Slush avalanches, primarily known from uninhabited arctic and mountainous regions, every year cause damage to man and property in Norway. Seeking objective criteria to identify their hazard zones, and methods for their prediction and control, the Norwegian Geotechnical Institute (NGI) has started a research program on these topics.

Thirty-four slush avalanches with known locations of crown surface have been investigated with regard to geomorphic and climatic characteristics. Starting zones, crown surfaces and tracks were classified according to geomorphic conditions important to avalanche release and motion.

Weak cohesionless snowpacks of coarse grains, and hard layers or crusts of ice in snow cover or on the ground, are critical conditions for slush avalanche release. Intense rain, falling on cohesionless new snow on these substrata, is the most striking feature of current weather situations.

The main starting zones were drainage channels, Sloping bogs, depressions and open fields. Channel-like, scar-like or bowl-like features appeared in the snow cover due to avalanche release. Three main types of release were identified: 1) Sudden release from crown surface, 2) drainage of snow-embanked, water-saturated snowfields through narrow outlets and 3) rapid headward growth from first point of release.

Enlargements and confinements along the avalanche tracks were caused by distinctly defined features in the tracks.

Reported are results of incorporating recent snow avalanche processes into hydrodynamic uniform flow equations, used to model motion of snow avalanches. Actual modifications include the relating of dissipative coefficients of the flow model to slab release depth, the representation of the material as a locking fluid, and the mini-segmentation of the avalanche path at low flow speeds in order to numerically accomodate viscous transition and avalanche cessation of motion. The purpose in looking at different formulations of the uniform flow hydrodynamic flow equations is to reduce the variation in the drag coefficients when the theory is applied to different avalanche paths, as compared to what has been previous experience. The model that reduced parameter variation the most was one in which the total drag force decreased in an intermediate velocity range, a mechanism that has had recent experimental verification.

Experimental data on avalanche impact pressures and speeds from an avalanche path at Rogers Pass, B.C., are presented for both dry and wet avalanches. Data from small load cells (645 mm2 surface) at various heights through the vertical cross-section of the avalanches show that the frequency of particle impact and hence flow density increases rapidly from top to bottom of the flow.

Data from a large circular impact plate (196 000 mm2) just above the ground surface show that dry avalanches have relatively higher peak pressures and relatively lower average pressures than wet avalanches for a given speed. Longitudinal wave-like characteristics in the flow may be seen for both wet and dry avalanches, but they appear to be the exception rather than the rule.

Comparison of pressure results with those of other studies shows good agreement for peak pressures, but average pressures are generally less than those of previous studies. This discrepancy is attributed to the past use of cells for which loading surface size approached typical particle size. The size effect may be demonstrated by comparison of small cell data with results from the large impact plate at Rogers Pass.

The startup dynamics of three wet snow avalanches in central Japan were evaluated by computer modeling the occurrences and comparing leading edge position-time data. Results were compared between three finite difference based computer codes, which were used to model the startup transients. Two of the computer codes use equations of uniform flow hydrodynamics, the third uses transient viscous fluid mechanics. The latter two codes also incorporate a material description of snow as a locking material. Results show a general increasing of frictional and/or viscous coefficients in the avalanche startup zones in order to match the kinematics of startup. Differences in results between the codes is attributed to the shape of the startup zones, whether convex or concave. The results indicate the magnitude of perturbation of startup on total avalanche runout time, which is likely to be negligible on long-duration avalanche occurrences.

The Ryggfonn project is a Full scale experiment carried out to investigate the impact of avalanches on structures and the effects of a retaining dam in the avalanche path. The vertical drop of the avalanche path is 910 m and the volume of the avalanches is 20 - 100 000 m3.

The experimental set-up consists of: a 15 m high retaining dam in the runout zone, instrumented with strain gauges on a 6.5 m steel mast and a load cell on a 1.0 m mast, plus a 4.5 m high concrete structure instrumented with three 0.72 m2 load cells, and three transmission line conductors strung across the avalanche path. The recorded analogue signals are digitized using Pulse-Code-Modulation (PCM) and recorded on a magnetic tape recorder.

Up to May 1984, speed data from five avalanches have been analysed, and impact pressures have been recorded from three of them. The maximum speeds vary between 38 m/s and 60 m/s. The maximum recorded impact pressure was 541 kPa, and averaged 220 kPa over a 15 second interval. A maximum pressure of 83 kPa was measured on a load cell buried under snow at the base of the concrete structure.

Over 30 individual slushflows were documented and observations were made of both high and low flow rates during 1979, 1980 and 1981 at Atigun Pass, central Brooks Range. Air temperature, precipitation and global radiation were recorded in 1980 and 1981 during the slushflow season. Meteorological conditions that exist in this arctic alpine environment at the time of break-up apparently are factors that encourage flow activity. Persistent low temperature prior to break-up allows the snowpack to remain intact until late spring. An influx of warm air from the interior of Alaska and long hours of radiation input produce an accelerated rate of snow-melt that plays a major role in slushflow occurrence.

Continuous avalanche flow speed measurements from start to stop of dense flow avalanches, and localized flow depth and slope perpendicular flow speed profiles, have been measured using X-band radar equipment- Maximum flow speeds of up io 65 m/s have been measured in large avalanches. Comparison of measured flow speeds with Voellmy-type model calculations shows that traditionally chosen sets of parameters, evaluated from runout measurements, cannot model the measured high speeds in the track.

To get an improved quantitative understanding of flowing avalanches, it is proposed that from the start to the stop different flow regimes with different mechanical characteristics are passed through. This is based on theoretical studies of the movement of granular materials, where in adddition to the flow velocity a mean random fluctuation velocity of grains is introduced. A partly or fully fluidized state may be taken according to whether the ratio of frictional to collisional energy loss is smaller or larger than unity. In a third regime the snow is non-fluidized, ie pure gliding of the rigid snow. A good agreement of measurements with the proposed mechanisms is shown.

This paper describes errors and corrections in snow heat flux when it is calculated using the numerical differentiation and integration method. The data obtained by the 20th Japanese Antarctic Research Expedition during GARP-POLEX in 1979 are used as a test case. Four factors are considered as causes of errors: a) temperature resolution, b) integration, c) determination of snow density and d) the deepest boundary condition. Factors a) and b) are significant in short term estimation, and the total error exceeds 90% if a daily value is calculated. The errors from a) can be reduced if the temperature is averaged over a long period, while those from b) become small in long term flux calculations. The total error can be reduced to 10% in monthly flux, while the improvement is limited by c) and d). If a constant thermal diffusivity is assumed between two levels of temperature measurement, a numerical filter which compensates for the effect of b) is composed. By using running averages in deep layers (z > 0.5 m) and the numerical filter in shallow layers (z <0.5 m), hourly flux can be calculated with errors of about 30%.

Heat and mass transport by an air and water-vapour mixture (pore air) in snow due to thermal convection has been computed. It can be shown theoretically that thermal convection is always occuring in newly fallen snow, but never occurs in snow with rounded grains. Rayleigh numbers of depth hoar and partly settled snow are close to the critical Rayleigh number, therefore a small change in one or more of the snow parameters will decide whether there is convective or conductive heat transport.

The influence of thermal convection on the mass transport rates of water vapour in snow is computed based on the assumption of the ventilation approach. Although thermal convection occurs more often than previously assumed, its influence amounts only to about 2-3%.

Thermal convection is observed in snow and in a compact of water-saturated glass beads. While uncertainty in the permeability of the snow limits our ability to compare the observed and calculated onset of convection, agreement between the observed and calculated effects of convection on heat transfer in snow is good. Experimental results with glass beads agree with both the calculated onset of and heat transfer by convection. Attempts are made to assess the effects of convection on snow metamorphism. While much is still uncertain about the significance of thermal convection in snow, it is clear that the phenomenon does occur.

This paper presents a small scale modelling of a snowdrift using activated clay particles.

Characteristic properties of activated clay particles, which are different from model snow particles proposed previously by other investigators, are fineness, high angle of repose and wide range of cohesion. Such properties may provide a similitude of a snowdrift and the phenomena caused by wet snow particles in a small scale model.

Experimental results presented in this paper show that activated clay particles are the most suitable substitute for natural snow particles in modelling, and indicate also the possibility of using them to model wet snow particles.

The equations of motion (continuity and momentum balance) for a dispersed, negatively buoyant particulate of snow entrained in a turbulent airflow contain apparent turbulent forces or turbulent particle buoyancies. These turbulent buoyancies arise from the constitutive assumption that the turbulent fluctuations of the snow phase velocity vector U’s, and the drift snow density ρ’s, are proportional to the deviatoric mean rate of deformation tensor for the airflow.

For an established, discretized airflow regime, the momentum balance equation for the snow phase can be solved by finite difference techniques for the snow particle velocity field. The snow phase continuity equation can then be solved for the drift snow density field.

The solutions for the snow phase equations of motion for a one dimensional airflow adjacent a solid surface show that the theory can reproduce an inertia! snow particle effect. The snow particle decelerates less rapidly than the airflow, resulting in the snow particle having a positive horizontal impact velocity at the solid surface, where air velocity goes to zero.

The solutions for the snow phase equations of motion for mixture flow and subsequent wind-aided snow accumulation on the immediate lee of a model mountain slope show that the theory can reproduce the geometries typical of wind-aided snow accumulation profiles, measured on the lee of mountain slopes.

An acceleration of the katabatic winds during periods of blowing snow was observed in Adelie Land, Antarctica. Data collected by Automatic Weather Stations' (AWS) showed a change in the relationship between the katabatic term of the surface geostrophic wind (katabatic force) and the wind speed for periods of blowing snow. When measurements of the katabatic force were plotted against the cube of the wind speed, the slope was steeper for wind speeds at less than a threshold speed for blowing snow. The difference between these two slopes was partly explained by the effect of blowing snow entrained into the atmospheric boundary layer.

Blowing snow was produced artificially in a cold wind-tunnel, and various measurements were conducted including particle diameters, concentrations, saltation lengths heat transport and electric charge. The mean diameter of blowing snow particles decreased only slightly with increasing height; in the saltation layer, standard deviation was large and velocities were scattered in a wide range, suggesting the complex dynamic process on taking-off. The mean saltation length ranged from a few cm to 40 cm increasing with wind velocity.

When wind blew without snow drifting, the static air pressure on the snow surface was smaller at higher levels, the vertical pressure gradient being negative. The pressure gradient became positive when blowing snow was initiated eg +9.6 Pa/m at 11.2 m/s and -8.3 °C. The magnitude of à downward force acting on a saltating snow partice caused by the pressure gradient was not large enough to explain the downward acceleration found from photographic analyses of particle trajectories.

Blowing snow particles were charged negatively the magnitude of charge increased with lowering temperature. Increase in vertical heat transfer was found in blowing snow by measuring the temperature of the air at various levels; the increase is reflected on that in the apparent turbulent diffusion coefficient.

The paper describes experience gained in Norway regarding the design criteria and use in practice of snow fences. The paper is based on theoretical studies on drifting snow, model experiments and experience accumulated through practical consulting work.

Snow fence design is a compromise between the storage capacity and minimization of dimensioning forces. Design considerations include fence height H, total snow fence density, and the gap between ground and fence. A gap of 0.15H - 0.2H and a total density of 45% are usually recommended. On ridge crests the gap can be reduced to 0.1H and in areas where snow depths exceed 2.0 m, it can be increased to 0.3H. In such cases the fence density should be varied such that the total density, including the gap, will remain near 45%. The height of the snow fences should be kept within 3.5 -4.5 m and the snow fences should be erected on the windward side of obstacles that create snowdrifts. The minimum distance from fence to road should not be less than 15H; in certain circumstances in coastal climate, this can be reduced to 10H.

Observations of drifting snow were carried out at Mizuho Station (70°42'S, 44°20'E, 2230 m above sea level), East Antarctica, in 1982. Drift flux was proportional to about the 8th power of wind velocity above 1 m and about the 4th power below 0.1 m, while snow drift transport rate was proportional to about the 5th power. For drift flux at 1 m height, the power had a temperature dependence, decreasing above -20 °C. Visibility was proportional to about the -8th power of wind velocity; this is explained by the power relation between drift flux and wind velocity. The repose angle of drifting snow particles was observed by the inclination of a cone-shaped deposit on a disk; it was more than 80° when snow was falling and less than 80° without precipitation. The fall velocity of drifting snow particles, obtained by time-marked trajectories of particles, was between 0.3 and 0.9 m/s, and depended on wind velocity and snow particle shape.

The shift of the equilibrium line in response to climatic disturbances is calculated for the International Glaciological Greenland Expedition (EGIG) profile of the Greenland Ice Cap. Data of heat balance studies carried out during EGIG 1959 and EGIG 1967 are used according to Kuhn’s algorithm. The following parameters are applied; the heat transfer coefficient of the sensible heat flux; the effective emissivity of the atmosphere for long wave downward radiation; a factor relating net radiation to cloudiness; the duration of the ablation season; the altitudinal temperature gradient; the altitudinal accumulation gradient.

The altitudinal shift of the equilibrium line results for the EGIG profile + 77m/°C and -4 m per 1/10 cloudiness with 35d of ablation. Changes in heat balance by formation of superimposed ice are taken into account.

Crystal size in polar ice caps increases with depth from the snow surface down to several hundred meters. Data on crystal growth in isothermal polar snow and ice show the same linear relationship between the size of crystals and their age. This paper reviews the mechanical behavior of polar ice which exhibits grain growth. Grain boundary migration associated with grain growth appears to be an efficient accomodation process for grain boundary sliding and dislocation glide. For grain growth to occur, strain energy must always be lower than the free energy of boundaries. The sintering of ice particles in polar firn is energized by the pressure due to the overburden of snow. Dislocation creep must be taken into account to explain the densification rate in the intermediate and final stage Constants of power law creep should depend on the crystal growth rate.

Pressure-sintering experiments of ice particles were carried out at pressures between 0.1 and 2.0 MPa and temperatures between -25 and -9°C. At each experiment, logarithm of the strain rate of densification was proportional to the density, and the slope of proportionality changed at a critical density implying the alternation of predominant densification mechanisms. The critical density varied with applied pressure and temperature, but its variation was restricted between 700 and 750 kg/m3.

At densities below the critical density, two-dimensional coordination numbers of constituent particles increased though specific areas of internal free surface were almost constant. The result was explained by the predominant contribution of the process of particle rearrangement in this density range. On the other hand, the contribution of dislocation creep was concluded to be dominant at densities above the critical density; the creep caused by boundary diffusion also contributes especially at higher densities.