Predicting a wildfire's next move? USC researchers using AI to forecast fire's likely path

Researchers at the University of Southern California are developing a way to accurately forecast how a fire will spread using artificial intelligence.

The model combines AI with real-time satellite data.

"There are models that are based on physics that you can use to forecast the progression of a wildfire, where a wildfire is going to go," said Assad Oberai , the study's co-author. The model is not too different from forecasting the weather, he added.

The model then feeds that information into a computer algorithm to forecast a fire's likely path, intensity and growth rate.

"As we look at any technology to help us do our job better, to be faster and more accurate, we're going to saves lives, protect property better," said Pasadena Fire Department Chief Chad Augustin.

The USC study comes as firefighters in California and other western states respond to an increasingly severe wildfire season. California has had 344,553 total emergency responses in 2024, according to CAL FIRE .

A total of 4,472 have been wildfires, a significant increase from 2023, which saw 3,649 wildfires.

These natural disasters also create widespread health impacts .

"All the emissions that are sort of lofted by the wildfire into the atmosphere which then cause pollution, not necessarily exactly where the wildfire is but several hundreds of thousands miles away," Oberai said.

Having had a tool like this could further prove the power of AI, potentially helping evacuation teams and firefighters on the front lines.

"It's not going to replace humans, but it's going to allow us to make much better-informed decisions," Augustin said.