Wildfires and AI: How tech can help in Canada’s natural disaster fight

Click to play video: 'Wildfires and AI: How tech can help in Canada’s natural disaster fight'
Wildfires and AI: How tech can help in Canada’s natural disaster fight
With wildfire season in full effect in Western Canada, we turn now to the ways in which AI can help prevent their spread. As Brody Ratcliffe reports, the latest took for information could come from California. – May 16, 2023

Wildfire season across western provinces has affected several communities very early in the season, but a look outside of the country could offer a solution to help prevent future fires.

Alberta’s wildfire season has been hard and fast, burning well over 500,000 hectares across the province already, putting it into a state of emergency on May 6.

B.C. has been getting hit on two fronts, contending with both wildfires and flooding.

The B.C. River Forecast Centre anticipated further accelerated snowmelt in the mountains, creating an elevated flood risk across the province.

And Saskatchewan has not been left out, seeing a high number of wildfires across the province, almost doubling the five-year average, with the Saskatchewan Public Safety Agency saying almost all of them have been caused by people.

Story continues below advertisement

To put that in perspective, the Prairie province has seen 163 fires this year so far, with 2022 seeing 119 wildfires by June 7.

Neal Driscoll, a professor of geology and geophysics at the University of California San Diego, is no stranger to the destruction that can come from wildfires, sitting as the director for ALERTCalifornia, a wireless network that uses more than 1,000 cameras to watch and monitor wildfire activity across California.

Credit: ALERTCalifornia | UC San DiegoThe command and control center located in Sacramento. This video wall uses a four-by-three screen display allows firefighters to monitor cameras across Northern California. Credit: ALERTCalifornia | UC San Diego

The public safety program is based out of the University of California San Diego but works with a range of partners from technology groups, government agencies and wildfire services.

It’s the third generation of wireless networks, the first being funded by the National Science Foundation on the heels of the Cedar fire that burned more than 270,000 acres of land in 2003 in California.

Story continues below advertisement

“It really changed the landscape. It brought us into recognition that 270,000-acre fires were possible, and the spread rate was just incredible,” Driscoll said.

He said these cameras allow early confirmation of ignition, and that they really started rolling out cameras in 2017.

“Just five or six years ago you’d have to put a battalion in a truck, drive out to where the reported fire is, or go out with an aircraft, fixed wing or rotary, and confirm ignition before you can scale your response.”

Credit: ALERTCalifornia | UC San DiegoALERTCalifornia cameras monitor Ortega Ridge near Santa Barbara, CA. The network of more than 1,000 cameras across California is an essential tool for firefighters and emergency managers. Credit: ALERTCalifornia | UC San Diego

He said this puts firefighters on the offensive, rather than allowing the fire to get too large, which puts them on the defensive.

Breaking news from Canada and around the world sent to your email, as it happens.

Driscoll said they are also mapping out the fire threat region using light detection and ranging data, as well as measuring the health of the forest.

Story continues below advertisement

He explained the plan as three legs of a stool:

  • Characterize the fuels (trees and shrubbery) and provide data for decision-making.
  • Sensors for early confirmation and how to respond during an event.
  • Surveys after the event along the burn scours to know more about the recovery.

To help manage the volume of camera feeds that ALERTCalifornia has, an artificial intelligence was trained for months to watch for anomalies like smoke and flag them to the Watch Stander for them to look at and verify.

Credit: ALERTCalifornia | UC San DiegoALERTCalifornia Camera captures the Mill Fire near the town of Weed in Northern California on Sept. 2, 2022. Credit: ALERTCalifornia | UC San Diego

Driscoll said they have a feature where once an anomaly is found, any camera that can look in that direction can spin and focus on that point.

“We have many firefighters that are on our team and advise us on how to make systems better.”

Story continues below advertisement

He said the response from emergency responders to this system has been overwhelmingly positive.

“Once they use the tool many of them will tell you that they don’t know how they sat in the watch room or incident command without it. It’s one more tool in the toolbox.”

Credit: ALERTCalifornia | UC San DiegoALERTCalifornia wildfire camera captures dramatic scene during the Radford fire near Big Bear, Calif., on Sept. 26, 2022. Credit: ALERTCalifornia | UC San Diego

Driscoll said this system is being set up in other states in the west, but stressed the importance of having it kept in the university environment to make it open source.

In terms of whether this model is feasible in places like B.C., Alberta or Saskatchewan, Driscoll said they recommend creating pilot projects in high-threat regions and building off of those.

“We have a number of sensors, we have forward-looking infrared, we have multispectral, hyperspectral that gives us better constraints on the fire perimeter, especially in smoky conditions. And as you know, the sky turns orange, even in the middle of the day, it’s frightening.”

Story continues below advertisement

Driscoll said they have portable instruments that sit as sentinels and don’t impact the ground or habitats.

“We’re very sensitive to leaving no scar on the landscape.”

He said another benefit of the university model is only charging for services rendered with no profit-making.

Driscoll said they’d like to offer this same model to other universities around the world, and allow those universities to run the platform and help create the next generation of computer engineers, field engineers and meteorologists.

Click to play video: 'Canada’s Wildfire Forecast'
Canada’s Wildfire Forecast

He said this will help in mitigating the impact of future wildfires.

“Trust us, we’re moving the needle a very small amount, but we have to keep trying and we have to keep using technology and data to drive our decisions.”

Story continues below advertisement

Driscoll said another benefit of keeping this system open source is allowing the public to access these cameras as well.

“The system is designed to keep firefighters safe while they keep us safe, but it also gives the public the ability, and enables them to make better decisions and to know when the threat is close by.”

Neil Sahota, the United Nations’ artificial intelligence advisor, said there has been a lot of work with AI over the past couple of years, saying it’s been used for predicting wildfires, along with robotics to help first responders on the front lines.

Shark Robotics I think has created six or seven different types of robots to actually assist firefighting,” Sahota said.

He said some are there to help transport equipment in difficult to get to areas, and others can actually be deployed to fight the wildfires.

“They have other robots that are actually designed to work with the human firefighters side by side.”

He said systems like ALERTCalifornia have helped automate the process.

Click to play video: 'Alberta wildfires: Officials encourage people not to tackle fires themselves, but to reach out to authorities to help'
Alberta wildfires: Officials encourage people not to tackle fires themselves, but to reach out to authorities to help

“Historically you have essentially human spotters, literally out there with binoculars scanning around trying to see the early instances of a wildfire.”

Story continues below advertisement

He said the level of precision with these cameras has increased, saying they can detect even smaller fires, and can potentially delineate between a wildfire and a campfire.

Sahota recognized that cost will always be part of the discussion, noting that cameras can be cheap, but robots are a different story.

He said better data was needed, and that detailed terrain information could help predict where wildfires start.

“It becomes a question of how serious of a threat do you think wildfires are and how much damage do you think they’ll actually do? Is that worth the cost of prevention?”

Sahota said there are an average of about 10,000 wildfires a day around the world. He said that number is going up, which is a concern.

He said it’s more than just the damage to homes and the wrecked lives that need to be taken into consideration with wildfires — it’s also the damage to the environment.

“If you actually do a good job in preventing, it’s hard to measure the return on that because less bad stuff happens.”

Sahota believes the cost of prevention is worth the return, saying we are used to being reactive.

— with files from Andrew Benson, Simon Little and Emily Mertz


Sponsored content