Each year, about 18 people are killed in avalanches in the United States. About one third of these fatalities occur in Colorado, which has a large number of high, snowy peaks, covered with a light, airy snow considered ideal for winter sports. Colorado also has one of the most effective avalanche forecasting systems in place in the United States.

"We have a network of 40 observation sites around the state," explains Knox Williams, who heads a team of four forecasters at the Colorado Avalanche Information Center in Boulder, Colorado. These observers check snowpack conditions by digging snow pits. "Keeping track of the snowpack layers as they develop and metamorphose over time is a big part of being an avalanche forecaster," says Williams. "This field data gives us a snapshot of current snowpack conditions. The other basic elements that go into a forecast include current temperature, wind conditions, snow fall, and weather forecast. We put all these pieces together and issue an advisory twice a day during the winter season, which runs from November to April."

They also track all the avalanches reported in Colorado. "Avalanches tend to run in groups or cycles," Williams explains. "A big storm, lasting four or five days, can result in 500 avalanche reports. We also pay careful attention to wind directions, because slopes that face similar directions will tend to become avalanche-prone. If we hear that there is an avalanche on [an] eastern-facing slope, we can predict that other eastern-facing slopes, with similar gradients, may also be prone to avalanche."

After evaluating all the available data, the forecasters issue a hazard rating, based on an internationally used five-tiered system. The ratings ranges from a low risk to an extreme risk of avalanche. The higher the risk, the smaller the trigger needed to set off an avalanche. "To a certain extent, these forecasts are subjective," Williams says. "We can't predict exactly what the weather will be, and we can't predict where the snow will be deposited. It requires a great deal of professional judgment to make a good forecast."


According to Dr. Bob Brown, a professor at Montana State University and one of the leading avalanche researchers in the United States, "Right now, we can forecast hazard levels, but it will be several years, at least, before we can say that there will be an avalanche at a certain place at a certain time. The science is still too imprecise."

Dr. Brown knows avalanches. He was nearly killed by one eight years ago, while skiing with two colleagues on a mountain just outside of Bozeman, Montana. They were collecting information about avalanche conditions by digging pits in the snowpack.

Snow pits are the basic field tool of avalanche research. They are usually several feet square, and set near an area prone to avalanche. The pits, cut from top to bottom, reveal a cross-section of the various layers that make up the snowpack. Each layer of snow has its own unique characteristics, depending on temperature and weather conditions. Some layers may be firm and compact, while others are brittle and unstable, and still others may be riddled with frost or icy deposits. Using a paintbrush to skim its sides, scientists can check the status and strength of the individual snow layers and the size of the individual grains. "We study the complete stratigraphy to look for weak layers and sliding layers," explains Dr. Brown.

"We had already dug a couple of pits, and made a risk evaluation for the slope, and began skiing down," says Dr. Brown. "I was first. The slope was about 45 degrees, and as I headed down I saw a snow pillow, where the snow had blown and accumulated into a mound, so I skied around it. You don't want to ski over a heavily loaded spot like that because it can trigger an avalanche. But one of the guys with me didn't recognize the hazard and skied right into it, kicking it off."

"I heard a yell, looked up, and it was coming right at me. There was nothing I could do about it. You can't outrun an avalanche. It wasn't very large, and I was on the edge, but it carried me about 100 yards. I tried to grab a tree, but I couldn't hold on. It was [strong] enough to carry me to the top of the flow, so when the avalanche finally dumped me off, I didn't get buried."

"The guy who hit the pillow wasn't very experienced with avalanches," says Dr. Brown. Roughly 95% percent of avalanche victims get caught in slides they trigger themselves, or ones triggered by a member of their party. "We were just taking him out, and getting him used to the slope conditions. Since then, he's taken an avalanche course, and become quite good [at] recognizing the conditions."

"If you are careful," says Dr. Brown, "back country skiing is not that dangerous. There are lot of people skiing, and not that many get killed. But we can't predict conditions exactly, so you try to hedge your bets. If you think the snow is unstable, or fairly unstable, you should stay off the open slopes and stay in the trees. Trees and brush tend to anchor the snow, and keep it from avalanching."


The future of avalanche prediction may lie with sophisticated computer programs designed to model the behavior of snow when it's confronted with a variety of climatic and geographic circumstances. One of the most advanced systems is being developed by Swiss researchers, who have spent the last ten years developing a computerized avalanche prediction system.

"In Switzerland, 65% of the population lives in areas that are threatened by avalanches," says Dr. Brown. "They need to know the avalanche tracks and predict the hazards on a daily basis." In 1999, a unusually severe winter dropped a large amount of snow, creating avalanches that did billions of dollars in property damage and killed more than 100 people across the Alps.

Dr. Brown recently spent a year at the Swiss Federal Institute for Snow and Avalanche Research in Davos, Switzerland. Researchers at the center have built a central computer system linked to a countrywide network of electronic monitors collecting meteorological data like wind speed, wind direction, air temperature, humidity, pressure, incoming solar radiation, and reflected radiation. They also measure the surface temperature of the snow, and the temperature every five meters (16 feet) into the snowpack.

"All this information is fed into a central computer program, which does an update of the snowpack conditions every 15 minutes," says Dr. Brown. "The computer program takes the daily weather, and tries to predict the characteristics of the snowpack from the top to the ground, and then comes up with a stability index for various slopes. The Swiss want these computer programs to become their mainline of defense against avalanches."

Dr. Brown is currently adapting the Swiss computer programs to create a general forecast for large regions in the United States. "We've been running the program to see how it compared to actual conditions," says Dr. Brown. "We did it last winter and found a few things that were wrong. So this coming winter, we will try it again. Right now, I wouldn't recommend that anyone make a prediction using this program. The model is still in development, and it will take time before it is reliable enough to use. It will be at least ten years before we can stop digging snow pits."