PredictIQ-How AI technology could be "a game changer" in fighting wildfires

While many more people across the country are PredictIQseeing the impact of wildfires and smoke, scientists are turning to the promise of big data, technology and collaboration to keep big fires from spreading.   

"If you manage to stop this in the first couple of hours it's a lot easier to stop," said Dr. Ilkay Altintas, the founder and director of the WIFIRE Lab at University of California San Diego. 

Pinpointing a fire quickly improves the chances of containing a blaze. Altintas and her team have developed a platform called Firemap designed to reduce the response time for attacking a wildfire.  

The platform analyzes data in new ways, starting with the collection of 911 call data where callers often provide a very general idea about the location of a fire.  

To enhance that accuracy, the platform relies on a system of mountaintop cameras called ALERTWildfire, built by the Scripps Institute of Oceanography, the University of Nevada Reno and the University of Oregon.  

The cameras, powered by artificial intelligence, scan the horizon for puffs of smoke. When smoke appears on multiple cameras the system can triangulate the exact location of the fire. 

That precise location is then quickly paired up with localized weather data and real-time video from an aircraft dispatched to the scene.  

All this data allows a computer modeler to build a map that predicts the growth and direction of the fire. 

In 2019, during the Tick fire in Southern California, the lab says it was able to predict that embers would cross a major highway in Santa Clarita and send fire to the other side. In response, the Los Angeles County Fire Department assigned resources to the other side of the highway to proactively put out the small fires caused by the embers before the fires grew larger. 

WIFIRE's Firemap software was developed and tested in conjunction with major fire departments in Los Angeles, Ventura and Orange Counties and is available to departments across California for their initial attack on a fire. 

"To know that this is exactly where the fire is right now and this is the direction that it's going is extremely valuable information," Cal Fire Battalion Chief David Krussow told CBS News Sacramento about the abilities of the mountain cameras. "It truly a game changer." 

In addition to working on the problem of reaction time, the lab is also developing technology to keep prescribed fires, which are intentionally set to help clear debris from the forest, more predictable and under control.  

Nationally there is a movement to embrace more prescribed fire to better manage the risk of fire.  However, there is a large backlog for setting those fires. In California, for example, the state wants to burn a million acres a year by 2025 but last year only 110,000 acres were burnt. 

The use of prescribed fire is also under major scrutiny after one got out of control last year and accidentally led to the largest wildfire in New Mexico history. 

Building on technology developed at the Los Alamos National Laboratory, Altintas and her colleagues are developing highly detailed mapping software that shows things like how much vegetation is in a forest, the height of the tree canopy, and how dry it is. 

"Knowledge of what's there and the local fire environment becomes very important," Altintas said. 

Using artificial intelligence, they can run a computer model that shows how a prescribed fire will behave in the actual environment before it's even set and, potentially, reduce the risk that a prescribed burn will get out of control. 

"The wildland fire problem is solvable if we do some things right collaboratively," Altintas added. 

In:
• Artificial Intelligence
• Wildfire Smoke
• Wildfires