terrace. The fan delta seaward of the retreating kame terrace is undergoing partial collapse to produce kettles ringed by concentric fractures. Vegetation, now as much as 5 m below high tide, suggests a fan-delta collapse rate of about 0.5 m a−1.

Temperature records from several randkluft sites at the margins and headwall of Boundary Glacier (lat. 52° 12′ N., long. 117° 12′ W.) in the Canadian Rocky Mountains are presented. These records indicate that during the ablation season a diurnal freeze-thaw air and rock-surface temperature regime occurs in a 2 m wide zone centred on the randkluft lip. Deeper in the randkluft, stable sub-zero conditions prevail whereas above the randkluft an above-freezing temperature regime prevails. The freeze-thaw temperature regime, observed freezing of melt water and rain water on randkluft rock surfaces, and copious in-situ loose weathering products on the headwall and marginal free faces suggest an active frost-shattering environment. Down-slope migration of the randkluft lip during the ablation season results in a migration of the freeze-thaw zone and thus a seasonal extension of the rock surface area exposed to the favorable weathering environment. Longer-term glacier fluctuations, with attendant thinning and thickening of the ice body, could result in large areas of marginal and headwall rock surfaces being exposed to the randkluft weathering environment over long periods of time. The data and observations from Boundary Glacier support an idea, suggested by Battle (1960), that open and shallow randklufts, rather than closed and deep bergschrunds, are a focus of weathering at glacier margins and thus a potential factor in cirque development.

Evidence of a recent (1985) ice-cliff avalanche from an outlet lobe of a small plateau ice cap on north-west Ellesmere Island is discussed. Former avalanche events are evidenced by debris lying outside the 1985 avalanche material. Periodic activity seems to be linked to the build-up of melt water in the crevasses of the outlet lobe during the melt season. The exact magnitude and frequency of events are unknown. Some implications to geomorphology and the sedimentology of sub-polar glaciers are discussed.

Inclinometry measurements in a 126 m bore hole slightly down-glacier from the crest of a riegel on Storglaciären showed what appears to be indisputable extrusion flow during the last 3 weeks of July 1984. The velocity at the bed was about five times that at the surface, and lay in a direction deviating 50–60° from that at the surface. The increase in velocity with depth began more than 50 m above the bed. In August and the first half of September the flow returned nearly to “normal” with only a slight tendency toward extrusion. The flow in July appeared to be related to a period of large diurnal water-pressure variation and cavity opening at the bed up-glacier from the riegel. Decoupling of the ice from the bed on the lowest parts of the riegel may have permitted the extrusion. The return to more “normal” conditions in August appears to have been related to a decrease in ice flux over the riegel and a widening of the zone of decoupling.

Our earlier monitoring program on Lewis Glacier, Mount Kenya, indicated a slow-down of the ice flow to January 1982, while based on the numerical modeling of the ice dynamics a further drastic decrease of the ice flow was predicted from the 1978 to the 1985 datum. This paper presents velocity measurements over the years 1982–83, 1984, and 1985. Changes of ice-flow conditions from 1978 to the mid 1980s are characterized by a velocity decrease by nearly half; a decrease of the maximum mass flux by more than half; a flattening and up-glacier shift of the velocity and mass-flux maxima; an up-glacier displacement of the transition between prevailingly longitudinal crevasses in the lower glacier and transverse crevasses in the upper glacier; and a terminus retreat by about 50 m. In consequence of the very weak ice flow remaining in Lewis Glacier, thinning and terminus retreat of the glacier are now primarily controlled by the in-situ net balance.

In a tunnel at the base of sub-polar Urumqi Glacier No. 1, China, three new mechanisms of glacier flow at sub-freezing temperatures have been observed. Taken individually or in combination, these modes of flow can account for nearly all (60–80%) of the overall glacier motion and, yet, they act only within the lowermost 1–2% of the effective glacier thickness. These mechanisms are: (1) enhanced deformation of the frozen and ice-laden subglacial drift; (2) motion across discrete shear planes or shear bands within the frozen drift or at the ice-drift interface; and (3) basal sliding at an ice-rock interface at a temperature of nearly −5° C. The ice-laden drift has an effective viscosity of more than one hundred times less than that measured in the overlying ice, thus allowing very rapid shear deformation. The observed rate or basal sliding at the ice-rock interface agrees favorably with that predicted by the recent work of Shreve (1984) if proper account is taken of the measured surface roughness and reduced ice viscosity.

A tunnel was excavated in Urumqi Glacier No. 1, at the Tianshan Glaciological Station. Ice temperature, ice displacement, deformation, and basal sliding, etc. were observed at regular intervals. It is shown that the temperature near the glacier bed is close to the melting point and that the largest proportion of the overall glacier motion is within the lowermost ice layers. The glacier ice is in a state of compression. The maximum shear strain increases towards the entrance of the tunnel, corresponding to the change in slope of the ice surface, and also towards the bedrock.

The formulation and application of a onedimensional sea-ice thermodynamic model is presented in this paper. The model’s sensitivity to changes in oceanic and atmospheric parameters is analyzed and compared with previous studies. The model is next applied to three locations in the Arctic: Cambridge Bay, Frobisher Bay, and Alert Inlet to study the model’s ability to simulate the annual cycle of first-year ice. The model’s results are compared with available climatological data and discussed in terms of the main thermodynamic processes, the combined effects of oceanic tides, and of sea-ice deterioration by melting on the break-up of sea ice.

It is shown that the model is effective in simulating the climatology of the first-year ice thickness at the three Arctic locations. The study also suggests that improved model performance can be expected from additional research and application of flexural forcing of the ice by waves and tides, and of deterioration of ice strength during the melting process.

Thin ice deposits, grown on a prismatic ice substrate and the first layers of large deposits grown about a cylindrical ice collector, are studied. Experiments were performed at 30 m s–1 wind speed and Ta near –20° C. Progressive structural changes in thin accretions with time of water-injection were observed, when the substrate temperature was Ts > –5°C. After 60s injection, which determined a deposit ~1 mm thick, the preferred orientation was established and the mean grain area was ~0.01 mm2. Such structural changes did not appear along the initial layers of thick deposits. These differed mainly from the subsequent zone because of the larger value of the preferred angle between the c-axis and the growth direction. These effects are discussed by considering the nucleation and orientation selection processes which occurred during droplet freezing and the annealing effects that occurred below the growth front. The different c-axis orientations in the initial and main zones are related to the different Ts values during their formation.

Mass-balance measurements have been renewed on two small ice caps on north-eastern Ellesmere Island. Original stake networks were established in 1972 and 1976. Since then, both ice caps have experienced significant mass losses averaging –70 to –140 kg m−2 a−1. They have also decreased in area. The equilibrium line in this area has averaged around 1150 m for the last decade or so. The ice caps are remnants of former climatic conditions and are out of equilibrium with contemporary climate.