What makes wildfires so dangerous is their unpredictability, according to MU researcher Ming Xin.
“Currently, the (nation’s) firefighting, or fire management system, is not very effective and efficient,” said Xin, an associate professor of mechanical and aerospace engineering. “One of the main issues is we cannot predict where fires spread.”
Last year, almost 56,000 wildfires burned 8.5 million acres across the country, according to the National Interagency Fire Center. In November, the Camp Fire became the deadliest and most destructive wildfire on record in California, killing almost 90 people.
Xin has teamed up with two other scholars from Kansas and Georgia to create better support for wildfire management.
He is working together with Haiyang Chao, an assistant professor of aerospace engineering at the University of Kansas, and Xiaolin Hu, an associate professor of computer science at Georgia State University, to develop technology that will better monitor where wildfires will spread by using analysis of real-time weather patterns to make predictions.
The idea is to use a swarm of unmanned aerial vehicles, or drones, to collect data about wildfires and send that data back to firefighters to help them control fires more effectively.
The $1.2 million project, sponsored by the U.S. Department of Agriculture and the National Science Foundation, began last month and is in the first of four years.
How drones can help
Using drones to fight wildfires is a recent technological development and has been tested in only a few fire departments throughout the county.
The U.S. Forest Service has also been testing drones during wildfires in recent years to examine the drones’ effectiveness, according to the U.S. Department of Agriculture’s website.
Xin said his drones will facilitate more efficient fire management because they will follow a simulation that will predict fire spread.
The simulation is based on advanced data assimilation techniques that fuse thermal images of an area, wind data and terrain and vegetation data from a specific area into a dynamic simulation model, Xin said.
That data will then be transmitted to the simulation model. When the model gets the data, it can run the simulation and predict the spread of fire, Xin said.
“From the simulation, we can predict precisely where the fire will go in for the next 10 to 30 minutes,” Xin said.
He said his drones will be equipped with thermal imaging cameras to collect data of the fire front, which represents the head of the fire spread, and with sensors to estimate the wind field, which is a key factor affecting fire spread.
The traditional way to monitor wildfires was through satellites, manned aircraft and ground-based sensors, Xin said.
“(Those ways) cannot provide real-time data,” Xin explained. “They are relatively static due to their limited temporal and spatial resolutions.”
Manned aircraft cannot cover a large area while ground-based sensors are static and not flexible, he said.
However, if a large number of drones are deployed, they can provide high-resolution images and real-time data. In terms of flexibility, the drones can cover a larger area, Xin said
What affects a fire’s spread
According to Xin, the three most significant factors that affect the spread of a wildfire are an area’s terrain, vegetation and weather.
Although information on an area’s terrain and vegetation can be easily collected from a geological survey, information on an area’s weather patterns is more difficult to collect because weather is uncertain and dynamic, Xin said.
Drones can help collect that data. Specifically, Xin is interested in having drones gather data on an area’s wind patterns, the most important factor in determining fire spread, according to Xin.
During a wildfire, the wind in the affected area is often turbulent. The strength and direction of the wind can change quickly in an uncertain way, which makes predicting the fire spread more challenging. With multiple drones, it is easier to sense the wind accurately, he said.
According to Xin, the most widely used method to collect weather data is to install weather stations in the affected areas, but these are too static, lack spatial resolution and are not as mobile as drones.
Although Xin is well aware that weather and geographical conditions are different from region to region, he believes his drones, combined with data assimilation and the precise simulation model, will work well to control fires in any area. These technologies are designed to be autonomous and adaptive to different environments, Xin said.
“The technology we are developing can be used for any wildfires, not just in California,” he said.
Another unique feature of Xin’s drone development project is that the developers can test their technology on real fires.
“In Kansas, they have prescribed burns of the forest annually,” Xin said. “Every April, they will burn certain areas of forest so that we can test our technology on real fire.”
Increasing safety for firefighters
Xin hopes his technology can provide more safety to firefighters.
Traditionally, firefighters can only be aware of their current situations — they cannot see the fire as a whole, he said.
“With the system and with our predictions, we can have a big overview of the fire spread. In the safety sense, we can give firefighters more useful information,” Xin said.
Because the firefighters will be able to see the scene of the fire on a larger scale, the technology will help firefighters with their decision-making, he said.
The data collected by the drones will be available to the public through online resource sharing, publications and proposed annual workshop, he said.
Xin and his colleagues plan to start test flights of their drones at the University of Kansas Field Station next summer.
Supervising editor is Kaleigh Feldkamp.