How satellites and AI help fire crews fight wildfires

As wind-driven wildfires spread through the Los Angeles area in January 2025, fire-spotting technology and computer models were helping firefighters understand the rapidly changing environment they were facing.

That technology has evolved over the years, yet some techniques are very similar to those used over 100 years ago.

I have spent several decades studying combustion, including wildfire behavior and the technology used to track fires and predict where wildfires might turn. Here’s a quick tour of the key technologies used today.

Spotting fires faster

First, the fire must be discovered.

Often wildfires are reported by people seeing smoke. That hasn’t changed, but other ways fires are spotted have evolved.

In the early part of the 20th century, the newly established U.S. Forest Service built fire lookout towers around the country. The towers were topped by cabins with windows on all four walls and provided living space for the fire lookouts. The system was motivated by the Great Fire of 1910 that burned three million acres in Washington, Idaho and Montana and killed 87 people.

Today, cameras watch over many high-risk areas. California has more than 1,100 cameras watching for signs of smoke. Artificial intelligence systems continuously analyze the images to provide data for firefighters to quickly respond. AI is a way to train a computer program to recognize repetitive patterns: smoke plumes in the case of fire.

NOAA satellites paired with AI data analysis also generate alerts but over a wider area. They can detect heat signatures, map fire perimeters and burned areas, and track smoke and pollutants to assess air quality and health risks.

Forecasting fire behavior

Once a fire is spotted, one immediate task for firefighting teams is to estimate how the fire is going to behave so they can deploy their limited firefighting resources most effectively.

Fire managers have seen many fires and have a sense of the risks their regions face. Today, they also have computer simulations that combine data about the terrain, the materials burning and the weather to help predict how a fire is likely to spread.

Fuel models

Fuel models are based on the ecosystem involved, using fire history and laboratory testing. In Southern California, for example, much of the wildland fuel is chaparral, a type of shrubland with dense, rocky soil and highly flammable plants in a Mediterranean climate. Chaparral is one of the fastest-burning fuel types, and fires can spread quickly in that terrain.

For human-made structures, things are a bit more complex. The materials a house is made of—if it has wood siding, for example—and the environment around it, such as how close it is to trees or wooden fences, play an important role in how likely it is to burn and how it burns.

Weather and terrain

Terrain is also important because it influences local winds and because fire tends to run faster uphill than down. Terrain data is well known thanks to satellite imagery and can easily be incorporated into computer codes.

Weather plays another critical role in fire behavior. Fires need oxygen to burn, and the windier it is, the more oxygen is available to the fire. High winds also tend to generate embers from burning vegetation that can be blown up to five miles in the highest winds, starting spot fires that can quickly spread.

Today, large computer simulations can forecast the weather. There are global models that cover the entire Earth and local models that cover smaller areas but with better resolution that provides greater detail.

Both provide real-time data on the weather for creating fire behavior simulations.

Modeling how flames spread

Flame-spread models can then estimate the likely movement of a fire.

Scientists build these models by studying past fires and conducting laboratory experiments, combined with mathematical models that incorporate the physics of fire. With local terrain, fuel, and real-time weather information, these simulations can help fire managers predict a fire’s likely behavior.

Advanced modeling can account for fuel details such as ground-level plant growth and tree canopies, including amount of cover, tree height, and tree density. These models can estimate when a fire will reach the tree canopy and how that will affect the fire’s spread.

Forecasting helps, but wind can change fast

All these tools are made available to firefighters in computer applications and can help fire crews as they respond to wildfires.

However, wind can rapidly change speed or direction, and new fires can start in unexpected places, meaning fire managers know they have to be prepared for many possible outcomes—not just the likely outcomes they see on their computer screens.

Ultimately, during a fire, firefighting strategy is based on human judgment informed by experience, as well as science and technology.

John W. Daily is a research professor in thermo fluid sciences at the University of Colorado Boulder.

This article is republished from The Conversation under a Creative Commons license. Read the original article.