Colorado Will Beta Test New Wildfire Prediction Model
Arizona's Yarnell Hill fire claimed the lives of 19 firefighters. Colorado's Waldo Canyon, Black Forest, and High Park fires were some of the costliest ever in terms of homes and property lost.
A better wildfire weather prediction system might have saved more lives and property. The state of Colorado thinks so, and has agreed to beta test a new system pioneered by the Boulder-based National Center for Atmospheric Research, starting late in the 2016 fire season.
William Mahoney, deputy director of the research applications lab at NCAR, said the technology has the potential to make firefighters safer, because it combines two models: one that predicts weather and another that predicts fire.
Scientists know that in large fires, the heat and moisture from the fire "actually changes the local weather," said Mahoney, but right now "everyone is kind of blind to that reaction." This is a problem, because the interaction between the fire and the weather often leads to some pretty extreme events, he said.
"Like mini-tornadoes or firebursts, back burnings, reversals over the fire, sometimes things like pyrocumulous clouds will form over the fire, which is thunderstorms that are caused by the fire itself."
During California's and the Pacific Northwest 2015 fire season, "the firefighters were saying this fire is huge and it is creating its own weather and they have absolutely no tools on how the fire will behave," Mahoney said.
That's dangerous. A fire acting unpredictably can suddenly change direction, racing over a ridge or blowing up where a plane is trying to drop retardant. Predicting what might happen will "make the fire mitigation and firefighting operations more efficient," said Mahoney. It will also help emergency managers plan evacuations and know where to concentrate fire fighting efforts.
Right now, as Mahoney explained, there are weather models, and there are fire models, but the two don't mix. Since the modern, super-wildfires of today are creating their own weather, neither model works all that well alone.
Yet even though NCAR developed this model in their laboratory and let the federal government and states know about it, they have not had the chance to test it in real time. Until now.
In the 2015 legislative session, Colorado lawmakers, frustrated by what they viewed as slow federal responses to devastating wildfires in 2012 and 2013, voted to fund the NCAR model for a real-world trial over the next five years. State legislators also approved new aerial firefighting capabilities in 2013 as part of this effort.
Mahoney said one challenge is taking the model from the lab bench to the local emergency manager. Ideally, when lightning strikes a tree and a small fire starts, people in the state's Division of Fire Prevention and Control can go to a computer, click on a map where the fire is, and that will launch the NCAR model.
That model will generate an 18-hour prediction of where the fire is going and how it is likely to behave.
"And they'll be able to see where the fire is evolving every hour out to 18 hours," Mahoney said.
The prediction will also be updated every 3 to 6 hours.
For the next six months, Mahoney's team will be working on making this happen. They plan to have the tool ready for operations late in the 2016 fire season.