Skip to content

Cookies 🍪

This site uses cookies that need consent.

Learn more

Zur Powderguide-Startseite Zur Powderguide-Startseite
SnowFlurry

SnowFlurry 2 2018/19 | Interview with ski touring guru Günter Schmudlach

How did the automatic, daily updated risk assessment come about?

by Lukas Ruetz 11/26/2018
Günter Schmudlach is the inventor, founder and programmer of the Swiss online portal Skitourenguru.ch. The avalanche report assesses the risk of countless ski touring routes in Switzerland and soon in the entire Alpine region on a daily basis. A highly innovative project, in our opinion. We spoke to Günter about Skitourenguru in general, his motivation and his vision for the future.

For anyone who has never heard of Skitourenguru: What is Skitourenguru and how does the platform differ from a conventional slope gradient map?

With the help of a slope gradient map, the avalanche situation report (LLB) and a reduction method, I can make a statement about a point in the terrain, e.g. the key point of my planned ski tour. Skitourenguru, on the other hand, evaluates complete ski tours for their entire length on a daily basis. Each route is assigned a risk indicator. The risk indicator shows a low risk (green), increased risk (orange) or high risk (red), similar to a traffic light.

Simply explained: how does the algorithm behind Skitourenguru work?

Skiing avalanches are caused by an additional load on an unstable snowpack in suitable avalanche terrain. Classic reduction methods therefore combine the danger level from the LLB (snowpack) with the slope gradient (avalanche terrain). Skitourenguru goes three steps further. Firstly, Skitourenguru takes into account not only the slope gradient, but also the slope size, slope shape and ground cover. Secondly, Skitourenguru not only uses the danger level to describe the snowpack stability, but also the special danger points (exposures and altitude levels), as communicated by modern LLBs. To combine the factors "avalanche terrain" and "snowpack", Skitourenguru has developed the Quantitative Reduction Method (QRM). This method derives the necessary knowledge from 1500 avalanche accidents and a collection of GPS tracks as to how the factors must be combined. An avalanche trigger probability can now be assigned to each point on a route. By combining these probabilities along the route, its final risk indicator can be determined.

SnowFlurry
presented by

Where do the GPS tracks for the tours come from and how good is their quality?

It has been shown that Skitourenguru's algorithm only produces consistent results if the routes are digitized according to a standardized procedure. Routes that originate from conventional ski tour maps or web portals are not suitable due to their heterogeneity. For five years, I have been digitizing the necessary routes with the help of first-class data material in a geoinformation system. I have access to 1:10,000 scale maps, high-resolution aerial photographs, collections of GPS tracks and an avalanche terrain hazard map. When ski touring maps were still drawn, you could only dream of such possibilities.

It goes without saying that Skitourenguru's route collection also has inconsistencies. Thanks to lively feedback from the ski touring community, the routes are constantly being optimized. Ultimately, however, I would also like to point out that laying an ideal line is basically only possible in real terrain.

How did you come up with the idea for the project and where were the biggest hurdles?

For almost 40 years, I have been asking myself before weekends what would be suitable ski tours in view of the current LLB. To answer the question properly, you would have to meticulously compare the 300 day ski tours accessible from Zurich using a reduction method. This is rather tedious, repetitive work. As a software developer, I quickly realized that the computer could do this much better. That's why I started developing version 1.0 on New Year's Day 2013.

There are a number of technical hurdles. One major challenge is that the results have to be available a few minutes after the LLB is published. However, the most difficult thing was to find a good way of dealing with the occasionally superficial prejudices of the "avalanche experts".

In the meantime, Skitourenguru has developed into an alpine-wide platform. Which organizations and players are you now working with?

A partnership has existed for over three years with the Swiss Council for Accident Prevention (bfu) and with Mammut. In technical terms, there is a cooperation with Gipfelbuch.ch. Thanks to my involvement in the development of an avalanche hazard map at the SLF during 2017, there is also a close exchange with the Davos avalanche researchers. Last but not least, I am also in direct contact with the Swiss Alpine Club (SAC). After initial skepticism, the SAC now recommends Skitourenguru for route selection. Since the fall of 2018, Achtung Lawinen!, a leaflet published by the "core team for avalanche training" in Switzerland, also refers to Skitourenguru.

There are many skeptics. How do you respond to them and what are the most common arguments against Skitourenguru?

Yes, are there many skeptics? If that's the case, then very little criticism reaches me. Basically, my experience is that criticism is often based on false assumptions about Skitourenguru. It is not possible to grasp the diversity of the project with three clicks on the website. A well-founded judgment is only possible after you have spent a winter planning your tours with Skitourenguru and reading up on the project. Skitourenguru helps you to select suitable ski tours. These must then be prepared and carried out according to the 3x3 by W. Munter. Skitourenguru therefore does little to change the valid "avalanche doctrine".

There are two arguments that are repeatedly put forward. Firstly, it is claimed that the danger level should not be projected onto the individual slope due to its high level of uncertainty and generalization. You could translate this as follows: The data is bad, so we'd better throw it away. In the light of modern research on dealing with uncertainties, this argument is rather skewed in the landscape. Another fear relates to the legal consequences. However, the article Does Skitourenguru.ch change case law? clearly shows that such fears are unfounded. I also wonder what we are actually talking about. Is it about the interests of specific occupational groups or about accident prevention? I agree with W. Munter: "Fear the avalanche and not the courts".

Don't you think that some of the criticisms are also justified and if so, how did you incorporate the criticism into the project?

Skitourenguru enables beginners to quickly and easily arrive at a suggested route. As soon as the beginner then embarks on this ski tour in real winter terrain, they are confronted with the full complexity of avalanche assessment on site and on individual slopes. It is obvious that beginners are not up to these challenges, even if experts admit behind closed doors that there are often situations that they cannot assess adequately.

In the end, however, my main concern is not the beginners. In recent years, studies have become more frequent, concluding that experienced winter sports enthusiasts in particular expose themselves to high risks. The fact is that experienced winter sports enthusiasts are often involved in accidents. Skitourenguru is therefore primarily aimed at experienced winter sports enthusiasts and not primarily at beginners.

Skitourenguru prioritizes "green" ski tours with a low level of difficulty. Such routes also allow for the occasional mistake without the worst-case scenario occurring. However, the residual risk cannot be ruled out. It is therefore very important to me to use the website to make it clear to users what Skitourenguru can and cannot do.

SnowFlurry
presented by

The highly complex structure of snow avalanche science and the usually equally complex LLBs, which hardly anyone understands anymore who is not out and about on a daily basis, seem to be slowly being split into different tracks for different user groups. Why does this make sense?

In the end, there is a smooth transition between beginners, experienced winter sports enthusiasts and avalanche experts. Even avalanche experts are not immune to heuristic traps and cognitive traps. I am therefore not sure whether the user groups will split up or whether they should split up. Skitourenguru can offer great added value to experienced users in particular, because only experienced winter sports enthusiasts can make optimum use of all the information that Skitourenguru provides. In any case, I would like to see beginners embark on the adventure of "avalanche awareness" and get to grips with avalanches in practice and theory.

The Munter reduction method was celebrated as a breakthrough in practical avalanche awareness. Today we know that it and the methods derived from it are only used consistently enough in training, but hardly ever in independent practice by users of all levels. Is the actual breakthrough of linking slope steepness and danger level according to statistical probability only now coming through tools and automatisms such as Skitourenguru?

A recently published study from Norway shows that neither reduction methods nor other tools are widely used. In real-life touring practice, a great deal of work is done using intuition. Nobel Prize winner Daniel Kahneman has shown that intuition must systematically lead us astray when we are confronted with low event probabilities and weak hazard signs. In view of these new findings, it surprises me that probabilistic avalanche science is currently on the defensive. Admittedly, probabilistic avalanche science has two handicaps: on the one hand, it takes a few spikes out of the expert's crown and, on the other, it is not particularly user-friendly. If we leave the calculations to the computer and also develop more complex models, we can address both problems. In this sense, yes, I think developments like Skitourenguru can help probabilistic avalanche science make a breakthrough. Ultimately, however, it's about combining the three major trends in avalanche science (probabilistics, analytics and intuition) in a clever way.

A lot has happened in avalanche science for winter sports enthusiasts in recent times. Be it the now much more meaningful snow cover tests such as ECT and PST or avalanche problems. Basically, more has developed in the last ten years than in 100 years before. How does Skitourenguru incorporate the latest developments - in terms of technology and science? Does this make Skitourenguru the pinnacle of current developments in avalanche prevention?

The major weakness of Skitourenguru is the LLB. Avalanche research has made enormous progress in recent years. The aim is to develop a model chain that starts with the weather data, models the snowpack and finally derives meaningful stability indicators. However, anyone who talks to the researchers quickly realizes that there are still a whole series of unsolved problems. In any case, avalanche warnings are eagerly awaiting a stability map that is derived fully automatically from weather data.

Skitourenguru is keeping a close eye on progress and will replace the LLB with better successor products in due course. Skitourenguru is therefore not at the top, unfortunately the top is only on the drawing board.

Where do you think risk management in the snow is heading? Will there eventually be a chip in your head that not only reports the risk, but also identifies the exact places on the slope where an avalanche could be triggered on that day? I'm afraid there won't be a physical prediction for the individual slope for two reasons: Firstly, the chain listed above can at best model the probable snow cover. Secondly, we will never really fully understand the physical relationships within the snowpack. In the future, there will certainly be apps that warn us when we are approaching a slope that shows all the signs that make triggering an avalanche likely. There will be no actual physical prediction in the foreseeable future.

What are the next, foreseeable development steps for Skitourenguru?

As mentioned, I would like to replace the LLB with the above model chain, but this will not be possible for at least a few years. There are two important tasks on the agenda for Skitourenguru: Firstly, covering the entire Alpine arc with a full range of routes. Secondly, the possibility of having user-generated routes rated. On the one hand, this means routes that have been digitized by hand on the map before the ski tour. On the other hand, it is also about GPS tracks of real ski tours. In the second application, the user would see what risks they have exposed themselves to. In other words, they would receive feedback on their route planning. At the same time, Skitourenguru is constantly working on optimizing the algorithm.

How is Skitourenguru financed? Do we have to pay to use the portal in the long term?

Skitourenguru is financed by sponsorship from the Swiss Council for Accident Prevention (bfu) and Mammut. In my opinion, paid offers in the field of accident prevention are pointless. Ultimately, it comes down to the following question: are we addressing a very dynamically changing ski touring community or are we maintaining an ivory tower for insiders? In any case, I will do my best to ensure that Skitourenguru remains accessible without barriers.

Thank God avalanche accidents are still rare events. That's why we turn to the positive aspects of snow. To conclude: Do you prefer powder or firn, or as you say in Switzerland: "Sulzschnee"?

I'm a "Gfrörli", as they say in Switzerland. That means I love the sun. I also like to leave the cutting of virgin powder slopes to the risk nerds. In technical jargon, this is called "risk transfer". So powder snow is the answer, although an untracked powder snow slope is nothing to sneeze at.

Maybe one day Ski Touring Guru will be able to tell us when and where the best snow is. Until then, we will certainly have to think for ourselves, at least when it comes to snow quality. Günter, thank you for your commitment to avalanche prevention and the interesting conversation.

The article about the background to the Skitourenguru algorithm can be found in English here.

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

SnowFlurry
presented by