Scientists Employ AI to Forecast the Movement of Wildfires
Researchers at USC have created a new technique for accurately predicting the spread of wildfires. By integrating satellite imagery with artificial intelligence, their model represents a significant advancement in wildfire management and emergency response.
Described in a preliminary study published in Artificial Intelligence for the Earth Systems, the USC model uses satellite data to monitor a wildfire’s progression in real-time. This data is then input into a sophisticated computer algorithm that predicts the fire’s probable path, intensity, and growth rate.
Moreover, the study is timely, as California and much of the western United States face an increasingly severe wildfire season. However, numerous fires, driven by a dangerous mix of wind, drought, and extreme heat, are sweeping across the state. Notably, the Lake Fire, the largest wildfire of the year in California, has already burned over 38,000 acres in Santa Barbara County.
Bryan Shaddy on the Impact of the New Wildfire Management Model
“This model marks a significant advancement in our ability to manage wildfires,” said Bryan Shaddy, a doctoral student in the Department of Aerospace and Mechanical Engineering at the USC Viterbi School of Engineering and the study’s lead author. “By providing more accurate and timely data, our tool enhances the efforts of firefighters and evacuation teams working to control wildfires on the ground.“
The researchers began by collecting historical wildfire data from high-resolution satellite images. By analyzing the behavior of past wildfires, they tracked how each fire started, spread, and was eventually contained. This thorough analysis uncovered patterns influenced by factors such as weather, fuel sources (e.g., trees, brush), and terrain.
Next, they trained a generative AI model called a conditional Wasserstein Generative Adversarial Network (cWGAN) to simulate the impact of these factors on wildfire development over time. They taught the model to identify patterns in the satellite images that correspond with the fire spread patterns observed in their study.
Evaluation of cWGAN Model on Real California Wildfires (2020-2022)
They subsequently evaluated the cWGAN model using actual wildfires that occurred in California between 2020 and 2022 to assess its accuracy in predicting fire spread.
“By examining the behavior of past fires, we can develop a model that forecasts how future fires may spread,” explained Assad Oberai, Hughes Professor and Professor of Aerospace and Mechanical Engineering at USC Viterbi and co-author of the study.
Oberai and Shaddy were impressed by how well the cWGAN, initially trained with simplified simulated data under ideal conditions such as flat terrain and unidirectional wind, performed when tested on real California wildfires. They credit this success to the model’s use of actual wildfire data from satellite imagery, rather than relying solely on simulated data.
Oberai, whose research focuses on creating computer models to understand the physics of various phenomena, has modeled everything from turbulent airflow over aircraft wings to infectious diseases and cellular interactions within tumors. Among all these models, he finds wildfires to be particularly challenging.
Key Challenges and Modeling Needs
“Wildfires involve complex processes: fuel like grass, shrubs, or trees ignites, triggering intricate chemical reactions that produce heat and wind currents. Additionally, factors such as topography and weather affect fire behavior—fires spread less in moist conditions but can advance quickly in dry conditions,” he explained. “These processes are highly complex, chaotic, and nonlinear. Accurate modeling requires accounting for all these variables, which demands advanced computing.”
Additional co-authors include undergraduate student Valentina Calaza from the Department of Aerospace and Mechanical Engineering at USC Viterbi; Deep Ray from the University of Maryland, College Park (formerly a postdoctoral student at USC Viterbi); Angel Farguell and Adam Kochanski from San Jose State University; Jan Mandel from the University of Colorado, Denver; James Haley and Kyle Hilburn from Colorado State University, Fort Collins; and Derek Mallia from the University of Utah.
Read the original article on: TechXplore
Read more: Extreme Wildfires have Doubled in 20 Years
