By Roberta Burkhart
Pittsburgh Post-Gazette
MORGANTOWN, W.Va. — The wildfires that shrouded Pittsburgh in smoke in mid-July were burning hundreds of miles away. What if satellites could have detected, tracked and deployed firefighters to fight them sooner?
Researchers at West Virginia University are developing an artificial intelligence system they hope will give firefighters a critical advantage: satellites that can detect wildfires, think for themselves and reposition in space to keep watch as the flames spread.
| CHECKLIST: The First 5 Minutes: A fireground communications checklist
The concept, known as WildFIRE-DS, remains a proof of concept demonstrated through computer simulations rather than hardware already flying in space, explained Hang Woon Lee, director of WVU’s Space Systems Operations Research Laboratory and an assistant professor in the Benjamin M. Statler College of Engineering and Mineral Resources.
But team members say it represents a significant step toward making future wildfire-monitoring satellites more responsive while reducing the time it takes to get critical information into firefighters’ hands.
If we are to grant satellites the ability to reposition themselves, we should be able to capture more information out of the wildfire events,” Lee said.
The research, led by aerospace engineering doctoral researcher Brycen Pearl with Lee and undergraduate researcher Joshua Warner, appears in the Journal of Aerospace Information Systems.
While many are accustomed to thinking of satellites as passive observers, Pearl says the idea is surprisingly straightforward. The team is studying how an AI program could help satellites detect and confirm burgeoning fires — especially those in remote areas — and then reposition themselves to better monitor the blaze as it develops.
As satellites repeatedly circle the planet, onboard AI analyzes images for signs of wildfire. It would continuously compare new images with previous observations, increasing or decreasing its confidence that a fire is actually burning. If confidence becomes high enough, the satellites can determine whether to move to get better views of the fire.
Current satellite constellations for wildfire monitoring already promise dramatic improvements in early detection. But, Pearl said, most of those systems rely on satellites traveling in fixed orbits.
“The major difference here is the fact that the satellites can reposition themselves,” he said.
Today’s satellites already carry propulsion systems capable of maneuvering, although they typically use them only to avoid collisions with space debris or other satellites.
Future systems could potentially use similar technology as part of normal operations, although doing so presents new engineering challenges because every maneuver consumes valuable propellant. Finding the balance between conserving fuel and collecting more useful wildfire data remains one of the team’s biggest areas for future research, Lee said.
The researchers stress that repositioning satellites would not necessarily detect fires earlier than existing systems. Instead, they would provide firefighters with richer, more frequent information after a blaze is discovered.
One of the main benefits “is the amount of data that can be obtained over time,” Pearl said.
“As the fire is moving, the firefighters are going to want to know which direction is it moving? How fast? How intense is it in these areas?” Pearl said. “Using that information, they can determine who they need to evacuate, who they need to relocate, and what resources they need to dedicate to different parts of the fire.”
Rather than replacing drones, ground sensors or camera networks already being deployed in fire-prone regions, the researchers see maneuverable satellites as another layer of information, creating a “sensor web” that combines data from all these sources, Pearl said.
Ground cameras can detect very small fires shortly after ignition, Lee said, while satellites excel at covering enormous areas where no other monitoring infrastructure exists. But false alarms are still among the biggest technical hurdles.
“AI do hallucinate,” Lee said.
Because every orbital maneuver consumes fuel and shortens a satellite’s operational life, researchers must be highly confident a wildfire actually exists before commanding satellites to change course.
During simulations, he said, more than 20,000 low-confidence targets were evaluated, with only a little more than 100 ultimately reaching the highest confidence level.
Perhaps the most ambitious aspect of the project extends beyond repositioning satellites altogether.
Traditional Earth-observing satellites largely function as data collectors, capturing images to send back to ground stations for people to decide what should happen next.
The WVU researchers envision eliminating much of that delay by building in onboard computing powerful enough to run both AI models and sophisticated optimization software in space. Both researchers estimate the technology could become practical within roughly the next decade, although much work remains.
“The uniqueness of our automating the Earth observation pipeline comes with the concept of integrating satellite maneuverability,” Lee said. “If we infuse maneuverability, the benefit is significant.
© 2026 the Pittsburgh Post-Gazette.
Visit www.post-gazette.com.
Distributed by Tribune Content Agency, LLC.
