Scientists are digging out the secrets of lethal flows of snow

Dan Miller, like many graduate students, spends a fair amount of time crouched over a hotplate in a small room. What sets him apart from most others are the parka and gloves he wears while doing so. The temperature in Miller’s 4-by-6-meter room can drop to –30C. It’s like a meat locker. In fact, it is a meat locker, a commercial freezer that Miller and his colleagues have tucked into a laboratory at Montana State University’s Department of Civil Engineering in Bozeman.

DOWNHILL RACERS. In 1968, a 3-kilometer-long avalanche scoured Oregon’s Mount Bailey. Today, skiers enjoy the scar left behind. Gus Gustafson/Mount Bailey Snowcat Skiing

JUST ADD SNOW. Steep slopes are one ingredient in the recipe for avalanches, such as this one on Utah’s Mount Baldy. Gus Gilman/Alta Ski Patrol

Also, the hot plate Miller uses isn’t so hot. It heats up to only about –1C. So what could he and his colleagues in Bozeman be doing in a freezer with a cold hotplate?

Cooking snow.

These hardy researchers are studying how snow crystals change shape under fluctuating environmental conditions. What they’re learning could provide clues about how films of fluffy white flakes become thick layers of tiny ice blobs on mountainsides. Under the wrong conditions, these crystalline layers collapse and avalanches result.

During field experiments in the nearby Bridger Mountains, the scientists hole up in wood-frame shacks, trigger avalanches, and then scrutinize the seething, white flow that whooshes over and past them. Elsewhere, analysis of damage from past avalanches is revising researchers’ understanding of how particles within these flows interact as they rush downhill. The research is reaping data that engineers can use for designing better protective barriers for areas at high risk of snowslides.

Single snowflakes are beautiful, fluffy crystals with hardly any weight at all; traveling downhill en masse at more than 100 kilometers per hour, they are one of nature’s most sudden and destructive forces, exerting pressures that can snuff lives, raze buildings, and scour the landscape.

Icy evolution

Avalanches come in all types and sizes, from the relatively small slumps of dry snow that travel harmlessly just a few meters downslope to roaring, 100,000-ton slab avalanches that can sweep a 3-kilometer-long path through a forest or bury a village. Almost all but the smallest avalanches stem from weak layers of snow that either evolve within the snowpack or form at the surface and are buried by subsequent snowfall or windblown drifts. When the weak layers can no longer bear the weight of overlying snow, they collapse and the snow above them breaks free and slides downhill.

All sorts of changes in weather conditions can weaken layers of snow, says Robert L. Brown, a civil engineer at Montana State University. For example, when cold, dry snow falls upon a sun-warmed drift of snow, the difference in temperatures between the two masses triggers changes in each. Vapor moves from the warmer, lower layer and recrystallizes as it seeps through the cold snow above, altering the shape and size of ice grains there.

How those weak layers evolve and the various conditions that cause them to form in the first place are just some of the phenomena that the researchers at Montana State are studying in their meat-locker lab. Although it’s often much colder in the laboratory than it is in the back country of Montana, the controlled experiments there are vital for understanding what happens on the mountain slopes.

“Understanding the physics behind the problem is one thing, but understanding the resulting effect on the snow is another,” says Brown. “Sitting in that meat locker for hours on end is necessary, but it’s not a lot of fun.”

Perhaps it’s more fun to ride out an avalanche in the field. That’s what Brown and his colleagues do five or six times each winter when they crawl into a cramped shed bolted onto the downhill side of a car-size boulder. Once inside, they set off an avalanche on the slopes above by detonating a grenade-size bomb suspended over the snow. Then, the researchers monitor conditions inside the billowing flow as it races past.

As dramatic as that sounds, the avalanches the Montana State scientists produce are miniature–about 3 m wide and 0.5 m deep–and they slide only about 200 m down the slope. Nevertheless, the researchers are making observations that challenge conventional avalanche wisdom.

For example, Brown notes, some scientists had theorized that one reason the material in avalanches hardens so quickly after coming to rest is that friction between particles in the swirling flow partially melts the snow, which then refreezes in a near-solid mass. Recently, Brown and his colleagues found that the temperature of the material within the avalanche does not really increase enough to cause any melting, especially when the snow is cold and dry to begin with.

Other observations of the individual particles within the flow may help scientists better understand avalanche behavior. At least for small sluffs like the ones Brown and his colleagues have triggered, the avalanche slides like a block of material instead of flowing like a fluid.

Even though many models of avalanche behavior treat the flows as though they were frictionless fluids, two simple observations prove that avalanches are neither like a fluid nor free of friction, says David M. McClung, a geophysicist at the University of British Columbia in Vancouver. First, he notes, an avalanche can come to a stop on a steep slope. Also, many avalanche deposits indicate that particles jostle and scrape each other during their tumble down slope.

When an avalanche flows through a curved valley, it sloshes up the side of the valley on the outside of the curve–a phenomenon known as superelevation.

The amount of superelevation, which occurs when the centrifugal force of the material in the flow overcomes gravity, depends on the speed of the avalanche.

McClung’s new model of flow, which incorporates friction between particles, results in better predictions of superelevation, he says.

Just that advance in predicting avalanche behavior could save lives. In the early morning of Oct. 27, 1995, an avalanche swept down on the small fishing village of Flateyri, Iceland, killing 20 residents and destroying 19 houses. All but one of the affected structures were outside recognized avalanche danger zones. The mountain ravine that spawned the killer flow had been the site of numerous avalanches, but none had previously reached as far downslope. The unusually large volume of snow that broke loose probably stemmed from blizzard conditions that had been pummeling northern and western Iceland in previous days.

After dropping 300 m down the mountainside above Flateyri, about 300,000 cubic meters of snow funneled into a curved channel. The flow scoured marks on the channel’s sides up to 40 m high. If the channel had been filled to that level with flowing snow all at once, the avalanche would have lasted only 1 second, McClung notes. That’s almost certainly not what happened. According to his new model, it’s far more likely that the highest flow marks reflect superelevation of a 3-to-10-m-deep flow of snow that raced around and up the side of the curve during an avalanche lasting about 23 seconds.

McClung notes that conventional theories peg the speed of the Flateyri avalanche at about 54 meters per second. His model estimates the speed was actually between 60 and 70 m/s, a difference that could be significant for engineers who are designing barriers to deflect avalanches from populated areas or estimating the immense pressures that avalanche flows would impose on buildings in their path.

Watch out for that tree!

People and avalanches are a bad mix. More than 600 people have been killed in avalanches in the United States since 1950, says Dale Atkins of the Colorado Avalanche Information Center in Boulder. The rate of fatalities increased dramatically between 1950 and 2000; almost 40 percent of reported avalanche deaths occurred in the 1990s. Atkins suggests that the striking jump in avalanche deaths from the 1980s to the 1990s–from 143 to 234–is probably simply the result of more people enjoying recreation in the backcountry. For example, snowmobile registrations in Colorado almost doubled between 1989 and 1999.

More than 83 percent of all avalanche victims since 1950 were engaged in some sort of recreation at the time of their deaths, says Atkins. The profile of the typical avalanche victim has changed through the decades, however. During the 1960s and 1970s, mountain climbers topped the list of fatalities. In the 1980s, backcountry skiers led the pack, and in the 1990s snowmobilers accounted for the highest percentage of deaths.

Contrary to common belief, loud noises such as shouts, sonic booms, or the roar of low-flying helicopters don’t trigger avalanches, says Bruce Tremper, director of the Forest Service Utah Avalanche Center in Salt Lake City. Instead, he notes, as many as 90 percent of avalanches resulting in fatalities are triggered directly by the weight of a person traversing the snow.

Signs of instability in the snow almost always precede the few fatal avalanches that aren’t caused by people. Says Tremper, “Stock market crashes, meteor impacts, and lost love often occur without warning, but avalanches don’t.”

So-called dry-slab avalanches account for almost all fatalities, Tremper notes.

When the weight from a new snowfall or accumulating snowdrifts overloads a weak layer of underlying snow, the slab of material on top begins to slide downhill, shatters like a pane of glass, and then breaks into a billowing mass of powder.

Dry-slab avalanches typically accelerate to speeds between 100 and 130 km per hour in about 5 seconds. Victims on the slab have almost no chance to escape the flow, Tremper says. Almost one-quarter of people who die in avalanches are killed because they hit rocks and trees on the way downhill; most of the rest die from asphyxiation once they’re buried.

A much less dangerous failure of weak snow layers happens on flat terrain, when the frail internal layers suddenly give way, allowing overlying layers to collapse. Such rapid settlings are commonly known as whumpfs because of the sound that they make, says Bruce Jamieson, a civil engineer at the University of Calgary in Alberta. Whumpfs occur when the weight of the snow above a weak layer–or the added weight of an animal or a passing skier–causes the layer to suddenly collapse. Because a weak layer is often much thicker than the diameter of individual ice crystals in the layer, the surface of the snow can drop several millimeters during a whumpf. That sounds small, but it’s enough to wreak havoc.

Once begun, the weight and momentum of the collapsing upper layers of the snowpack compress the weak layer and the whumpf spreads outward like a ripple on a pond. Whumpfs can travel distances of several hundred meters and can trigger avalanches if the collapsing weak layer reaches steep terrain that’s supporting a lot of snow. One whumpf in Antarctica, set off by an explosion, traveled at least 8 km in one direction from the blast.

Tough to predict

Because scientists are just beginning to understand the varied and complicated phenomena that produce the weak layers of snow that can cause avalanches, there are few models that can help predict when the snowslides may break loose. Some of the better models can simulate the evolving conditions in the snowpack fairly well, says Montana State’s Brown. They can estimate the temperature of layers as a function of depth within the snow, as well as determine temperature gradients within the snowpack, the density of its different layers, and the depths at which the strong and weak layers occur.

Those models don’t do as good a job of predicting the size or shape of ice grains within the layers or the strength of the bond between different layers of snow–factors that also affect where the weakest spot in the snowpack lies.

One way of determining the risk of avalanche in any given spot is to dig pits and look for telltale signs of weak layers, such as so-called depth hoar and surface hoar–freezer-burn-like frost that occurs when snow vaporizes and recrystallizes. The technique is useful but costly and time-consuming.

Without digging, the best way to get an idea of where avalanches might occur is to compile massive databases of where avalanches have already occurred, along with the meteorological conditions that spawned them. Pertinent factors include the temperature, wind speed and direction, and the amount of recent snowfall, says Karl W. Birkeland, an avalanche forecaster at the U.S. Forest Service’s National Avalanche Center in Bozeman.

Using what’s called the nearest-neighbors approach, computer programs search a database for past days during which the weather was most like that for the current day. For example, if there’s been a 10-centimeter snowfall in the past 6 hours, if all of that snow is dry and can be blown about, and if the wind has been blowing from the south at 20 km per hour, then scientists will extract information from the database about days that had similar conditions. By noting where avalanches occurred on those days, researchers can determine whether a pattern exists and possibly identify areas at imminent risk for avalanches.

Because the detailed topography, and other factors that affect the occurrence of avalanches vary widely among different areas, information from one region can’t usually be generalized to predict avalanches in another. Also, says Birkeland, the nearest-neighbors method works only when there’s a complete database of pertinent weather conditions matched with detailed information about where in the region avalanches occurred on specific days.

In the United States, such detailed databases exist only for some national forests and for ski areas that have been in operation for a long time. Avalanche forecasters use the nearest-neighbors approach to analyze the historical database for some of the ski areas around Salt Lake City to help protect skiers participating in this year’s Winter Olympics.

Meanwhile, Miller and his colleagues are crouched over a frosty hot plate in a meat-locker lab in a Bozeman basement studying the microscopic details of ice and snow particles. It’s not glamorous, but these kinds of observations could help plow the route to better avalanche predictions.

More Stories from Science News on Earth

From the Nature Index

Paid Content