By Bob Berwyn
I started really listening to snow when I was about 12. On one of my first solo off-piste forays, I detoured through the woods on the last run of the day. When I stopped to adjust my goggles in the deepening blue shadows, I heard a clump of snow rustle off a branch behind me. A second later, a white weasel jumped off the same branch, landed with a soft puff, then scuttled away through a layer of surface crystals that broke apart with the soft tinkle of glass breaking in the distance.
My brain processed the sounds in slow motion. I gently shuffled through the hoar to see if I could recreate the sound, then pointed my skis toward the softest snow I could find.
As skiers, we all try to read those sounds, from the Styrofoam-squeak of packed powder, to the hollow thumps of wind-hardened sastrugi and slurpy splashes of spring corn. Even from the chairlift, we listen to the turns of the skiers on the mountain below because it helps us learn about the snow we're about to ski on.
Now, imagine you had ears that were 1,000 times more sensitive—so keen they could hear the low frequency rumble of an avalanche releasing before the snow moves at the surface. Imagine you had eyes that could peer six feet deep into the snow, picking out layers as thin as half a centimeter. And imagine that data was collected and stored, a blizzard of instantly accessible information that could tell if the snowpack is safe.
In Bergen, Norway, a startup called ThinkOutside is packing at least some of those ideas into an iPod-sized radar device that will be mounted on up to a couple of hundred pairs of skis and tested by avalanche forecasters and other snow scientists and experts in Scandinavia. The device clicks onto the skis in front of the binding and send radar waves into the snowpack as you move over it.
"You know how radar works—some of the energy reflects back, and it gives you a very detailed picture of what's there. It's like having eyes under the snow," says ThinkOutside CEO Monica Vaksdal. "With every move you make, it sends back multiple shots of data sets. It reads in real time what's underneath you every step of the way."
Vaksdal, a lifelong skier, says the 16 years she spent working for the oil industry as a geologist gave her the tech background needed for the venture. A drop in oil prices a few years ago got her thinking about wanting to be part of Norway's transition away from fossil fuels. The aim of her company is to focus on the transfer of oil exploration technology to new markets as part of the decarbonization process.
She said the final nudge to push for completion of the vision came last winter.
"My husband called me from home and said, 'I had a terrible day, I just triggered an avalanche.' It was on a slope near our cabin that we've skied a million times. I couldn't believe it. He triggered a 100-meter-long avalanche. It was distance triggered, so he was OK, but it became so clear to me—we just have to do this."
To fund the project, ThinkOutside partnered with research institutes, Ă…snes Skis, and also earned some European Union grants for innovation. After getting feedback from avalanche professionals this year, Vaksdal hopes to perfect a device that could someday help empower recreational backcountry riders to make better decisions in avalanche terrain.
"We started about a year and a half ago and made some ugly devices," she said, laughing as she described the early attempts to mount the radar on skis. "In the late part of last season, started collecting data, and we were stunned with the accuracy of the device, even on crappy, warm, and coarse spring snow that is mostly homogenous.
"We're searching for weak layers. We collect a lot of data, but we also do a lot of synthetic data. We are using machine learning to train the devices, and the computing cloud they're connected to, to look for these kinds of elements in the snowpack. Right now we have 150,000 features of what a deep snowpack can be," she says.
While high-tech tools won't ever replace hands-on avalanche skills, the future of avalanche forecasting probably will include huge data sets that are fed into powerful machine learning algorithms and then made accessible by devices and apps like the ones Vaksdal's company is working on.