Skip to content

Cookies 🍪

This site uses cookies that need consent.

Learn more

Zur Powderguide-Startseite Zur Powderguide-Startseite
news

World of Science | Review ISSW2018: Session Snow and Avalanche Dynamics

What's happening in snow science?

by Anselm Köhler 12/12/2019
Every two years, the International Snow Science Workshop (ISSW) brings together scientists and practitioners from a wide range of different, but always snow-related, subject areas. New findings and research results are presented in various thematic blocks - so-called sessions. We break the whole thing down into more or less digestible morsels and summarize the sessions of ISSW2018 for you every two weeks. This time: Snow and Avalanche Dynamics (Session 1).

The first session of ISSW2018 deals with the topic of avalanche dynamics, which examines the flow behavior of avalanches in the broadest sense. Current research is multi-pronged - some scientists study large avalanches experimentally, others try to simulate and measure effects in the laboratory, and others develop and use computer models. Nowadays, computer models are becoming increasingly important and so it is not surprising who got the very first talk at the conference:

What do avalanche dynamics have to do with Disney's animation studios?

For some time now, a win-win situation has been inspiring snow science. During a stay in Los Angeles, the young professor and professional snowboarder Johan Gaume developed a snow model together with Disney that not only looks good but also works exceptionally well (O1.1). If you want to know what it looks like, you can visit the current movie "The Ice Queen 2". If you want to follow the scientific scope of the model, we recommend Johan's Twitter channel or the website of the SLAB (snow and avalanche simulation laboratory) at EPFL in Lausanne.

The fascinating thing about the model is that it simulates the properties of snow and all applications such as fracture mechanics and flow dynamics arise from this. Technically, the model is based on the "Material Points Method" (short explanatory video). Instead of being based on a rigid computational grid, "Material Points" (quasi individual snow grains or granules/snowballs) are defined, which carry properties such as mass, momentum and deformation. This enables simulation over several orders of magnitude - i.e. from the weak layer in the centimeter range to the flow of the avalanche in the hundred-meter range. Furthermore, it is precisely this simulation method that is suitable for modeling transitions from solid mechanics to flow dynamics: The stationary snowpack behaves like a deformable solid, but the avalanche behaves more like a granular fluid.

So far, the model has mainly been validated in small-scale experiments for fracture propagation in the weak layer. More precisely, Propagation Saw Tests (PST) were compared - in the PST, the weak layer is sawn into an elongated block and it is seen from which saw width (critical crack length) the weak layer continues to break independently. The model not only manages to reproduce the critical crack length, but also cases in which the snow slab itself breaks off above it instead of the weak layer breaking completely. What is still missing is the validation of the model with the dynamics of the complete avalanche sequence - with velocities, runout lengths, pressure values, snow recordings, granular composition, and so on.

Who measures, measures crap, who models loses

Such a validation - also known as back calculation - works in principle independently of the underlying model. The existing avalanche data is taken as input and an attempt is made to recreate this with the model. To do this, all possible model parameters are varied (crack volume, existing snow in the avalanche path and its "absorption capacity", friction parameters ... ) and thus adapt the simulation to the data. The difficult thing is that there are a few model parameters that cannot be measured in reality, but are factors in the model, such as the parameterization of the absorption capacity of the existing snowpack.

The friction values cannot be determined directly in large-scale, i.e. natural avalanches, but have a very central influence on the runout length, for example. It is therefore important to know at least approximate friction values in order to determine the runout lengths of avalanches before the actual runout. Such analyses have been carried out frequently in the past; here in the current session, two papers deal with the friction coefficients of ice avalanches generated by large glacier avalanches. In contrast to snow avalanches, ice avalanches have less friction, flow faster and therefore advance further into the valleys (P1.11).

In addition to the aforementioned statistics through model validation based on many avalanches, probabilistics is another way of approaching the uncertainty of the model parameters. On the one hand, the "poor man's ensemble" can be carried out by applying different flow models; or the sensitivity of a model parameter to the result is considered as uncertainty in the results themselves. Accordingly, the article O1.4 defines the so-called "runout gradient", i.e. the change in a parameter that increases the simulated runout length by 100m. Roughly speaking, doubling the crack width or reducing the friction by 5% increases the runout length by 100m. Such a probabilistic concept thus makes it possible to compare the different errors of the input parameters and to transfer the uncertainty of the parameters to the simulation result.

For more information on avalanche modeling and back calculation of individual avalanches, we recommend the article O1.2 about the avalanche accident in Rigopiano (central Italy), in which an entire hotel was destroyed.

Continue on the next page -->

15 kg of explosives, 100,000 m³ of snow and lots of data

Experimental avalanche dynamics is absolutely fantastic. It is one of the few experimental geophysical disciplines where it still makes a real impact - and when the media are involved, there are not only spectacular images but also a complete car in the avalanche. Of course, the car was not only equipped with cameras, but also with acceleration and rotation sensors to measure the interaction with the avalanche.

Apropos flowing measurements: Two papers describe 3D-printed balls, which are equipped with motion and positioning sensors to record conditions in the avalanche. In principle, avalanche research overlaps here with research on rockfalls, where similar sensors are located in the stones. Individual colored stones can be located quite easily using camera images, but in an avalanche it is more difficult and only the acceleration directions remain to calculate the position of the balls back (P1.2). In response, contribution P1.3 is developing a positioning system based on WLAN, which may be supplemented by radar tracking in the near future.

Large avalanche test sites equipped with many different sensors can be counted on the fingers of one hand. This makes the article P1.8, which reports on a new site in Niseko, Japan, all the more pleasing. In other words, right in the well-known powder ski resort in Hokkaido - measuring and powder skiing at the same time is also an absolute advantage for avalanche researchers. Although the test slopes only have a height difference of around 200m, avalanches that would otherwise be labeled as a "dynamically uninteresting little skier's slide" can be investigated. That's nonsense, of course, because avalanches like this are on the border between flow and dust avalanches. Incidentally, virtually nothing is known about this transition.

Of course, there are also contributions from the most prominent avalanche test site "Vallée de la Sionne" in Valais. The two contributions deal with pressure measurements on a 20m high steel pylon, which stands in the avalanche path and is directly surrounded by water. Article O1.5 deals with the pressure and bending moment due to different flow regimes on the steel pylon. The authors find that dust avalanches mainly cause impulsive, short-term and very high dynamic pressures in the range of up to 1000kPa. In contrast, slow wet snow avalanches cause continuous pressures of around 400kPa due to their high density. As a comparison for the avalanche danger zones, the boundary from the red to the yellow zone is around 10kPa, and the boundary to the unzoned areas is 1kPa. More descriptive: a water column of 1m height causes about 10kPa static pressure.

The measured avalanche pressure is of course dependent on the geometry of the measuring device, so the ground plan of the steel pylon is decisive here. The measured pressures can only be transferred to other objects and geometries under certain assumptions, perhaps most likely to the masts of ski lifts. The second article (P1.5) from the Swiss test site reports on the development of a discrete element model that is optimized precisely for the investigation of pressures on general geometries, their drag coefficient and amplification factors. The velocity is linked to the pressure by the drag coefficient. The amplification factor comes from the piling up and accumulation of snow in front of the object, which virtually increases the area of attack for the avalanche.

Small but powerful...

In addition to such measurements on real avalanches, there are always measurements on controlled laboratory scales. The advantage of the laboratory and cold chamber is the good control of external conditions. The disadvantage, however, is that the results can only be transferred to real scales to a limited extent. Experiments on friction values or run-out lengths on small chutes are particularly difficult to compare, as certain flow structures only form with sufficient volume and flow length. For example, there is a small dust avalanche model in the exhibition at the Institute for Snow and Avalanche Research in Davos, which only works because the glass beads flow in a container filled with water and therefore the flow energy, density differences and drop heights are relatively similar to the real counterpart.

Despite these scaling difficulties, two contributions have ventured into the laboratory measurement of friction and run-out length. Contribution P1.17 lets different quantities of glass beads flow down a chute and finds longer run-out lengths with increasing volume. Contribution P1.15 uses a rotating drum instead of a chute and thus creates an infinitely long flowing avalanche. Snow of different grain sizes actually moves in it, but only marginally different friction coefficients can be determined.

Another laboratory article repeats a quasi-scientifically correct experiment from 3 years ago, in which snow of different temperatures was shoveled into a concrete mixer to observe the formation of granular snowballs. It was found that at temperatures warmer than -2°C, the snow quickly glopped together from fine-grained snow to fist-sized ballsTM - trivial for Bavarian primary school children, complex result for academics. This contribution P1.16 also uses a rotating drum with a diameter of 2.5m. The improvement over the concrete mixer is speed control and the direct measurement of temperature, liquid water content and flow profile. The limit value of -2°C to -1°C, at which the grain size changes quite abruptly, was also determined.

Wet snow vs. warm snow

After the concrete mixer experiments, the snow temperature was focused on by various avalanche dynamics experts. In contrast to liquid water content, temperature is a classically measurable variable that can be simulated quite well with snowpack models and can also be roughly estimated from historical data. The article P1.14 reports on a possible parameterization of snow temperature in avalanche models. The article on the hotel accident in Rigopiano mentioned above also comes to the conclusion that warm snow in the lower part of the avalanche path was decisive for the long run-out length. Even if the term wet snow avalanche may not be optimal, I personally find the term warm snow avalanche equally weak.

Conclusion

It is always difficult to show the current state of research in a field from contributions to a conference. A lot is certainly happening in avalanche dynamics at the moment, and it will be exciting to see what avalanche science will look like in 10 years' time. Outstanding innovations at ISSW 2018 were definitely the Disney model, the flowing sensor nodes and the fact that several contributions now talk about warm instead of wet snow.

Photo gallery

This article has been automatically translated by DeepL with subsequent editing. If you notice any spelling or grammatical errors or if the translation has lost its meaning, please write an e-mail to the editors.

Show original (German)

Related articles

Comments