In Franee, drifting snow is generally controlled using artificial snow fences. Living snow fences are not a new concept but they have only been used on a limited scale. Research directly related to natural plant barriers is limited. We therefore decided to study the behaviour of species that would survive and grow satisfactorily in the French Alps.

In the first experiment, we compared the storage capacity of several different kinds of living fences consisting of pruned spruces, unpruned spruces and sorbs.

Field observations during the winter of 1995-96 proved that deciduous trees such as sorbs are effective, and that pruning the lower 50 cm is not effective at the end of the season because of the weight of snow on low branches.

However, the use of natural plant barriers has disadvantages: a living snow fence takes time to reach an effective height and is difficult to establish on windy sites at a high altitude. This is a real problem. Therefore, in a second experiment, we studied the death rate of larches planted behind a fence. We noticed that the snow fence had several effects, snow accumulation (until the planted trees grew up) and protection of the planted trees.

A new gauge to detect drifting snow has been tested at an experimental site at La Molina, eastern Spanish Pyrenees during the 1996 winter.

Based on an acoustic principle, it consiste of a miniature microphone located at the base of an aluminium pole of a Variable length (depending on the average height of the snow accumulated in a season). While snowdrifting occurs and the pole is exposed to the flux (wind + snow particles), the snow grains impact on the pole and the sound produced by these collisions is recorded as an electrical signal.

30 m away from the snowdrifting detector, there is an anemometric station. The comparison of both records allows determination of thresholds and deposit wind velocities. In addition, this gauge allows users to know, during a wind episode, the exact amount of time during which snowdrifting occurs without being in the field. The interpretation of the values obtained using the acoustic detector is not yet an easy subject to discuss. We slill do not know what the drift-density equivalents for tile recorded noise levels are. However, these are only preliminary results and we expect to complete the calibration of the gauge during the coming months.

In experimental snowdrifting mass-flux measurements many different instruments have been tested (Takeuchi, CEMAGREE Mases, etc.). Very often the results obtained are a function of the gauge used. However, in order to compare data from different instruments, orders of magnitude have to be similar.

Since 1992, snowdrifting has been studied at an experimental plot at La Molina ski resort (eastern Spanish Pyrenees). The alpine site, characterized by a plateau topography, is located at 2250 m.

In this paper, different gauges used to measure snowdrifting mass flux at this site are presented: one snow-collector column and two types of snow traps. Snow-collector columns (prismatic boxes) are permanent installations and are used to measure the mass-flux episode. Snow traps (Takeuchi, 1980: modified) are lighter and more mobile, and they are used for short experiments during a wind episode during which mass-flux data are obtained.

The three different gauges are compared and the rate of trapping efficiency is suggested from a comparison of the field data with estimated mass-flux data deduced from empirical formulae (Mellor and Fellers, 1986; Naaim-Bouvet and others, 1996). The mass-flux values obtained at the experimental site are lower than the estimated values.

During the construction phase of the Pegasus runway on the McMurdo Ice Shelf, relatively large amounts of snow and ice were cleared to meet basic grade requirements for the runway surface. A considerable amount of material remains adjacent to the runway in two north–south extending mounds (berms). The runway was originally constructed on an experimental basis so attention was not focused on developing and executing a snow-removal/accumulation plan. After the runway was successfully constructed and supporting routine flight operations, concern developed over the possibility of snow accumulation adjacent to the berm area eventually inundating the runway. The intent of this project was to analyze snow accumulation and to recommend passive methods for removing some of the berm material and snow adjacent to the berm. We found that large quantities of excess snow could be removed by use of vortex fences which cause erosion on the leeward side of the fence. The vortex fence was designed to be portable (unlike traditional jet or blower fences) and self-orienting into the wind to allow snow removal regardless of the wind direction. The vortices generated by the fence do not dissipate rapidly, providing effective and sustained erosion.

Field data from Stillberg (Switzerland), provided by SFISAR, have been analysed in order to understand the spatial distribution of snow transported by wind in avalanche-starting zones. The results have been used to develop a new empirical model for snowdrift distribution. This model constitutes the “snow-wind module” of the knowledge-based computer system called ELSA.

The model is based on an empirical parameter called the wind or “aeolien” coefficient. It represents the relation between two snow heights at a given place: the snow accumulated at the end of a wind period (or wind episode) and the snow available for transport at the beginning of the wind period. The wind coefficient has been established using field data from the Stillberg site.

The problem of physically modelling snow avalanches with an aerosol cloud is considered. Using the simplest model of avalanche-body motion, the criteria of similarity for transit and deposit zones are written. The transit relations between model and natural scales for dynamic parameters of the avalanche body are established. On the basis of the similarity criteria, an approximate physical model of avalanches with an aerosol cloud is made using a mixture of ferromagnetic sawdust and aluminium dust. The setup allows us to start and stop the model avalanche during any stage of its motion and to model actual avalanches. A series of experiments was completed. It is shown that the basic mechanism of aerosol cloud generation for model avalanches is related to the hydraulic jump of fluid when a stormy flow changes to a calm one. A comparison of parameters for model and “home” avalanches (the vicinity of Elbrus) showed that the ratios of the sizes of dense and aerosol components and deposition zones of those avalanches are close to the geometric scale factor and the ratios of velocities, and durations of processes are close to the square root of this coefficient. The latter corresponds to modelling the process by Froude number.

A small avalanche path near the Bridger Bowl ski area in southwestern Montana has been instrumented to measure density, velocity and dynamic friction in a flowing avalanche. These measurements, made by an array of sensors mounted in the avalanche path, have been carried out for several dry-snow avalanches. Measurements of density were made using a capacitance probe that measures the dielectric constant of any material that passes in front of it. Through a calibration procedure, the dielectric constant of a given type of snow can be related to the density of that snow. Optical sensors were used to measure light reflected from the avalanche as it passed by the sensors. Signals from adjacent optical sensors were cross-correlated to determine velocity. Density and velocity measurements were made at several heights in the avalanche, with particular attention directed near the running surface. Results indicate that avalanche deformation is concentrated near the running surface where the snow density is found to be largest. Upward from the surface, the velocity gradient falls off greatly while the density also declines.

Finally, the dynamic-friction coefficient at the base of the avalanche was found by measuring shear and normal forces on a roughened 23 cm × 28 cm aluminum plate mounted parallel and flush with the avalanche running surface. The ratio of the shear force to normal force on the plate provides a measure of the dynamic-friction coefficient at the base of the avalanche.

Following Norem’s description of powder-snow avalanche formation and structure, we propose a mathematical model that consists of a suspension layer and a so-called saltation layer. The latter is only a few meters deep and is modelled by depth-averaged mass and momentum balances. In the suspension layer, the mass and momentum balance equations for the mixture are supplemented by the snow mass balance and the transport equations for turbulent kinetic energy and dissipation. Mass and momentum exchange between the two layers is determined by particle settling, turbulent diffusion against the concentration gradient and aerodynamic shear forces. The net erosion or deposition rate is a function of the kinetic energy of the impacting particles. The saltation layer reacts on the suspension layer in that saltating particles extract momentum from the air flow. The preliminary estimates of the model parameters can be refined by means of saltation-trajectory simulations. Three-dimensional simulations with a simplified model have clearly shown the importance of snow erosion and deposition in practical applications. This approach is well suited for coupling to a dense-flow avalanche model.

Dense-flow avalanches may be simulated as granular flows. In this work, large-scale experiments on ping-pong-ball avalanches are investigated. In one situation, channeled flows with up to 6000 balls are produced. Another set-up is the open, three-dimensional flow over the landing track of a ski-jump field, using up to 300 000 ping-pong balls. The measurements focus on the internal velocity distribution, of the balls as well as of the air. A video camera positioned above the flow allows the measurement of the location and distance of a single ball, which finally leads to its velocity. Different profiles in the head and tail as well as other flow properties are obtained. With vertically placed lubes, in and above the flow, the open ends pointing towards the ground, the static pressure depression induced by the air flow is measured and leads to qualitative air-velocity profiles in and above the tail of the ping-pong-ball avalanches.

Two models simulating snow avalanches impacting retaining dams at oblique angles of incidence are presented.

First, a lumped-mass model applying the Voellmy-Perla equation is used to calculate the path of the centre-of-mass along the side of a retaining dam.

Secondly, a one-dimensional continuum model, applying depth-integrated equations of balance of mass and linear momentum, is expanded to take into account that real avalanche flows are three-dimensional. The centre-line of the avalanche path is determined by the flowing material as it progresses down the channelized avalanche path. The nonlinear constitutive equations comprise viscosity, visco-elasticity and plasticity.

Both models are calibrated by simulations of a registered avalanche following a strongly curved channel. The path and the run-up height of the avalanche on the natural deflecting dam with oblique angle of incidence as calculated by the two models, are compared with the observations made.

The Savage-Hutter theory for granular avalanches assumes that the granular material is in either of two limiting stress states, depending on whether the motion is convergent or divergent. At transitions between convergent and divergent regions, a jump in stress occurs, which necessarily implies that there is a jump in the avalanche velocity and/or its thickness. In this paper, a regularizaron scheme is used, which smoothly switches from one stress state to the other, and avoids the generation of such singular surfaces. The resulting algorithm is more stable than previous numerical methods but shocks can still occur during rapid convergence in the run-out zone. Results are presented from two-dimensional calculations on complex geometry which illustrate that some necking features observed in laboratory experiments can be explained by the regularized Savage-Hutter model.

The Savage-Hutter model is generalized by including a velocity-dependent drag in addition to the usual Coulomb dry friction at the base of the avalanche. Both linear and quadratic velocity dependencies are considered, with either constant or asymptotically constant drag coefficients for large thickness h. The singular nature of the constant coefficient model for small h is demonstrated and it is shown that the asymptotic model allows the tail of the avalanche to move at a finite velocity. The inclusion of velocity drag changes the stress state in the avalanche and new earth-pressure relations are derived and investigated.

The paper describes mathematical and physical models for powder-snow avalanches developed by a group of researchers at Moscow State University.

We present regression models to estimate maximum avalanche run-out distances in the western Catalan Pyrenees (northeast Spain). The regression equations have been calculated from topographic parameters of 216 well-known avalanche paths, mapped on avalanche location maps at a scale of 1: 50 000. Historical information about avalanche maximum run-out zones has been available for some avalanche paths hut most of them have been delimited by vegetation and geomorphological indices.

The avalanche profiles, which are representative of avalanche paths, have been classified on the basis of avalanche run-out zone morphology, because this has a considerable influence on avalanche behaviour and run-out. Three classes have been defined and regression equations obtained for each class and also for the total. Finally, some examples have been calculated in order to appreciate the applicability of the models to improving avalanche mapping.

In this paper, numerical and experimental approaches are applied to analyse the dynamics of the front of a gravity current. This study focused on two parameters: internal density and velocity fields. The salt concentration was determined by a potentiometric process. The internal velocities were determined using an optical device and an image-processing system. The structure of the head of the gravity current was analysed. Its density was measured and two stages of evolution were observed. This analysis allows us to coufirm the existence of two important stages. For xf < xs, where the dynamics depend on the initial condition, the flow consists of a head and body and the front density is constant. For xf > xs, we show that the density of the front decreases and evolves towards the Hallworth and others (1993) law. From a comparison between the experiments and the numerical model, we show that the numerical model, which is based on Navier–Stokes equations and on the k−L turbulence model (where L is the height of the gravity current), can predict well flow in the slump regime and in the inertia–buoyancy regime with smoothed results in the transition from the head to the body of the gravity current.

Stability indices for skier-triggering of slab avalanches are discussed in terms of an adjustment to the skier-induced stress to allow for ski penetration and an adjustment to shear strength (measured with a shear frame) to allow for normal load due to the slab overlying the weak layer. The proposed adjustment to shear strength depends on the microstructure of the weak layer. These adjustments are incorporated into a refinement of the previously established Swiss stability index for skier-triggering. The percentage of correct predictions for the Swiss and refined indices are evaluated using data from 115 weak layers on skier-tested avalanche slopes, 83 of which were classified as persistent weak layers of surface hoar, faceted crystals or depth hoar and 32 as non-persistent weak layers. The refined index reduces the number of incorrectly predicted slab avalanches for the persistent weak layers.

The results from a numerical- probabilistic experiment aimed at evaluating avalanche-release probabilities are presented in order to show the problems and the possibilities of such an approach. A snow slab on a mountain slope is modeled as a thin elastic non-moment shell resting on a three-dimensional hard underlying surface. The thickness of snow is considered as an inhomogeneous Gaussian spatial random field with a predetermined covariance that was obtained previously by field measurements. The stress distribution σ(x, y) is calculated by solving the equations of equilibrium using a finite-difference method with a mesh of lines of curvature on a mountain slope. As a criterion of avalanche release, exceeding some threshold value of stress σthres; by the calculated value σ(x, y) is applied. The field P(σ(x, y) > σthres; x, y) derived, using the Monte-Carlo method, is regarded as a measure of the spatial distribution of avalanche-release probability. This simulation scheme at present has no claim to forecast avalanche release but may be used to pinpoint the importance of snow distribution on a slope.

Measurements and observations by others indicate that a potential slab avalanche consists of a relatively cohesive slab of snow overlying a thin weak layer that coniains flaws where locally the shear stress from the overburden is not fully supported. Under favorable conditions, snow will shear strain-soften, which provides the basis for applying a slip-weakening model to examine the size of flaw needed to initiate sub-critical crack propagation along the weak layer. Using typical values for snow properties, the model predicts sub-critical crack growth can initiate from a relatively small flaw well before the shear stress from the overburden approaches the peak shear strength at tin-bed. The occurrence of small flaws or imperfections in the basal layer would explain field measurements which usually indicate that avalanching occurs before the applied shear stress exceeds the shear strength at the basal layer.

Widespread slab-avalanche activity often increases significantly soon after the onset of rain on new snow. Measurements of temperature and mechanical properties show that only the upper 0.15 m or less of the slab has been altered at the time of avalanching; alterations at the sliding layer have not yet been detected. Results from the slip-weakening model indicate that the rain-induced alterations would reduce the size of flaw needed to initiate sub-critical crack growth by 10–20%. The observations and model results show clearly the importance of the slab properties; it is evident that both the slab and the weak layer act together to control slope stability. A further implication is that the stability of freshly deposited snow is often close to critical, because a relatively small surface perturbation is often sufficient to cause avalanching. This is not surprising, because it is well known from field observations that new snow on slopes should be treated with caution.

A compact, operational prototype for automatic seismic detection of natural avalanches, named SARA, is proposed as a means to improve real-time estimation of natural avalanche activity. A software combining multi-component signal analysis and fuzzy logic has been set up to reject spurious signals automatically. The average success rate of our system, estimated from a set of previously identified signals, is somewhat over 90%. Two prototypes have been installed in the French Alps dur ing the winter of 1996–97. The SARA system enables the user to follow up the evolution of natural avalanche activity with a high time resolution. Results obtained in 1997 are presented and compared with other data related to avalanche activity.

Seismic signals of avalanches, related video images and numerical models were compared to improve the characterization of avalanche phenomena. Seismic data and video images from two artificially released avalanches were analysed to obtain more information about the origin of the signals. Image processing was used to compare the evolution of one avalanche front and the corresponding seismic signals. A numerical model was also used to simulate an avalanche flow in order to obtain mean- and maximum-velocity profiles. Prior to this, the simulated avalanche was verified using video images. The results indicate that the seismic signals recorded correspond to changes in avalanche type and path slope, interaction with obstacles and to phenomena associated with the stopping stage of the avalanche, suggesting that only part of the avalanche was recorded. These results account for the seismic signals previously obtained automatically in a wide avalanche area.