New technique could help authorities conduct triage in multiple-blaze scenarios

UCI is home to a new kind of firefighting squad. But rather than a hose or pickaxe, their tool of choice is an advanced algorithm. A team of university scientists have successfully applied the emerging field of machine learning to predict which wildfires pose the greatest threat to people, wildlife and property. Their work is especially vital for California where fires have caused massive destruction with greater frequency.

Identifying and weighing key data points at an event’s beginning to project its final outcome, is the idea behind the team’s application of artificial intelligence to fire prediction. Their algorithm analyzes factors on climate conditions and vegetation at a fire’s starting point to predict its final size, with a current success rate of 50 percent.

It’s possible for people to run such calculations manually. However, as the project’s lead author Shane Coffield states, their machine-learning system is “really much faster and more efficient, especially for considering multiple fires simultaneously."

Lead author, Shane Coffield
Better predictive data can be invaluable in helping increasingly overtaxed firefighting authorities allocate their scarce resources, especially in the case of multiple, concurrent fire outbreaks. “Only a few are going to get really big and account for most of the burned area. We’re focused on identifying specific ignitions that pose the greatest risk of getting out of control,” says Coffield.

The team used Alaska as a study area for its rash of concurrent fires. However, cluster outbreaks are also afflicting The Golden State with greater frequency, a tragic effect of the Earth’s changing climate. Seven of the state’s ten most-destructive fires have occurred in the past decade. “Camp Fire” in 2018 was California’s most-devastating wildfire by far, killing 86, burning over 150,000 acres and devastating countless wildlife species.

As fires, including cluster outbreaks, occur with increasing frequency, the innovation of the UCI team could be an important aid in promoting community safety. Here’s to their efforts, and to more-effective firefighting through math and science.