The land is always changing, sometimes by human hands: cities are built, farms expanded, and forests logged. Other changes lie mostly outside people’s control: wildfires burn through communities, and hurricanes reshape coastlines. For most of the past four decades, observations from the Landsat satellite record show that humans have dominated changes to the U.S. landscape. Recent research revealed a shift in that trend, suggesting that disasters might be catching up.
In a NASA-funded study published in Nature Geoscience, scientists analyzed nearly 35 years of data from NASA/USGS Landsat satellites to better understand what has been shaping the continental U.S. landscape. The researchers, led by former Landsat science team member Zhe Zhu, found that “human-directed disturbances” like logging, agricultural expansion, and construction have declined, while “wild disturbances” like wildfires and hurricanes—disasters that can be influenced by human activity but are not controlled by people—have risen in frequency and intensity.
Robert Emberson, associate program manager for the NASA Disasters program and not affiliated with the study, said that understanding the forces transforming the U.S. landscape is critical for future planning. “If you know what’s causing them, you can begin to plan around disasters,” Emberson said. “Any understanding of causal factors impacts the adaptation strategy.”
This research is especially useful for policymakers working to prepare communities for resilience, he said. For example, a region expecting to see increased wildfires could strategically perform prescribed burns, remove brush or dry grass around homes, and construct new buildings with fire-resilient materials.
Between 1988 and 2022, 18 percent of the land area in the continental U.S. was disturbed at least once, the researchers found. Adding repeated disturbances, the cumulative area disturbed rises to almost 700,000 square miles, equivalent to nearly one-third of the continental U.S. Humans drove more than half of that change, clearing or developing over 446,000 square miles of land—that’s bigger than the size of Texas and California combined. For example, the animation above, composed of Landsat images from 1985 to 2025, shows the expansion of Reno, Nevada, into a previously undeveloped desert landscape.
Meanwhile, wild disturbances—disasters like wildfires, hurricanes, and landslides—drove much of the remaining change, transforming more than 165,000 square miles of the continental U.S. The Landsat images in the animation below show areas burned by wildfires in Eldorado National Forest west of California’s Lake Tahoe from 1985 to 2025. Major fires in 1992, 2014, and 2022 cleared large swathes of forest, leaving behind bare ground that slowly reforested.
Although human activity has disturbed a larger cumulative area than wild events, the trends over time are moving in opposite directions. That is, land disturbance caused directly by people has been decreasing, while wild disturbance has been increasing.
Specifically, human-directed land disturbances decreased by nearly 232 square miles (600 square kilometers) each year over the course of the study period. Researchers attribute this change to declines in construction, agricultural expansion, and logging, likely brought about by a combination of policy changes, technological improvements, and the 2008 financial crisis’s effect on construction.
In contrast, land affected by wild disturbances increased by more than 77 square miles (200 square kilometers) per year. Fire, drought-related stress, and wind disturbances all became more frequent, likely due to climate warming and other environmental factors, the study authors wrote.
“What this study basically tells me is that what we’ve been doing is not working,” said Ramakrisna Nemani, a retired NASA scientist and co-author on this study. “We have to go back and come up with new strategies on how to deal with these natural disturbances.”
The study’s findings drew on the deep archive of Landsat data, which has long been a key resource for detecting change on Earth’s surface. Think of it like a “spot-the-difference” game. Historically, identifying differences between images required scientists to manually identify the source of the change; for example, using ground observations combined with satellite imagery to determine whether a bare spot resulted from wildfires or logging. For this study, scientists trained a new machine-learning algorithm to do that differentiation work for them.
They fed the algorithm 40 years of land-change data acquired by satellites, manually inspecting and identifying changes at 50,000 locations. After a decade of work, they developed a product that achieves more than 75 percent accuracy across most disturbance types.
The resulting product details the causes of disturbance across the continental U.S. over the course of nearly 35 years. With this information, communities can analyze the past to better plan for the future. “The USA is entering a new era of disturbance,” the study authors wrote. “The challenge now is to transform our relationship with disturbance from one of control to one of coexistence.”
NASA Earth Observatory image by Lauren Dauphin, based on data from Qiu, S. et al. Animations by Ross Walter, Landsat Project Science Support. Story by Madeleine Gregory, Landsat Project Science Support.
