Ping-pong ball avalanche experiments have been carried out for the last 3 years at the Miyanomori ski jump in Sapporo, Japan, to study three-dimensional granular flows. Up to 550 000 balls were released near the top of the landing slope. The balls then flowed past video cameras positioned close to the flow, which measured individual ball velocities in three dimensions, and air-pressure tubes at different heights. The flows developed a complicated three-dimensional structure with a distinct head and tail, lobes and “eyes”. “Eyes” have been observed in laboratory granular flow experiments, and the other features are similar not only to snow avalanches” but also to other large-scale geophysical flows. The velocities attained showed a remarkable increase with the number of released balls. A power law for this relation is derived by similarity arguments. The air-pressure data are used to deduce the structure of the airflow around the avalanche and, in conjunction with the kinetic theory of granular matter, to estimate the balance of forces in the avalanche head.

We developed a method that predicts avalanche paths deviating from a stream on complex surfaces. This method is based on Nohguchi’s center-of-mass model with simple dynamic resistance. Avalanche paths from many starting points were calculated at two sites in Japan where avalanche disasters have occurred, and paths causing unexpected disaster are presented. The characteristics of avalanche paths at these sites are shown. Small differences of the release positions sometimes cause great changes in avalanche meanderings.

A probabilistic approach to modelling avalanche origin and dynamics is developed. An avalanche process is considered as a single event. As a basis, the physically established deterministic models describing the processes of avalanche origin and motion are used. On the other hand, input data and/or parameters of these models are treated as random variables. The study is restricted only with respect to the dynamic stage of the avalanche process. It is assumed that the initial volume of snow, entrained into the motion, and the coefficients of dry and turbulent friction of the avalanche body are random variables. The other input data are deterministic. Three kinds of distribution laws for random variables are examined: uniform, normal and exponential. Several hundred numerical tests, carried out for the selected avalanche path No. 22 in the Khibiny region, Russia, allowed construction of the distribution functions for the output parameters of the dynamic model. These parameters are the front velocity, the height and the volume of the avalanche body at fixed points of the avalanche path. No very strong dependence of distribution functions on the kind of distribution laws was found. The model and empirical distribution functions are very close to one another for run-out distances and sufficiently close for depths and volumes of snow deposits.

In this work, we study the size of the particles involved in a powder-snow avalanche. To determine this value, we study all the phenomena encountered by the particles before they arrive in the “body” of the avalanche. We study the boundary layer which is at the bottom of the avalanche. We determine, with the help of experimental data, the range of sizes of particles that can be entrained by the avalanche. We then examine the possibility of these particles reaching the top of the boundary layer and thus taking part in the avalanche. Our final result is that the typical size of particles suspended in a powder-snow avalanche is < 200 μm.

The Savage-Hutter equations of granular avalanche flows are a hyperbolic system of equations for the distribution of depth and depth-averaged velocity components tangential to the sliding bed. They involve two phenomenological parameters, the internal and the bed friction angles, which together define the earth pressure coefficient which assumes different values depending upon whether the flow is either diverging or contracting. Because of the hyperbolicity of the equations, since velocities may be supercritical, shock waves are often formed in avalanche flows. Numerical schemes solving these free surface flows must cope with smooth as well as non-smooth solutions. In this paper the Savage-Hutter equations in conservative form are solved with a shock-capturing technique, including a front-tracking method. This method can perform for parabolic similarity solutions for which the Lagrangian scheme is excellent, and it is even better in other situations when the latter fails.

Seismic signals from artificially released avalanches were studied in an attempt to characterize them for avalanche-monitoring purposes. The seismic signals generated by different sizes and types of avalanches were recorded and analyzed in the time and frequency domains. Synchronized recordings of the corresponding seismic signals and the video images of the evolution of the avalanches were obtained together with a detailed cartography. Characteristic signatures in the frequency and time domains were found to depend on the characteristics of the avalanche path and measuring location, but to be mostly independent of avalanche size. The source of the different parts of the recorded seismic signals was determined. A relationship was observed between the avalanche size and the amplitude of the signals. Given the presence of local site effects, a prior seismic characterization of the avalanche path in relation to the recording sites is necessary for monitoring purposes. Moreover” it was found that sliding slabs in the early phase of acceleration produce little seismic energy, resulting in a time lapse between the observable start of the avalanche and the arrival of the detectable seismic waves at the receiving station.

This paper presents the application of a pulsed Doppler radar for the measurement of dynamic properties of snow and ice avalanches. The instrument was developed and built by the Institute of Communications and Wave Propagation, Technical University Graz, and is operated by the Austrian Institute for Avalanche and Torrent Research (AIATR), Innsbruck. The data acquired by this radar during field campaigns are used to verify and optimize avalanche models and simulation tools. As well as describing the radar’s technical principle, how the dynamic parameters are measured and presented to the user and how these data can be used to accomplish the determination of avalanche velocity and runout distance, the paper deals with the participation of the radar in a successful full-scale snow-avalanche experiment in Ryggfonn, Norway, and evaluates the collected data. This measurement campaign was a joint experiment by the Norwegian Geotechnical Institute, Oslo, and AIATR.

One means of preventing areas from being hit by avalanches is to divert the flow by straight or curved walls or tetrahedral or cylindrical-type structures. Thus, there arises the question how a given avalanche flow is changed regarding the diverted-flow depth and flow direction. In this paper a report is given on laboratory experiments performed for gravity-driven dense granular flows down an inclined plane obstructed by plane wall and tetrahedral wedge. It was observed that these flows are accompanied by shocks induced by the presence of the obstacles. These give rise to a transition from super-to subcritical flow of the granular avalanche, associated with depth and velocity changes. It is demonstrated that with an appropriate shock-capturing integration technique for the Savage-Hutter theory, the shock formation for a finite-mass granular flow sliding from an inclined plane into a horizontal run-out zone is well described, as is the shock formation of the granular flow on either side of a tetrahedral protection structure.

We show that two different types of dry-snow slab avalanches may result from a tensile failure of the slab crown triggered by either a quasi-static or an unstable shear propagation of a basal crack at the slab/substrate interface The triggering conditions for both modes are derived analytically. The latter only takes place if the depth of the top layer is above a critical value, and in a “window” centred around a critical slope of 35.3°, if residual friction on the basal crack surfaces is neglected. The width of this window increases with snow thickness. The introduction of a residual friction on the basal crack surfaces increases the critical slope.

The reduced societal acceptance of living in regions exposed to snow avalanches, and the increased economic consequences when houses are located within a hazard zone, highlight the uncertainty concerning avalanche run-out prediction. The limitations of today’s zoning procedures are especially pronounced in potential avalanche terrain where there are few observations of snow avalanches, where old buildings are present in the potential run-out zone, and where the local climate does not favour severe snow accumulation. This paper combines a mechanical probabilistic model for avalanche release with a statistical/topographical model for avalanche run-out distance to obtain the unconditional probability of extreme run-out distance. For the mechanical model, a first-order reliability method (FORM) and Monte Carlo simulations are compared. The interpretation of the statistical/topographical model either as an extreme value model or as a single value model is discussed. Furthermore, both a classical approach where the probability of an avalanche occurring is a constant, and a Bayesian approach with stochastic probability, are compared. Finally, example applications in hazard zoning are presented, with emphasis on how the influence of historical observations, local climate, etc., on run-out distance can be quantified in statistical terms and how a specified certainty level can be found from constructing confidence intervals for, for example, the most likely largest run-out distance during various time intervals.

The present work addresses the urgent demand for methods of quantifying the uncertainties inherent in the current procedures for avalanche hazard assessment. A Monte Carlo approach to hazard mapping is proposed for this purpose. This statistical sampling-analysis method allows us to evaluate the probability distributions of the relevant variables for avalanche hazard assessment –– essentially runout distance and impact pressure-once the release variables and the model parameters are expressed in terms of suitable probability distributions. In this way it is possible to explicitly account for uncertainties both in the input-data definition of the dynamic models and in the mapping results. The overall methodology is presented in detail and applied to a real-world avalanche mapping problem. The one-dimensional version of the VARA models is used for avalanche dynamics simulations.

In this paper we present a new methodology for the study of drifting-snow phenomena at an Alpine site This method emphasizes the accurate determination of local wind directions and the distribution of snow denudation and accumulation zones through the observation, measurement and mapping of aeolian “snowforms” and drifts. The aim is to find out whether the local wind directions on extensive surfaces in mountain areas (of about a few km2) are independent of the main wind intensity and direction, and whether valid erosion and accumulation zones are spatially stable and can be established for dominant winds. The methodology proposed to study the spatial distribution of snow by the wind has been used in the redesign of an old ski slope at La Molina, a resort in the eastern Spanish Pyrenees. Results have been very useful for deciding the design of the slopes. Therefore, we believe that the prediction of spatial snow distribution by wind in mountain areas, without the need for expensive instruments, has a valuable part to play in the management of mountain and ski resorts and roads, and in avalanche forecasting.

In the mountains of the moist-subtropical zone of Western Transcaucasia seven zones of snowiness are distinguished, depending on the heights from sea level to the tops of mountain ranges. Data are taken from 17 years of snow-avalanche observations. In a legend of a map the quantitative characteristic of the avalanching conditions for each zone is given. Interannual variability of snow depth is up to 100 cm in low mountains and about 600 cm in the middle belt of mountains The inversion of snow accumulation is registered. The frequency cycles of heavy-snow (3–4 years) and of especially heavy-snow (10–12 years) winters are determined. The avalanche danger is provoked not by the total depth of the snow, but by the stormy snowfalls. The particular conditions of creation and development of nival processes are distinguished here: dry avalanches on the crests of mountains, wet avalanches in the middle mountains and slushflows in the low mountains. Despite low avalanche activity, avalanche risk is high due to the factor of unexpectedness.

In recent years, most avalanche fatalities have been due to dry-snow slab avalanches triggered by the victims themselves during recreational activities, at least in countries where recreational skiing takes place, as in Western Europe and North America. Simple analysis suggests, and previous measurements have shown, that the skier’s dynamic impact is relevant for triggering, decreasing with increasing depth within the snow cover and depending on the layering of the snow cover. The stress distribution below a skier in the snow cover is not known, but is important in view of the critical area required for fracture propagation preceding dry-snow slab avalanche release. The skier’s zone of influence was measured with load cells buried within the snow cover in a level study plot for different depths. The size of the area of influence is of the order of a few 0.1 m2, but also depends on the weak-layer strength. Stresses in the snow cover below skiers who walk behind each other are not cumulative. The effect on the influence size is in accordance with the independent estimate for the critical size (0.1–1 m) for fracture propagation in the case of rapid loading. It supports the hypothesis that skiers induce a brittle fracture within the weak layer or at a weak interface between layers, leading under certain conditions to slab release, without necessarily hitting a pre-existing flaw.

In winter 1998/99, high-frequency pressure measurements with 10 cm sensors mounted 1–19 m above ground were carried out in the upper run-out zone of the avalanche test site at Vallée de la Sionne, Switzerland. Two large dry-snow avalanches clearly revealed a three-layered structure, with surprisingly low pressures in the suspension (or powder-snow) layer. The height of the saltation layer varied between 1 and > 3 m. From the duration, impulse and frequency of single-particle impacts (observed in the saltation layer and intermittently in the dense flow),particle-size and velocity distribution functions as well as strongly varying saltation-layer densities were found. With improved methods for peak detection and correction for grazing impacts, pressure measurements will become premier tool for testing granular flow models.

In Switzerland three large snowfall periods led to numerous large avalanche events, causing the deaths of 17 people, damage to about 1000 buildings and blockage of many important traffic systems. This situation was a valuable test of the avalanche-hazard mapping procedure developed in Switzerland since the last extreme avalanche winter in 1951. In most cases, the avalanches stopped within the avalanche-hazard zones. However, approximately 40 of about 1200 large avalanche events passed the limits of the existing hazard zones. The primary deficiency was underestimation of the hazard of powder-snow avalanches. The other reason the hazard-zone boundaries were passed was multiple avalanche events in a single track within a short time period. In addition, some fracture depths were larger than those used in avalanche-dynamics calculations. Therefore, the guideline assumptions for defining the fracture depths are questioned. Winter 1999 impressively underlined the danger of avalanches in mountainous regions. It emphasized the importance of avalanche-hazard mapping, showed the existing deficiencies and provided a large amount of field data that will help to improve our knowledge of avalanche dynamics.

At 1455 h on 18 January 1997, an airborne powder avalanche with a volume of about 3 × 106 m3 flowed along the tongue of Brenva glacier, high Aosta valley, on the southern flank of Val Veny, Italy It descended 2000 m and covered a distance of 5.5 km, with a rate of movement on the intermediate stretch of > 70 m s–1. It killed two skiers following a valley-bottom piste, damaged a hotel, destroyed a wide belt of woodland on the opposite slope and caused other minor damage. The avalanche occurred immediately after a large rockfall on the southern slope of the Sperone della Brenva (3567 m a.s.L). However, the relationship between the two events can only be considered indirect, since the main mass of the rockfall stopped at the base of the scar, on a large plateau on the glacier surface. The rockfall and subsequent shock caused the contemporaneous detachment of masses of ice and/or firn from hanging glaciers at the head of a nearby cirque, leading to the formation of the avalanche. The complex mechanism of detachment hypothesized was induced from available documentation and from in situ investigations. Its phases also coincide with seismograms recorded at a station 45 km from the avalanche area. Finally, a mathematical model of the avalanche was used to reconstruct its dynamics and path.

SAFRAN and Crocus are simulation systems developed at the Centre d’Etudes de la Neige, the snow research department of Météo-France. SAFRAN analyzes weather parameters relevant to snow on the ground, and Crocus simulates the build-up of the snowpack. In this study simulated snow depths and measured snow depths at three locations in Iceland were compared. The main study was performed for the Hveravellir weather station located 640 m a.s.l. in central Iceland. The results from three winters, 1994–97, were analyzed. In Iceland the “Alpine” version of the models systematically underestimated the snow depth and density of the snowpack. Corrections for the effect of wind on snow density and on precipitation measurements led to significant improvements in simulated snow depths. The simulations are sensitive to the threshold temperature between snow and rain. The remaining problems in simulating the snowpack are mainly related to transport of snow by wind, which is not accounted for in the models, and to some extent to melting and sublimation in strong winds.

Snow avalanches threaten a large part of the residential area of Bolungarvik, a town of 1100 inhabitants located in the Vestfirðir peninsula, northwest Iceland. Methods for the complete protection of all the houses were requested by the Ministry of Environment. As conventional deflecting and catching dams were not feasible due to avalanche velocities of > 40 m s–1 in the actual defence area, a huge open pit built into the lower mountainside was proposed. The pit is designed to dissipate the energy of avalanches of any expected size. It is 1000 m long, and the upper and lower pit walls are 30–40 and 15–20 m high, respectively. The downhill side of the pit is increased to 30 m by a nearly vertical, 10–15 m high dam. The bottom is 25–40 m wide, and the effective cross-sections are 700–2300 m2. The total storage capacity is approximately 106 m3. Cost-benefit analyses indicate that the development cost of the pit protection is favourable compared to the value of the houses as well as alternative safety solutions.

Two deflecting dams were constructed above the village Flateyri, northwestern Iceland, after the catastrophic avalanche accident on 26 October 1995 when 20 people were killed. Both deflectors have since been hit by moderately large avalanches, in each case with a volume of > 100 000 m3 and an estimated return period of 10–30 years. The avalanches hit the deflectors with deflecting angles of 20–25° and a maximum run-up of 12–13 m. They flowed about 500 m along the deflectors after the initial impact and terminated in the ocean on the respective sides of the reef where the village is situated. Back-calculated impact velocities are of the order of 30 m s–1. In both cases, the impact channelized a part of or The whole of the width of the avalanche into a stream 20–80 m wide, the run-out of which is estimated to have been increased by > 100 m by the deflection of the avalanche by the dam. Although the avalanches are much smaller than the design avalanches of the deflecting dams, they have provided avalanche professionals and the public in Iceland with much-welcomed direct evidence of the effectiveness of the defence structures at Flateyri against moderately sized events. They also provide unique direct observations for further scientific investigations of the deflection of avalanches by man-made deflecting dams.