Lightning strikes are one of the main drivers of wildfire in the state, according to the Alaska Fire Service. They sparked most of the fires that swept across the Interior at the peak of this year's fire season, burning thousands of acres and dozens of structures.
But seasonal wildland fire outlooks usually don't factor in lightning because local lightning forecasts are difficult and time-consuming to put together. A group of researchers at the University of Alaska Fairbanks is now working to remedy that using machine learning.
Josh Hostler, a PhD student at UAF, heads the lightning modeling project. He said training AI to look at weather conditions can help estimate the likelihood of extreme lightning events over a season.
"We tell it what patterns occur on each day, and we have it predict categories of lightning intensity for that day," Hostler said. "Then we give it a low, medium, and high category."
The group released its initial findings in the journal American Meteorological Society last month. Hostler said they're working toward producing monthly lightning strike forecasts to help Alaska's fire managers stay on top of fire conditions.
Climate change could make those forecasts even more important. One recent study found that climate change is increasing lightning frequency and wildfire intensity across the globe.
Heidi Strader, a program manager with the Alaska Interagency Coordination Center, which organizes national resources to deal with wildfires, said it's hard to say for certain whether that's observable in Alaska yet. But she said she does see the conditions that spark lightning storms hang around longer.
"With the climate changing so rapidly, we're seeing that some of the old rules of thumb on these are really falling apart," she said. "And one thing that we see with climate change is that we're getting these periods of stalled weather patterns."
Strader helped guide the lightning model project at UAF. She said the team still has some fine-tuning ahead, but she's pleased that the groundwork has been laid to give fire crews a head start before next year's wildfire season.
"Now that we have this baseline," she said, "the goal is to take that and apply it to the forecast for an upcoming season and see if it can pick up on: are we going to have a year that has more of these patterns that trigger big lightning events, or is it going to be a slow year where we really don't see that happen?"
Strader said she hopes the team can someday repeat this process to predict other things that spark and accelerate wildfires, like big wind events or periods of hot and dry weather.
Copyright 2025 KUAC
