Previous studies of alpine glaciers have demonstrated that as water discharge increases through the summer, the predominant mode of subglacial drainage shifts from a distributed system to a more efficient conduit drainage system. We observed an early-melt-season speed-up and flood event lasting roughly 2 days in a small, uncomplicated Alaskan glacier that appears to have resulted from a sudden shift of the subglacial system in response to a significant accumulation of meltwater within the glacier. Calculated melt-water inputs exceeded discharge before the event; the implied change in storage over this 10 day period was equivalent to roughly 0.13 m averaged over the entire glacier bed. The pattern of discharge and suspended-sediment variations and the appearance of large ice chunks in the stream suggest that the speed-up occurred during a period of establishment of new subglacial conduits. A culminating flood and associated suspended-sediment pulse appear to have marked the final establishment of the new section of subglacial conduit. The flood ended the episode of high sliding velocity, but released water with high solute concentrations that reflect relatively long contact time with sediments. Discharge of stored water, inferred from high solute concentrations and lack of diurnal variation in discharge, continued for at least 3 days. While events such as this must recur through the melt season as the conduit system extends up-glacier and the locus of meltwater inputs shifts, their manifestations in the outlet stream will likely be more subdued later in the season.

The transition from a dry, sub-freezing snow cover to isothermal snow freely conducting water was observed near timberline in the Sierra Nevada, California, U.S.A. Although there were few major stratigraphic boundaries in the snowpacks observed, minor textural discontinuities were sufficient to both route water laterally downslope and later create ice layers at these strata. During the first few days of snowmelt, downslope movement of water parallel to the strata allowed water to bypass lower layers of the snow cover on slopes and converge in microtopographic depressions on level terrain, creating isolated drains through the snowpack. In addition to these drains, flow fingers associated with holes in the ice sheets delivered water to the base of the snow several days prior to the thorough wetting of the snowpack. Formation of a series of ice lenses just below the snow surface every night released much more latent heat than did ice-layer development within the bulk of the snowpack or at its base.

Within the framework of core-drilling through the permafrost of the active rock glacier Murtèl–Corvatsch in the Swiss Alps, subfossil stem remains of seven different bryophyte species were found at a depth of 6 m below surface and about 3 m below the permafrost table in samples from massive ice. The composition of the moss species points to the former growth of the recovered mosses in the nearest surroundings of the drill site. A total of 127 pollen and spores captured by the mosses and representing 23 taxa were determined. The local vegetation during deposition time must be characterized as a moss-rich alpine grassland meadow rich in Cyperaceae, Poaceae, Chenopodiaceae and Asteraceae, comparable to today’s flora present around the study site. For l4C analysis, accelerator mass spectrometry had to be used due to the small sample mass (about 0.5 mg Carbon content). The mean conventional 14C age of 2250 ± 100 years (1σ variability) corresponds to ranges in the calibrated calendar age of 470–170 BC and 800 BC to AD 0 at statistical probabilities of 68% and 95%, respectively. This result is compared with the present-day flow field as determined by high-precision photogrammetry and with information about the thickness, vertical structure and flow of the permafrost from borehole measurements. Total age of the rock glacier as a landform is on the order of 104 years; the development of the rock glacier most probably started around the onset of the Holocene, when the area it now occupies became definitely deglaciated. The bulk of the ice/rock mixture within the creeping permafrost must be several thousand years old. Characteristic average values are estimated for (1) surface velocities through time (cm a-1), (2) long-term ice and sediment accretion rates (mm a-1) on the debris cone from which the rock glacier develops, (3) retreat rates (1–2 mm a-1) of the cliff which supplies the debris to the debris cone and rock glacier, and (4) ice content of the creeping ice/rock mixture (50–90% by volume). The pronounced supersaturation of the permafrost explains the steady-state creep mode of the rock glacier.

Extensive damage to ice occurs during ice structure interaction by microcracking, recrystallization and melting. The objective of this work was to investigate this damage process under confined-stress conditions believed to be associated with impact zones that occur during ice–structure interaction. “Damage” refers to microstructural modification that causes deterioration of the mechanical properties. Prior experimental work has shown that a small amount of deformation causes permanent damage in ice, leading to enhanced creep rates during subsequent loading. To investigate this softening, freshwater granular ice was deformed under moderate confinement (20 MPa) at –10°C, at two rates which bracket ductile and brittle behavior (10−2 s−1 and 10−4 s−1). Samples were deformed to different levels of axial strain up to 28.8%. Thin sections were examined to assess the progressive changes in microstructure.

Both grain-boundary and intra-granular cracking began at strains corresponding to the peak stress (1–2%) for tests at both strain rates. The peak stresses were 23.4 MPa for the tests at 10−2 s−1 and 9.8 MPa for the tests at 10−4 s−1. At strains of > 1–2%, dense clusters of intra-granular cracks began to develop in the samples tested at the higher rate. At the lower rate, dynamic recrystallization was apparently the dominant deformation mechanism beyond the peak stress. The average grain-size decreased strongly during the first few per cent strain and then maintained a relatively stable value.

To develop new strategies for global mass-balance monitoring, data for the period 1961–90 have been compiled for 80 glaciers with a variety of mass-balance and morphological parameters. This dataset is significantly larger than that used in previous studies. This allows us to check the mass-balance data for both strong and weak correlations with different glacier parameters. In many cases, the strong correlations suggest new approaches to monitoring glaciers on a global scale. For example, the mass balance at the terminus is strongly correlated with the difference in elevation between the terminus and the glacier’s mean elevation. These easily measured parameters could be particularly useful in assessing maximum ablation and meltwater potential based on altitudes derived from maps and photographs. Good correlations also exist between differences in mass-balance parameters (e.g. net balance minus terminus balance) and several other morphological properties (e.g. elevation range and length). Equally important, the weak correlations demonstrate that some relationships commonly used on individual glaciers are not appropriate when considering global monitoring strategies. For example, the correlation between net mass balance and terminus balance is very poor. Likewise, the correlation between the net mass balance and equilibrium-line altitude is weak, and the correlation between the net mass balance and activity index is almost non-existent. This suggests that although these climatically sensitive parameters may be closely related on individual glaciers, these same relationships are not reliable as tools for monitoring glaciers on a global scale.

For more than 30 years the quantitative method of evaluating stability (e.g. Roch, 1966; Föhn, 1987; Jamieson, 1995; Jamieson and Johnston, 1998a) has been focused on calculation of a strength-to-load ratio (or stability index): when the shear stress applied to the weak layer reaches the shear strength, failure is imminent. However, field observations combined with experience and measurements indicate that snow-slab temperatures and slab hardness can have a strong influence on dry-snow slab stability. In this paper, we present a simple static analysis of the stability index, and discuss the importance of slab temperatures and hardness and macroscopic size effects (factors not contained in the stability index) on snow-slab stability. Our conclusion is that the traditional method lacks some elements which are very important in snow-slab stability, particularly when skier triggering is involved.

Previous work on the relation between glacier volume and area and on accumulation area ratios suggests that balance rates measured at the glacier terminus are not constant or random from glacier to glacier but instead scale with glacier length. Using mass-balance data from a collection of 68 valley and cirque glaciers, we show that the terminus mass-balance rate scales roughly linearly with surface area and scales with length raised to an exponent constrained to fall roughly between 0.5 and 2 with 1.7 preferred if a glacier’s length is dependent on the mass-balance conditions (rather than balance being dependent on length). When these exponents are used to predict valley-glacier volume–area scaling, the results are very close to empirical volume–area observations. Although the data are noisy and the proposed fits could be modified by improved observations, the scaling trend for terminus balance vs length remains clear. Although the exact value of the scaling exponent is not well determined, establishing the existence of this scaling relation will be important for studies of climate change and the impact of glacier recession on sea level.

The effect of subglacial lakes upon ice-sheet topography and the velocity patterns of subglacial water-sheet floods is investigated. A subglacial lake in the combined Michigan–Green Bay basin, Great Lakes, North America, leads to: (1) an ice-sheet lobe in the lee of Lake Michigan; (2) a change in orientations of flood velocities across the site of a supraglacial trough aligned closely with Green Bay, in agreement with drumlin orientations; (3) low water velocities in the lee of Lake Michigan where drumlins are absent; and (4) drumlinization occurring in regions of predicted high water velocities. The extraordinary divergence of drumlin orientations near Lake Ontario is explained by the presence of subglacial lakes in the Ontario and Erie basins, along with ice-sheet displacements of up to 30 km in eastern Lake Ontario. The megagrooves on the islands in western Lake Erie are likely to be the product of the late stage of a water-sheet flood when outflow from eastern Lake Ontario was dammed by displaced ice and instead flowed westward along the Erie basin. The Finger Lakes of northern New York state, northeastern U.S.A., occur in a region of likely ice-sheet grounding where water sheets became channelized. Green Bay and Grand Traverse Bay are probably the products of erosion along paths of strongly convergent water-sheet flow.

This work investigates the deformation of ice under deviatoric stresses and confining pressures expected during ice–structure interaction. Granular ice was tested under a range of confining pressures (5–60 MPa) and deviatoric stresses (up to 25 MPa), with sample temperatures between –8° and –10°C. Samples were deformed to increasing end-levels of axial strain, and were thin-sectioned and photographed immediately following testing.

At all confinement levels, the original texture of the sample is completely transformed within the first 10–15% strain, to a fine-grained matrix with a few larger, isolated grains. At low confinements, grain-size reduction is associated with extensive microcracking. At high confinements, few cracks are observed. Observations suggest that microcracking, melting and recrystallization are active at all levels of confinement, though the relative importance of each depends on the confinement, stress and accumulated strain.

Deviatoric stress is a strong factor in controlling the creep, reflected in both the time required for the sample to reach accelerated creep and the tertiary creep rate itself. Two exceptions to this pattern were noted. First, some samples experienced strain localization and eventual rupture. These specimens tended to have higher creep rates even in the initial stages of strain. Second, prior damage resulted in rapid softening compared with the behavior of undamaged specimens. However, when strain rates are compared among all samples at a given level of cumulative axial strain, the creep behavior obeys a power law over the whole range of strain levels tested. Effective viscosity decreased from 107.8 to l06.4 MPa−n s within the first 10% strain, during which the most substantial microstructural changes occurred, and then stayed relatively stable. The stress exponent, n, remained within the range 4.0–4.6.

The dominant deformation mechanism appears to depend strongly on confining pressure (cracking at low pressure and dynamic recrystallization at high pressure). Creep rates at high confinement appear to increase relative to those at intermediate confinement, but the influence of temperature must be addressed further.

Two 10 cm lengths of firn core from the Dyer Plateau (70°31 ′S, 65°01 ′W), Antarctic Peninsula, were used to carry out a laboratory experiment to investigate the migration of methane sulphonate, the anion of methane sulphonic acid (MSA), in natural firn. Each length was cut vertically into four pieces, and a dopant solution containing Cl−, NO3−, SO42−, F− and MSA− pipetted onto the top of three sections, the fourth being kept as a blank. The doped sections were stored vertically, with the doped end uppermost, for 8 months at a range of temperatures (nominally –5°, –10° and –22°C) before subsampling at 1 cm resolution and analysis by ion chromatography. The two firn lengths were treated identically and the results were consistent. Profiles of the doped firn sections showed that C1−, NO3 and SO42− remained in the uppermost subsample, although the NO3− concentrations were variable compared to the blank. The F− profile shows slightly elevated concentrations in the second sample down compared to the blank, at temperatures of -10°C and above. The MSA− showed higher concentrations in the second, third and fourth samples down at -10°C and above, which indicates that some percolated downwards from its original position at the top of the core. This experiment shows that MSA is mobile in warm firn even over a short period of time. We propose that the mechanism for the mobility of MSA− in natural firn is via liquid MSA drainage, though we cannot yet discount vapour phase transport.

We performed microscopic observations and a statistical study of the number, size and shape distribution of clathrates in the GRIP (Greenland Ice Core Project) deep ice core, using 185 samples from a depth range of 1016–3014 m, spanning a period of 6 to >110 ka BP and encompassing the Holocene, Wisconsin and Eemian periods. The number concentration of the clathrates varied considerably with climatic changes. It was possible to detect the rapid climatic oscillations in the last glacial between 13 and 110 ka BP, the Dansgaard–Oeschger cycles, in the number-concentration profile of clathrates. The mean volume of clathrates is less clearly influenced by climatic factors, with a tendency towards greater volumes in warmer periods, but also a growth of clathrates with depth could be detected. This growth rate was calculated to be 3.1 × 10-12 cm3 a-1. The amount of gases captured in the clathrates is estimated to be significantly smaller than the total amount of air determined by gas-concentration measurements. This points to diffusion processes of atmospheric gases within the ice matrix.

To predict the distribution of motion beneath glaciers on soft beds, the strength of the coupling between the ice and the bed and its variation with effective pressure must be known. A record of shear strain, acquired with a tiltmeter emplaced in till beneath Storglaciären, Sweden, indicates that fluctuations in water pressure cause variations in the local shear stress on the bed and that the bed deforms elastically in response to these variations. To estimate the shear stress from the elastic component of the total shear strain, the shear modulus of the till was measured in relaxation tests conducted in the laboratory with a ring-shear device. After accounting for the elastic compliance of the device, these tests yielded shear moduli of about 1000 and 1800 kPa at confining pressures of 85 and 280 kPa, respectively. These values are comparable to those of other granular materials undergoing recoverable shear strains of the same magnitude. The local shear stress on the till, calculated by applying the measured shear moduli to the tilt record, scales with Pe1.7, where Pe is the effective pressure. This relation implies that as Pe decreases at the ice/till interface, shear stresses on the till are reduced and concentrated elsewhere on the bed, perhaps where the till is absent or the glacier is frozen to the bed. When compared with the strength of the till determined from ring-shear tests, this relation also accounts for the lack of permanent deformation at depth in the bed during periods of low Pe and indicates that most basal motion was by sliding or ploughing.

During a full-scale iceberg-impact study conducted in July 1995 on the Labrador coast, Canada, a sample of ice was retrieved from the impacted surface of an iceberg. The sample was thin-sectioned and the observations of the contact-zone microstructure are presented in this paper.

Thin sections were prepared from two slabs cut parallel to the impacted surface. In each of the thin sections taken from the impacted-surface slab, fine-grained material was found to surround parent-size grains (as observed in the second slab). A boundary between the parent grains and the grains of modified microstructure was found running approximately parallel to the impacted surface in each of the thin sections taken from the impacted-surface slab. This boundary was pronounced towards the edges of the contact zone. Lateral movement of grains outward along this boundary was observed in thin sections near the edges but not near the centre of the contact zone.

The thin sections were compared to the results of medium-scale indentation tests in 1989 and 1990 from the Arctic Ocean. The same type of fine-grained material and layer formation of modified microstructure was found in the contact zones.

With the use of a numerical three-dimensional (3-D) model the flow dynamics of the confluence area of Unteraargletscher, Bernese Alps, Switzerland, are studied. Previous predictions, based on conceptual two-dimensional models, about flow characteristics at confluence areas are tested against results from the fully 3-D model. Measured winter velocities are used for model verification. Despite some consistent systematic differences, good overall agreement between measured and calculated surface velocities is obtained. The calculated vertical strain-rate variation with depth is in good agreement with available measurements from boreholes. The ice is found to be almost three times stiffer than standard estimates of rheological parameters for glacier ice would predict. The model predicts a complicated yet realistic pattern of vertical velocity variation along the surface. The most noticeable features of the vertical velocity distribution across the surface are listed, and their relation to topographic surface undulations and the overall dynamics of the confluence discussed. In accordance with previous results from analytical models, a strongly localized surface trough and a concomitant negative (downward orientation) vertical velocity anomaly develop at the junction point. Although depth-integrated strain rates are positive (extension), the basal layer is compressed vertically. The ice-cored medial moraine is formed by differential ablation. The flow mechanics of the confluence area play only an indirect role, by enabling transfer of debris-covered marginal ice towards the confluence center. In the absence of differential ablation, an elongated surface depression would be formed in the down-glacier direction from the junction point instead of an elevated ice-cored medial moraine.

The relation between the local effective pressure and shear stress on till beneath Storglaciären, Sweden, discussed in Iverson and others (1999), provides an empirical basis for studying the processes that control the strength of the ice/bed coupling. Particles in the bed that protrude into the glacier sole support shear stresses that are limited by either ploughing or the traditional sliding mechanisms. Model calculations, based on studies of cone penetration through fine-grained sediment and sliding theory, agree with the observed relation between shear stress and effective pressure if the water layer at the ice/bed interface is assumed to thicken rapidly as the effective pressure approaches zero. Studies of the hydraulics of linked cavities provide support for this assumption, if the mean thickness of the water layer reflects the extent of microcavity development at the interface. Comparison of the calculated shear stress with the ultimate strength of till suggests that bed deformation limits the shear stress on till beneath Storglaciären only at intermediate effective pressures; at very low effective pressures, like those inferred at the site of the tiltmeter discussed in Iverson and others (1999), and at sufficiently high effective pressures, ploughing and sliding should focus motion near the glacier sole. A calculation using parameter values appropriate for Ice Stream B, West Antarctica, suggests that ploughing may occur there at shear stresses not sufficient to deform the bed at depth. This conclusion is reinforced by the likelihood that pore pressures in excess of hydrostatic should develop down-glacier from ploughing particles, thereby weakening the bed near the glacier sole. However, given the apparent sensitivity of the ice/bed coupling to basal conditions that may be highly variable, any blanket assumption regarding the flow mechanism of ice masses on soft beds should probably be viewed with skepticism.

Radar interferometry, ice-penetrating radar profiles and an elevation model are used to determine the velocity fields, rates of ice discharge, approximate states of balance and catchment area for three large outlet glaciers in northeast Greenland. Discharge through flux gates is calculated for Humboldt and Petermann Gletscher, which are found to be in balance (at the level that the accumulation is known). A large difference between the measured and estimated fluxes for Ryder Gletscher may be a reflection of unsteady flow behavior for this glacier. The patterns of ice flow for the three glaciers considered are each unique, showing that the nature of ice discharge varies substantially from basin to basin, controlled by bed conditions and the presence of subglacial troughs and obstructions.

The interannual variability of glacier mass balance is expressed by the standard deviation of net balance, which varies from about ±0.1 to ±1.4 m a−1 for a sample of 115 glaciers with at least 5 years of record. The standard deviation of net balance is strongly correlated with the mass-balance amplitude (half the difference between winter and summer balances) for 60 glaciers, so the amplitude can be estimated from net balance standard deviation for the other 55 glaciers where winter and summer balances are unavailable. The observed and calculated mass-balance amplitudes for the 115 glaciers show contrasts between the Arctic and lower latitudes, and between maritime and continental regions. The interannual variability of mass balance means that balances must be measured for at least a few years to determine a statistically reliable mean balance for any glacier. The net balance of the Greenland ice sheet is still not accurately known, but its standard deviation is here estimated to be about ±0.24 m a−1, in agreement with other Arctic glaciers. Mass-balance variability of this magnitude implies that the ice sheet can thicken or thin by several metres over 20–30 years without giving statistically significant evidence of non-zero balance under present climate.

This paper examines the debris-rich basal ice layer from Worthington Glacier, Alaska, U.S.A., a small valley glacier overlying rigid bedrock. The debris-rich basal ice layer studied shows evidence for large-scale longitudinal compressive deformation (isoclinal folds and nappes), similar in style and magnitude features to that reported from push moraines formed in glacial sediments. The debris-rich ice largely comprised stratified solid ice (layers of alternating debris-rich and debris-poor ice) which we suggest results from the tectonic attenuation of folds produced from the deformation of the frozen debris, glacier ice and bubble-rich ice that comprise the initial basal layer of Worthington Glacier. Beneath the glacier lies a thin bed of saturated diamicton which contains evidence of limited movement. It is suggested that this is the result of the partial melt-out of the debris-rich basal ice layer which then behaved as a local (and seasonal) thin deforming layer.

It is suggested that this example, from a valley glacier flowing over rigid bedrock, provides further evidence that the processes of sediment transport, incorporation and deposition in the debris-rich basal ice are similar to and linked with those in the deforming layer.

A quasi-one-dimensional dense-snow avalanche model has been developed to predict avalanche runout and flow velocity in a general two-dimensional terrain. The model contains three different dense-snow-avalanche flow laws. These are: (1) a Voellmy-fluid flow law with longitudinal active/passive straining, (2) a Voellmy-fluid flow-law advanced by Russian researchers in which the Coulomb-like dry friction is limited by a yield stress, and (3) a modified Criminale—Ericksen—Filby fluid model proposed by Norwegian researchers. The application of the Voellmy-fluid law with active/passive straining to solve practical avalanche-dynamics problems is evaluated by applying the model to simulate laboratory experiments and field case-studies. The model is additionally evaluated by comparing simulation results using the Russian and Norwegian models. In a final analysis the influence of the initial conditions on avalanche runout is investigated. We conclude that the model resolves many of the shortcomings of the Voellmy–Salm model, which is traditionally used in Switzerland to predict avalanche runout. Furthermore, since the model contains the three well-calibrated parameters of the Swiss Guidelines on avalanche calculation it can be readily applied in practice. We discuss why we believe the Russian and Norwegian models are not ready for practical application. Finally, we show that many problems remain, such as the specification of the initial release conditions. We conclude that numerical models require a more detailed description of initial fracture conditions.