OpenSnow, a specialized weather forecasting service designed for the mountains, has added two artificial intelligence models to its quiver of forecasting tools.
Most recently, forecasters introduced the StormNet model, which is specific to predicting severe weather and lightning. OpenSnow engineer Andrew Brady said it more accurately identifies when there will be strong winds, hail and tornadoes.
“What AI is helping us do, enabling us to do, is building a computer algorithm that's actually able to analyze the current state of the atmosphere, so the current conditions, and then simulate what it'll look like in, six hours, 12 hours, all the way out 15 days into the future, just using pattern recognition and just tying those steps together one by one,” he said on KPCW’s “Local News Hour” March 27.
From a user's perspective, the forecasts look the same but Brady says the models make each forecast more accurate.
The second AI model now in use is called PEAKS.
“That addresses improving mountain weather forecasting, going beyond conventional weather models, for models fine tuned for mountainous weather and complex terrain,” Brady said.
While many meteorologists and weather centers like the National Oceanic and Atmospheric Administration are also beginning to implement AI models for broad forecasting, OpenSnow’s models are unique to mountain conditions.
“It is fine tuned to mountainous terrain and to ski conditions. So we're not looking, for example, when we're improving the model, we're not only looking at the error over the entire US or the entire globe,” he said. “We're looking at that era. We're also specifically focusing on ski resorts. We're focusing on complex terrain where conventional methods fail.”
Brady said OpenSnow is also working on a new avalanche prediction model that could examine avalanche danger up to 15 days in the future.
The United States Department of Defense and other emergency management organizations have also implemented the StormNet model which can be used for aviation, insurance and logistics purposes.
