safety topics

Understanding avalanche problems! Or the danger of the danger levels

Thoughts from Drew Hardesty, long-time avalanche warden at the Utah Avalanche Center (USA)

by PowderGuide • 03/09/2016

For some years now, avalanche problems have been included in the situation reports of many warning services. The Utah Avalanche Center, among others, has been in charge of this. The avalanche problems are based on an article by Roger Atkins, published as part of the International Snow Science Workshop 2004. They indicate "typical, recurring and mostly obvious" hazardous situations. The avalanche problems provide an initial, rather rough overview of possible sources of danger, while the "danger patterns" developed by the LWD Tirol focus more on the causes.

The European warning services (eaws) have agreed on the following avalanche problems and now also use standardized symbols: Fresh snow, drifting snow, wet snow, old snow, sliding snow

In addition, there is an icon for "favorable situation" (explanation of avalanche problems). In North America, other problems are sometimes named, such as cornice breakage, loose snow (wet and dry), as well as "deep slab" and "persistent slab". The latter are summarized in Europe as "old snow".

Drew Hardesty has worked intensively on avalanche problems and, together with his colleague Wendy Wagner, has developed the "Avalanche Toolbox". In the toolbox you can see schematically which "tool" is appropriate for which problem. (See also) Three factors are considered for each problem: Predictability of "behavior", probability of remote triggers, usefulness of snow observations and tests. In this way, regardless of the hazard level, the difference between old snow problems and, for example, a new snow situation is worked out.

In the following commentary, Hardesty explains more about avalanche problems and the difficulty of summarizing complex snowpack situations in a hazard level. The text originally appeared on the Utah Avalanche Center blog and has been translated and lightly edited by PG in consultation with the author:

The Danger in the Danger Levels

"Democracy is the worst of all forms of government, except for all others." Who said that again? Mark Twain? No, Winston Churchill. But it sounds like he stole it from someone (Twain?). It's a bit like democracy with our avalanche danger levels.

Recently, at a lecture, I asked the participants to determine the danger level in two different situations. Beforehand, I gave them a little crash course: When we write our situation report, we evaluate each potential avalanche problem based on a number of factors:

The likelihood of triggering

Bruce Jamieson did a study a few years ago where he asked experts (avalanche watchers and the like) about the likelihood of triggering at different danger levels. (See illustration). The probability of triggering increases approximately 10-fold with an increase in the danger level.

In the past, the factor "probability of triggering" was considered the most important point in determining the danger level. Terms such as "unlikely", "possible", "probable" played a major role. As a rule, the level was indicated as significant when self-triggering came within the realm of possibility. This system did not work well. Imagine someone offered you 10 million dollars to play Russian roulette. Would it influence your decision if you knew how many rounds were in the magazine? What if I told you it had a rubber bullet in it? A bullet? A cannonball? A nuclear missile? Would you change your mind?

I asked my lecture participants to estimate the probability of triggering per danger level. The result was something like:
Low - 1-5%
Moderate - 25-35%
Significant - 60-65%
Large - 75-85%
Very large - 90-100%

I found the jump from moderate (35%) to significant (60%) particularly interesting. What is the key difference between these two levels? After all, most people are at levels 2 and 3. Where did the missing 25% go?

safety topics

presented by

The possible size of the avalanche in the event of triggering
The size of an avalanche is important for determining the danger level, as are the possible consequences. On the one hand, this has to do with the choice of terrain (is there a run-out or are you over a bergschrund or a rockfall), but basically goes back to the "destructive scale" (German: damage potential). We use the following scheme in operations:

The spatial distribution of the problem (individual locations, widespread, specific altitude zones/exposures)

How widespread or small-scale is the problem? The following graphic by Andrew McLean shows schematically how our representation of the hazard areas can be transferred to 3D. Most importantly, the spatial distribution is also an indicator of how often trigger points are found in each of the "exposure-height squares".

So after these explanations, I asked my audience to imagine the following situations.

The picture happened to be from the day of the talk. I had seen several of these self-detachments that had come off during the day due to wind and radiation. The cracks were widespread on sunny slopes and neither particularly large nor particularly powerful. The participants went through the checklist we had just discussed and noted their danger level on an index card. They were not allowed to talk to each other.

Situation B: This accident happened in the San Juan Mountains in Colorado, a fairly large area. This avalanche was the only one that was triggered in the entire area that day.

The result was clear:
The participants had selected level moderate for both situations. Given the definition of the danger levels, it is clear how this came about. However, it also shows how problematic level two can be and that there are different types of moderate (and all other levels).

I have looked at many incidents and have come to the conclusion that the moderate level is particularly critical in conditions that are unmanageable (i.e. uncontrollable, unpredictable). I describe old snow problems as unmanageable, which are often accompanied by hard snow slabs and a high potential for remote triggering.

Manageable: loose snow, fresh snow, drifting snow, most wet snow situations, cornice breaks. At least at the regular level, these problems can be estimated in terms of trigger point and probable size.

Unmanageable: Old snow problems (deep slabs, persistent slabs). Only the unpredictable is predictable here.

Two is not equal to two, three is not equal to three

The definition of hazard levels was expanded in the US a few years ago to include size and distribution, but I have a feeling that's still not enough. For example, the scale says, "Small avalanches in certain areas, or large avalanches in single spots". The skier in the video above would probably have said that the danger on that slope that day was major. (Unfortunately, he did not survive.)

European scale for comparison.

We are trying to reconcile two completely different things with this scale. Would you rather be thrown into a pond with 6 small reef sharks that might nibble your toes a bit, or into a tank with a single great white shark? Maybe there's additional blood in the water...

Over the years, we've come to realize that we generally don't have a problem understanding the snowpack. We know pretty well what's going on and what the problems are. We have a communication problem.

How high is the risk? Moderate? Significant? Sinister moderate? How do you put that into a danger level? I've never been comfortable with that, especially with old snow problems that persist for a long time. Here in Utah, we use the avalanche problems as additional information, but those who only take note of the danger level don't get it. Many perceive moderate as a kind of new low.

Everything depends on the character of the potential avalanche, which is reflected by the 'problems'. For me, the aspect of "manageability" mentioned above is crucial. Sometimes moderately is issued instead of significantly because the spatial distribution of the critical spots is no longer as problematic as in the days or weeks before, or the probability of triggering is somewhat lower. However, the existing old snow problem does not suddenly disappear. Maintaining the higher danger level just to protect the public means that the danger levels are no longer taken seriously in the long term.

When climbing, there are well-secured routes that are rated 5.9, for example (note: Sierra/USA scale). Sometimes a 5.9 is marked with an additional X, i.e. 5.9X. The difficulty is the same, but the X means that accidents can lead to "serious injury or death", for example because the route is poorly secured or the rock is very unstable. When you see the X, you stop and think about it. Mistakes here have worse consequences than in a normal 5.9.

If such an X were also introduced for the danger levels, situations could be better described for which significant or large is not appropriate due to the distribution of danger points and the sensitivity of the snowpack, but the consequences of triggering are very high. X is a function of the avalanche problem, the size and the possible consequences, as well as the "manageability", i.e. assessability or predictability.

In the end, it is not up to the warning services to decide how the danger scale can be improved, but to the public. What can be done to convey to casual readers the difference between "low probability, high consequences" and "high probability, low consequences" in a number or a single word?

How would you have judged the situations in Hardesty's example? Do you pay attention to the avalanche problems in the situation report, or is the danger level more important to you? How do you deal with the current problem of old snow this winter? Is the danger level scale and its definition sufficient to adequately describe all avalanche problems?

In this country, there have been repeated discussions in the past as to whether level three should be split up or whether it would make sense to rename it from considerable to major and major to extreme. What do you think of Hardesty's climbing comparison? Would a moderateX or considerableX be a way to summarize complex old snow problems in one keyword?

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)

Avalanche at Burgstall: How we deal with mistakes

Lea Hartl • 02/20/2017

She actually knew where the dangers lurked that day. Why didn't it stop her from climbing? A story about failure and the courage to talk about mistakes.
SnowFlurry 14 2016/17 | The Unconscious Incompetence in Snow & Avalanche Science

Lukas Ruetz • 02/02/2017

For a change, we immerse ourselves in theoretical thought processes relating to people, the psyche and snow. A fundamental assumption: the distinction between experience and knowledge.
The 10 decisive avalanche hazard patterns, Part II

Rudi Mair • 03/16/2011

When analyzing hundreds of mostly tragic avalanche accidents, Rudi and Patrick discovered that most accidents occur in typical, recurring hazard patterns. They therefore developed 10 typical, particularly accident-prone and frequent hazard patterns. Recognizing these avalanche hazard patterns should help winter sports enthusiasts to behave with appropriate caution – and thus avoid accidents.