Is there a problem with the Avaluator?
Calgary Herald April 19, 2009
As a season in which 25 people have died in Canadian avalanches nears its end, snow scientists are arguing about whether a new risk-assessment tool recommended by most Canadian avalanche experts is making backcountry users safer, or giving them a false sense of security.
Until 2006, avalanche awareness education for backcountry users was based on the assumption that more avalanche safety knowledge would lead to less accidents. But after studying hundreds of avalanche accidents in North America, snow scientists Pascal Haegeli of Vancouver and Ian McCammon of Salt Lake City found that in most cases the avalanche hazard was obvious on the day of the accident. What's more, they noticed that most of the victims had taken at least one avalanche awareness course.
So, they created the Avaluator, a tool designed to help a recreationist make an informed decision about whether to risk travelling a slope based on an objective look at the conditions and the user's own level of experience.
In the winter of 2006-07, the Canadian Avalanche Centre, a non-profit group based in Revelstoke, B.C., began recommending the new Avaluator: Avalanche Accident Prevention Card to backcountry users. That same year, the CAC incorporated the new risk-assessment tool into its basic avalanche course curriculum taught at mountain schools across Canada.
Students and instructors gave mostly favourable reviews. But there were already rumblings from some snow scientists, who were skeptical of the science behind the new tool.
How it works
The Avaluator provides a simple framework for assessing the seven most obvious clues that a slope might slide.
Using a colour-coded chart on the Avaluator card, the recreationist can combine this knowledge with the danger rating from the regional forecast and their level of backcountry experience to come up with a recommendation for how much caution they should use on a mountain. Recommendations on the chart range from green (normal caution) to yellow (extra caution) to red (avoid a slope).
What do users think?
Last year, the CAC interviewed 17 avalanche skills instructors on the value of the tool. Overall, instructors complimented the Avaluator for helping a user focus on the most important factors when assessing avalanche terrain. But the Avaluator was criticized for being too limiting, restrictive and vague in its recommendations.
The heads of the biggest avalanche skills providers in Alberta and British Columbia, however, both report that most of their beginner students appreciate the Avaluator's simplicity.
Brian Jones, the owner/operator of Canada West Mountain School in Squamish, B.C., points out that students who want more in-depth knowledge can take more advanced courses.
"The CAC's safety model is very appropriate to their target audience," says Jones.
And Albi Sole, a snow scientist and the avalanche safety programs co-ordinator for the Outdoor Centre at the University of Calgary, says that until the Avaluator was adopted as part of basic avalanche course curriculum, students often left a course feeling confused and uneasy about their skills.
"There were no concrete clues then," Sole explains.
So what's the problem?
The Avaluator has some vocal critics.
Snow scientist Bob Uttl of Calgary argues that the tool is based on unproven research and is actually causing more deaths than it is preventing. He's called on the CAC and Parks Canada to recall it immediately and issue a public safety warning to users.
At the centre of the controversy is one page in the 30-page booklet: it includes a chart of "prevention values" showing that 77 per cent of recorded avalanche accidents where four or fewer obvious risk signs existed might have been avoided if the victims had used the Avaluator's seven-question obvious clues test.
According to Uttl and other snow scientists, the Avaluator's statistic, which implies that more than three-quarters of avalanche accidents could be prevented, is based on flawed reasoning.
Uttl has conducted three studies using data from about 250 to 300 accidents reported in either Canada or the U.S. that he says prove the Avaluator's prevention values are greatly exaggerated.
In his latest study, Uttl found that just nine per cent of past accidents in Canada where four or more obvious avalanche signs were present, and 18 per cent of those he studied in the U.S., could have been prevented using the Avaluator's obvious clues test. So he reasons that the Avaluator is giving users a false sense of security.
Sole didn't bother with a separate study. But he reasons that the Avaluator's prevention values aren't verifiable because they don't take into account all the people who went out and didn't get caught in an avalanche.
"And you will never get that information," he adds.
Sole has instructed avalanche safety instructors with the Outdoor Centre to tell their students to disregard the obvious clues prevention values in the Avaluator. Contentious statistics or not though, the Avaluator's obvious clues are still based on sound science, he argues.
"Those clues are so strong, and they're easy to understand in a weekend course. And if the course is longer than a weekend, most people aren't going to take it," Sole argues.
James Floyer, a Revelstoke-based snow scientist, also criticized the Avaluator's prevention values in his independent report on the Avaluator for the CAC last September. After studying 100 avalanche accidents in the U.S., Floyer found that 48 per cent of the accidents he studied where four or more obvious signs were present could have been avoided.
Floyer, who took a job with the CAC as a public avalanche forecaster last November, recommends in his report that users be counselled not to take the prevention values of the obvious clues seriously. He describes the prevention values as confusing, and recommends they not be published in future editions of the Avaluator.
"I don't believe those prevention values really give the Avaluator any added value," Floyer says. "I don't think they're appropriate and I don't think they're being used properly by most people."
How we got here
The problem partially stems from the fact that the scientists all used different avalanche accident data to come up with their statistics.
Haegeli and McCammon based their studies on hundreds of past avalanche accidents in Canada and the U.S. that either included a fatality or a serious injury. Such serious accidents prompt professional rescue attempts, which lead to thorough, credible accident reports written by industry professionals.
"When you get a professional report, you can rely on the data," McCammon says.
The obvious clues are based on the most common eyewitness observations McCammon noted in 750 well-documented reports from accidents that occurred on U.S. mountains between 1972 and 2003.
McCammon focused on well-documented accidents — usually involving at least one fatality — that offered definitive details on all the obvious clues present leading up to the accident.
His co-author, Haegeli, backs this approach, noting that avalanche accidents that don't include a fatality are much more likely to be misreported, because the resulting reports are often incomplete.
Uttl counters that McCammon has skewed the Avaluator's prevention values statistics by not including 82 per cent of the 1,400 accidents he studied because they didn't mention the presence or absence of one or more of the seven obvious clues leading up to an accident.
Uttl contends that if an obvious clue was not reported, it likely wasn't present leading up to the accident. In fact, Uttl is just finishing up a study of accident reporting that he says proves eyewitnesses to accidents rarely report the absence of obvious signs in their accident report, but almost always report such signs when they observe them.
"According to (McCammon), the average number of clues when an accident happened was five clues. But that's crazy because he eliminated all the accidents with missing values," Uttl says.
Uttl also charges that McCammon is withholding access to the coded data that shows whether McCammon judged an obvious sign to be present or absent in each accident report used in the Avaluator study. But McCammon said he has explained to Uttl that the data is available to the public at the Colorado Avalanche Information Centre.
The problem is that the records at the Colorado Avalanche Information Centre are only stored on paper, necessitating a visit to the facility. Uttl has yet to make the trip. Floyer wanted to for his report, but the CAC told him the trip would be too expensive for the report's budget.
Floyer actually agrees with McCammon that the accident data used by the Avaluator authors is of higher quality than the online reports he used in his study, because there are fewer professional reports in his online data-set.
Uttl, however, says McCammon is the one who used an inferior data-set.
McCammon would like to see Uttl's "attention-getting claims" and studies put under the same scientific microscope that McCammon and Haegeli's studies have gone under as part of the peer-review process before being published in 2007 in the Cold Regions Science and Technology, a scientific journal. Uttl says he plans to do just that, but declined to name the journal and an expected publication date.
Is there an alternative?
Uttl would like to see the CAC adopt the Nivo test developed in 1997 by French snow scientist Robert Bolognesi. The 25-question test is used by thousands of French backcountry ski tourers to assess avalanche risk. Reached via the Internet in France, Bolognesi wrote in an e-mail that the Avaluator is "an effective tool." But he qualified this assessment by adding that the Avaluator is "sometimes too simplified," because it gives equal weight to each of its avalanche-risk factors, unlike his weighted system.
"Avalanche prediction is usually reliable when A LOT OF independent indicators converge to the same diagnosis," he wrote.
But Haegeli says the Nivo test wouldn't be applicable in Canada, since its factors were weighted based on observations made at European ski resorts, where avalanches are often artificially released using explosives and the weather patterns are different.
Sole prefers the Avaluator as well, because he believes the Nivo test is too complicated for most backcountry users.
"I say keep it simple. Seven clues is plenty," Sole says. "My students like it."
Indeed, the Reduction Method used in Switzerland, Germany and Austria is even simpler than the Avaluator. It uses only two factors: the avalanche danger rating from the regional forecast, and the slope angle for each slope to be travelled. But this is easier for Europeans to use, since maps in the Alps often include slope angles, something that has yet to be added to Canadian maps.
Sole says that Uttl is "on a crusade" and characterizes his peer's criticisms of the Avaluator as a "little, picky academic debate (about) an extremely dense, technical question."
The Avaluator isn't perfect, admits John Kelly, the operations manager for the CAC. That's why, for the past two years, the CAC has been gathering data and feedback from users of the Avaluator across Canada, with plans to revise the tool based on the feedback.
Kelly remains a staunch defender of the Avaluator, saying it is the best tool available to the average backcountry user.
"The fundamental principle of the Avaluator is to match your trip selection to the current conditions. That is the best answer that we have so far to helping people stay out of harm's way," he says.
Even Floyer argues that lives would have been saved over the past three years if more backcountry recreationists had used the Avaluator and adjusted their route to match its recommendations.
Haegeli and McCammon say that Uttl's concerns go against decades of study findings by the snow-science community.
"What Uttl is fighting against is pretty well-established," McCammon says. "Pretty much all of the avalanche books out there say that when avalanche accidents happen, victims usually ignored the obvious clues."
Indeed, in the introduction to the best-selling avalanche book Snow Sense: A Guide to Evaluating Snow Avalanche Hazard, authors Jill Fredston and Doug Fesler write: "Usually, when avalanche accidents are investigated, it is found that not just one or two clues were ignored, but three, four, or five clues by the time the group got into trouble."
Aside from the scientific arguments, Kelly says that as an avalanche professional with decades of experience, he knows that the disputed "obvious clues" test draws a backcountry user's attention to the most important factors affecting a slope's risk of avalanche.
"I have to say, the world is a better place with the Avaluator than without it."
As much as they are at odds about the Avaluator, all the experts seem to agree on one point — after 25 avalanche-related deaths this season, backcountry users need all the help they can get to make better decisions in the mountains.
- Contributors can log in for advertising-free pages.