Agreement between Sentinel-1 C-band backscatter change and Sentinel-2 burn severity indices across land-cover types in a low-biomass transitional eco-region

Wildfires are no longer just a summer threat confined to dense pine forests. As the climate warms, fire seasons are stretching longer and spreading into new territories, including the rolling plains and scrublands of the American South. In these transitional landscapes, where grasses mix with scattered oak trees and shrubs, measuring the damage after a fire is surprisingly difficult. Traditional methods rely on optical satellites that take pictures of the land in visible and infrared light. These images work well in forests because they can see the difference between healthy green leaves and blackened, dead ones. However, in open, low-vegetation areas, the ground itself is often bright and reflective, and grasses can grow back so quickly that the satellite sees green again within weeks, hiding the true extent of the destruction. To solve this, scientists are turning to a different kind of eye: radar. Unlike cameras that wait for sunlight, radar satellites send out their own microwave signals that bounce off the ground and vegetation, allowing them to see through smoke and clouds and detect changes in the physical structure of the land, even when the plants are still green.

In March 2022, a massive fire complex swept through central Texas, burning across more than 62,000 acres of land that included four distinct fires: the Kidd, Wheatfield, Ramsey, and Oak Mott fires. This event provided a perfect opportunity to test whether radar could see what optical cameras missed. A researcher at The University of Texas at Dallas set out to compare two types of satellite data for this specific fire: the standard optical burn maps and a new analysis using radar signals. The goal was to understand how well the radar could track the loss of vegetation structure and moisture in a landscape dominated by shrubs and grasses, rather than tall trees. The study focused on a region where the vegetation is short and the soil is often visible, a setting where traditional fire damage maps have historically struggled to be accurate.

The researchers began by mapping the fire's perimeter and then looked at the data from two different satellite systems. One system, Sentinel-2, captures images in light wavelengths that are sensitive to plant health. The other, Sentinel-1, uses radar to measure how much energy bounces back from the surface. When vegetation burns, it loses its structure and its water content. In a forest, this usually means the radar signal drops because the trees that were scattering the signal are gone. But in the open Texas scrub, the story is more complex. The analysis revealed that the radar signal behaved differently depending on the type of plant. In areas with taller, woody shrubs, the radar signal dropped significantly, indicating a loss of structure. However, in grassy areas, the signal sometimes increased. This happened because the fire removed the grass that was dampening the radar waves, exposing the rougher, drier soil underneath, which reflected more energy back to the satellite.

When the researchers compared the radar changes to the optical burn maps, they found a clear mismatch in how the two sensors saw the fire. The optical maps, which rely on the difference in plant health before and after the fire, showed that much of the fire perimeter was either unburned or only lightly burned. In reality, the fire had skipped over many patches of land, leaving a fragmented pattern of burned and unburned areas. The optical maps struggled to distinguish between areas that were truly damaged and those that were just recovering quickly. The radar data, however, provided a sharper picture of the physical changes. It showed that the cross-polarized radar signal, which is particularly sensitive to the three-dimensional structure of plants, was a much better indicator of structural loss than the standard optical maps. The radar could detect the removal of woody shrubs and the drying of the soil even when the optical sensors were confused by the rapid return of green grass.

A key finding of the study was that the way scientists calculate burn severity matters immensely in these low-biomass landscapes. The standard method for optical maps often adjusts the severity score based on how much vegetation was there before the fire. The researchers found that this adjustment actually made the data worse in the Texas scrublands. Because the pre-fire vegetation was already sparse, the adjustment amplified random noise rather than clarifying the damage. Instead, a simpler measure of absolute change in the optical data worked better when paired with the radar. The study confirmed that the radar signal and the simple optical change measure were linked, but only in areas with enough woody structure to scatter the radar waves. In the grasslands, the connection was weak, suggesting that while radar is powerful, it cannot fully replace optical data in every situation.

The study also looked at what drove the fire's intensity. They examined whether the steepness of the land or the type of vegetation played a bigger role in how badly the fire burned. The results showed that the slope of the land had almost no effect on the fire's severity. Instead, the type of fuel was the deciding factor. Areas covered in shrubs and woody plants burned with significantly higher severity than open grasslands. This supports the idea that the encroachment of woody plants like mesquite and juniper into grasslands is changing the fire regime, making fires more intense and damaging. The radar data was able to highlight these differences in structural loss, showing that the woody areas suffered a more profound change than the grassy ones.

Perhaps the most practical discovery was the ability of the radar to create a map of moisture loss. By combining the radar signals, the researchers generated an index that showed where the land had dried out the most after the fire. This map aligned well with the areas where the fire had burned, even in places where the optical maps were ambiguous. This is crucial because radar can see through the smoke and clouds that often block optical satellites immediately after a fire. It also sees the physical drying of the soil and plants before the green grass has a chance to grow back and hide the damage. This means that emergency managers could use radar to get a reliable picture of the fire's impact and the resulting water deficits in the soil much sooner than they could with traditional methods.

The researchers noted that their findings are specific to this type of low-biomass, transitional environment and should not be assumed to apply to dense forests or other ecosystems without further testing. They also pointed out that while the radar data was promising, it still needs to be validated with ground-level measurements to create a perfect scale for measuring fire severity. For now, the study demonstrates that combining radar and optical data offers a more complete view of fire damage in the Southern Great Plains. It reveals that in these semi-arid landscapes, where the ground is visible and the plants grow back fast, the physical structure of the vegetation tells a different story than the color of the leaves. By listening to the radar's echo, scientists can finally see the true shape of the burn scar, even when the land looks green again.