Physical, social, and biological attributes for improved understanding and prediction of wildfires

Photo Credit

USDA photo by Kari Greer

Wildfire hazards have increased worldwide, and while it’s become easier to predict and respond to lightning-caused wildfire ignitions, human-caused ignitions are still difficult to predict, prevent, and prepare for. Part of this is due to ongoing expansion of residential development into the wildland urban interface, as extreme weather creates conditions more favorable to wildfire. Because of this, the need to identify the areas most vulnerable to human-caused wildfire hazard grows more pressing each year. 

The best source of fire ignition data in the U.S. is the Fire Program Analysis Fire Occurrence Database, which researchers have recently augmented with a complementary dataset of 267 additional attributes related to climate, weather and fire danger, topography, land cover and vegetation, jurisdiction and management, infrastructure, and social context. Each attribute is linked to the date and location of ignitions that became fire events between 1992-2020. 

With the help of AI and machine learning, researchers and analysts can use this dataset to build our understanding around the factors and conditions that influence wildfire ignitions. Users of this database can create visuals that easily communicate the distribution of ignition data across time and space. Research possibilities are endless, and the attribute data will provide more in-depth explorations of past fires, the conditions they started from, and insights around their outcomes within specific social and ecological contexts. By understanding fires of the past, we can better predict, prepare for, and even detect wildfires of the future.

The Forest Service is working through a years-long, science-based roadmap for creating safer and more fire-resilient landscapes and communities, reducing wildfire risk to critical infrastructure and natural resources, and fostering strong collaboration across all lands.