California researchers use machine learning to predict growth of wildfires

IRVINE, California — Researchers at the University of California, Irvine are using machine learning to predict how big wildfires will grow. They’re using information on hand when a fire ignites to make the prediction.

“A useful analogy is to consider what makes something go viral in social media,” said lead author Shane Coffield, a UCI doctoral student in Earth system science. “We can think about what properties of a specific tweet or post might make it blow up and become really popular – and how you might predict that at the moment it’s posted or right before it’s posted.”

Viral philosophy used for fire prediction

Coffield and his fellow researchers applied the same line of thinking to an imaginary scenario in which multiple fires break out at once. As residents of the western US know, sometimes fires spark at the same time across multiple locations, particularly with our dry climate. Knowing which of the fires is going to become big or potentially threaten homes can help firefighters decide where to send more resources.

“Only a few of those fires are going to get really big and account for most of the burned area, so we have this new approach that’s focused on identifying specific ignitions that pose the greatest risk of getting out of control,” Coffield said.

Alaska used as an example

The team zeroed in on Alaska for its study. One reason was because of the existence of multiple fires at the same time in the state’s boreal forests. Boreal forests are considered critical to the global ecosystem because of their impact on climate worldwide.

Researchers used a “decision tree” algorithm to gather and access information. They used specific information from the first six days of a fire, including climate, atmospheric conditions, the types of vegetation in the area, and the vapor pressure deficit – how little moisture is in the air.  With the machine learning model, they were able to accurately predict how large the wildfires would grow about 50% of the time.

Coffield said a major advantage with the algorithmic method is speed. With each new bit of information, he said, the algorithm “learns” and adjusts its prediction.

Specific plants are indicators

Professor and co-author James Randerson said even the existence of a specific type of plant could have a major impact on whether a wildfire grows large. And,  how quickly it does so.

“Black spruce, which are dominant in Alaska, have these long, droopy branches that are designed – from an evolutionary perspective – to wick up fire,” Randerson said. “Their seeds are adapted to do well in a post-fire environment, so their strategy is to kill off everything else around them during a fire to reduce competition for their offspring.”

How Utah can benefit

While the research was conducted in Alaska, Randerson said other states like Utah should be able to benefit from the machine learning study. He said wildfires are expected to increase in number every year as a result of climate change. The ability to forecast fire growth would help local fire departments protect property and infrastructure and even save lives.

“In places like Alaska, there’s a need to limit the area affected by fire, because if we keep having these unusual, high-fire years, more carbon will be lost from the landscape, exacerbating warming,” Randerson said. “If we let the fires run away, we could be in a situation where there’s a lot of significant damage to both the climate system and ecosystems.”

Follow @KSLbbruce