IE 11 Not Supported

For optimal browsing, we recommend Chrome, Firefox or Safari browsers.

Can AI Give California an Edge in the Fight Against Wildfires?

Sonoma County, Calif., is implementing artificial intelligence technology to help emergency management workers detect wildfires before they spread out of control. Could this be the blueprint for other at-risk counties?

A forest fire burning at night.
Shutterstock/Mikhail Roop
In October 2017, the Tubbs Fire devastated Sonoma County, Calif., destroying 5,600 structures and causing 22 deaths. The extensive damage pushed county officials to investigate new technologies to help identify and prevent another wildfire of this scale.

One solution under investigation works with an existing camera network to constantly watch for signs of wildfire. As of March 17, 2021, the county of Sonoma is implementing Alchera artificial intelligence technology to make this process even more efficient.

The decision to invest in AI technology was made possible by the Federal Emergency Management Agency Hazard Mitigation Grant Program, which awarded $2.7 million for improvements and hazard mitigation, $225,000 of which was allocated specifically for AI monitoring.

According to Chris Godley, Sonoma County director of emergency management, the grant provided 75 percent of funding and local governments are expected to provide the remaining 25 percent. In this case, the remaining 25 percent was covered by an additional grant the county applied for, a Community Development Block Grant from the Department of Housing and Urban Development.

The technology itself is not an additional hardware or even software, but rather offers software as a service to which the county is subscribing, Godley said. The machine learning component of this technology is what sets it apart from the current monitoring system. Not only does it allow for 24/7 monitoring, but it should also allow for fire detection within the first five minutes, which could lead to a significant reduction in response time.

Godley described the importance of AI technology in early fire detection, especially in cases of nighttime fires and increasingly extreme weather events caused by climate change. He believes AI technology for early fire detection will see widespread adoption in the next three to five years.

“That’s a significant difference,” explained Godley. “If we can dispatch firefighters to a specific location that quickly, there’s a good chance we can get the upper hand quickly, as opposed to giving that fire a head start.”

As Robert Grey, Alchera’s product manager for the wildfire detection solution, described, the AI technology “fits in the puzzle” of early detection of wildfires, as it can be applied seamlessly across the existing camera monitoring system. There is one primary difference between the new technology and the previous monitoring system, according to Grey: The algorithm does not sleep.

Additionally, the machine learning capabilities allow the technology to determine the difference between smoke and something that looks similar, like clouds or “waterdogs,” a weather phenomenon when fog looks and behaves like smoke. For the technology to understand the difference between a threat and natural occurrence, millions of images of smoke must be collected over years of monitoring, allowing the system to make a determination based on the diffusion and flow patterns.

The system can also be set up to notify different agencies. In Sonoma County, it will notify dispatchers who can verify if a detected anomaly is a prescribed burn; if it is not a prescribed burn, they can immediately dispatch resources to that location to control and put out the fire if necessary.

Implementation will begin May 1, leading into what Godley acknowledged will be a long learning season for the algorithm. The goal is to evaluate the technology until February 2023, at which point county officials and stakeholders can decide if the technology is effectively completing its intended task.

“I think what’s important for local governments is that you will have to invest in technology,” said Godley. “This is really, essentially an experiment to see if we can make practical use of a developing technology… But it does highlight now how far in front local government has to be to serve its residents in terms of adapting and adopting new technology.”
Julia Edinger is a staff writer for Government Technology. She has a bachelor's degree in English from the University of Toledo and has since worked in publishing and media. She's currently located in Southern California.