Wednesday, April 11, 2012

Avalanche Triangle & Human Factors

Two seemingly simple questions on my Level 1 course review quiz often flummox my students.
One is:
List the three proximate physical conditions that contribute to the occurrence of snow avalanches.
I get all sorts of answers, but what I'm looking for = terrain (i.e., steep and open) + snowpack (i.e., unstable) + trigger (i.e., you or a "natural" trigger, which is really just a change within the snowpack) = avalanche.

For my students' confusion I blame in part the traditional avalanche triangle.  Here are a couple graphical representations:

The new 2011 Fifth Edition of the classic Snow Sense explains:
"The interaction of three variables - the terrain, snowpack, and weather - determines whether or not an avalanche is possible.  Terrain is the foundation of avalanche, weather is the architect, and the snowpack is the winter's blueprint.  However, to determine whether an avalanche hazard exists, we must add an important fourth variable, us.  Without the present of people or property, there is no hazard."
The second edition of Bruce Tremper's Staying Alive in Avalanche Terrain rightfully downplays the causation element by referring to it as a "data triangle" yet still reproduces the same isosceles triangle from Snow Sense.

So what don't I like about the traditional triangle graphic and the Snow Sense explanation?

  1. First, it mixes together both direct and indirect factors.  Specifically, we care about weather only because of the snowpack it creates.  If we possessed a magical means of somehow scanning the entirely snowpack on a mountain (and such research is even in its initial stages), then we would not care at all about recent or season-long weather in evaluating slope stability.  But since we have only very limited and time-consuming means of examining a snowpack, and hence face considerably uncertainty regarding the snowpack condition, we rely up our (typically) more certain knowledge of recent and season-long weather.
  2. Which brings me to the second problem:  not only do the equal sides of the triangle imply equal weights to what are direct and indirect factors, but those equal sides also mask the unequal certainty with which we can know those factors.  Specifically, we can know terrain with absolute certainty.  (Although exceptions certainly exist with unfamiliar terrain in poor visibility.)  Weather we can also know with reasonable certainty (given some combination of personal observations and telemetry).  Snowpack though (except at the extreme ends of the stability spectrum) will always be uncertain to some significant extent.
  3. The third problem is what's inside the triangle.  It's us, but as a passive "presence" - the reality though is that we're there because we made a decision, and AIARE has emphasized decisionmaking throughout its curriculum.

Which leads me to the second question that flummoxes many students, although admittedly I deliberately phrase it in a way to lead them into the tradition thinking that I want to critique:
"About what percentage of avalanche incidents are attributable to human factors?"
A typical answer is 82 percent, based on a citation in Bruce Tremper's Staying Alive in Avalanche Terrain to research by Dale Atkins.  Staying Alive also includes a table for "Primary Factors in Fatal Avalanches 1990-2000" which is reproduced from "Human Factors in Avalanche Accidents" presented by Dale Atkins at the 2000 International Snow Science Workshop ("ISSW").  Out of the sample size of 41 fatal incidents, 1 is attributable to snowpack, 2 to weather, 4 to terrain, and 34 to human factors.  So just under 83 percent in the table for human factors.

However, the correct answer I am looking for is . . . 100 percent!  An avalanche incident can't occur unless human factors led to the decision to be exposed to the potential avalanche terrain.  (Even an incident such as the Alaskan avalanche hitting a warehouse would not have happened had someone not made the decision to locate the warehouse near an avalanche path.)

I lack information on the details of the 41 fatal incidents of the 2000 ISSW paper, but nevertheless, I can assess the primary factors in each incident:

  • Snowpack = Yes, the snowpack had to be a primary factor, since without an unstable snowpack, the avalanche never would have occurred.
  • Weather = Yes, the weather had to be a primary factor (albeit indirect), since without the weather to create an unstable snowpack, the avalanche never would have occurred.
  • Terrain = Yes, terrain had to be a primary factor, since without a steep and open pitch, the avalanche never would have occurred (as the unstable snowpack would have stayed put, despite its instability).
  • Human Factors = Yes, the human factors had to be a primary factor, since without the human decision to be in an unstable snowpack, the avalanche incident (with humans and/or our infrastructure) never would have occurred.  (Perhaps an avalanche might have occurred - via a natural trigger - but as a variation on the old line about "If a tree falls in the forest and no one is around to hear it does it make a sound?" if a slope slides and no one is around to be hurt by it, then no, we don't really care about it.)

So where am I going with all this (if anywhere)?

  1. Ditch the avalanche triangle graphic.  For a substitute, AIARE has the ever-evolving Decision Making Framework ("DMF") graphic, but it's much more complicated, and not addressing quite the same issue.  Perhaps some sort of yin yang graphic, representing terrain and snowpack, with weather feeding into snowpack, and the human element represented . . . somehow.
  2. Acknowledge human factors as a universal and critical factor in all avalanche incidents.  Although speaking of the "White Death" as the enemy makes for fun drama, the reality is more introspective.  Staying Alive mentions the following classic line in the context of triggers, but it is even more appropriate and insightful in the context of the decision making that is at the core of every avalanche incident: 


  1. First, I think you're overthinking it. :-) I'd wager that few folks beyond yourself look at the shape and see an isosceles triangle and equate each side equally. I'm not arguing that the avalanche triangle is sacred, only that it doesn't seem to be as complicated as you make it out to be.

    Second, there are direct weather factors, particularly with respect to the snow surface: solar, wind, temperature, so we do care about weather in this sense.

    Third, it's not always possible to know terrain with absolute certainty. Maps and of course seeing it with our own eyes (with snow on it) help a lot and tell us much, but it doesn't tell us what's beneath that snow. We can infer, but there are still traps lurking there.

    I'd agree that knowing the snow pack is the most uncertain and time-consuming problem.

    Lastly, while the graphic may imply the human presence as passive, you don't have to describe it like that; you can make it clear that the opposite is more true.

    All that said, your points are of course valid and make for good discussion.

  2. Me, overthinking something? Nawh, you must have someone else in mind! (Besides, I've still done less thinking on this than all the discussion we've had on the AIARE DMF graphic in my ITC & IRC's over the years...)
    As for direct weather factors, all those examples are still just indirect factors, in the sense that they're changing the snowpack, a snowpack that was previously created by previous weather. And the snowpack, in combination with terrain, is what produces an avalanche ... that is, when a trigger changes the tenuous balance.
    Even a "natural" trigger like rain or new snow, or rapid warming, once again, that's just the snowpack changing -- on account of weather, but still, it's terrain + snowpack = avalanche, with weather acting only indirectly by changing the snowpack.
    I suppose weather could act as a trigger if lightning struck the slope, but that's kind of a stretch...

  3. We'll have to agree to disagree on direct vs. indirect in this case. It's probably more a matter of semantically split hairs.

    And of course terrain capable of producing an avalanche is necessary. I'm perhaps not as certain as you are that I always know terrain, that can be a subtle thing sometimes.