A gas flare burns at a fracking site in Pennsylvania.Les Stone / Reuters

At 3 a.m. on the morning of May 17, 2012, the town of Timpson, Texas, was awoken by the largest earthquake ever measured in the eastern half of the state. The 4.8-magnitude tremor shattered glass cabinets and knocked deer heads off the wall. “One respondent reported his fireplace came down inside his residence, and his south exterior brick wall ‘blew off’ the house,” reported a definitive study.

Earthquakes do not often strike Texas: Timpson is closer to tornado alley than the Pacific ring of fire. Timpson isn’t even in West Texas, where the state’s worst quakes have historically taken place. So seismologists soon felt comfortable suggesting that the quake was at least partly manmade or induced; that is, not something that would have happened by itself, but a product of the large fracking operation less than five miles from Timpson.

Yet there were dozens of other fracking wells throughout Texas that never appeared to trigger tremors. Why did Timpson shake when other towns like it didn’t? And once that happens, can you predict an induced earthquake before it strikes? Manoochehr Shirzaei, Bill Ellsworth, and a team of satellite analysts and Earth scientists have a new hypothesis, published last week in Science.

The answer could matter far beyond East Texas. Induced earthquakes have become increasingly common in the United States over the last decade, a kind of geological gurgle tailing the surge in natural-gas extraction. Between 1973 and 2008, the central United States saw only about 20 earthquakes per year with a magnitude of 3.0 or higher (which is roughly the threshold for when people start reporting shaking). But then, starting in 2009, the numbers began to surge. By 2013, there were 99 observable quakes per year. Then, in just 2014, there were 659, according to the U.S. Geological Survey. Somehow, in less than a decade, the once-solid seeming North American continent had been destabilized.

The quakes clustered around fracking wells, but seismologists came to believe the quakes aren’t actually caused by the removal of natural gas itself. Rather, they’re caused by injecting waste water left over from the fracking process back into the earth. Often this water oozes with salt, heavy metals, and toxic chemicals, so drilling companies try to place it beneath the water table.

After fracking began nearby around 2008, four wells of this type were built around Timpson—two to the east and two to the west. Some of these wells were enormous, pumping more than 2 million gallons of waste water underground every month. And in 2011 and early 2012, tiny earthquakes started to rattle the town.

According to new data from Shirzaei and Ellsworth, the quakes weren’t the only thing that was changing about outlying Timpson. According to observations they obtained from a Japanese satellite radar sensor called ALOS, all those gallons of water had started to nudge the ground. Land between the injection wells lifted up at a rate of about 3 millimeters per year. This is not an enormous distance, as far as the changing Earth goes—California slides into the Pacific every year at about 11 times that rate—but it is not negligible. In the months before the quake, the ground silently rose as tall as a sesame seed is long.

This change, though, didn’t happen where the earthquake eventually struck, on a basement fault running beneath the town. Rather, the ground lifted mostly around the wells to the east. The earth beneath the western wells didn’t seem to move.

And this is the key of Shirzaei’s hypothesis. One of the few differences between the wells was their depth: Eastern wells deposited water about half a mile below the surface into hard rock, where water eventually caused the ground to rise. To the west, the wells forced water much deeper, more than a mile underground, where, Shirzaei believes, it leaked down and irrigated the basement fault. He reached this conclusion by feeding the InSAR data into a model of the earth beneath Timpson and calculating the water pressure at certain depths.

“Wells to the west are deep and not sealed from the bottom by harder rock, so fluid injection could reach downward, lubricate faults and trigger the earthquake,” he told me. “The shallow system was sealed from bottom, so fluid had no opportunity to reach faults.”

His conclusions suggest a solution: Disposal wells shouldn’t go too deep; or, at least, they should only deposit into rock protected from the underlying fault system.

But Cliff Frohlich, a senior scientist at the University of Texas at Austin’s Institute for Geophysics, isn’t as sure. Frohlich wrote the first study about the Timpson earthquake. He praised the use of satellite data to investigate induced earthquakes, saying that InSAR, a narrow type of radar sensing used by ALOS, was an important new data source. But he wasn’t convinced by the paper’s central finding.

“When you just look at the paper, you think its gonna see a huge surface signal from InSAR where the earthquake happens. But what’s actually there is a huge surface signal 7 to 10 kilometers from where the earthquake happened,” he told me.

“What this paper didn’t show is that the earthquake was related to uplift,” he said.

“Say you looked at InSAR for 50 injection wells, five of which eventually had earthquakes and 45 of which didn’t. Then what is the relationship of the uplift to the seismicity? Is the InSAR method broadly applicable? Is it something one could apply only in retrospect to all induced earthquakes, or also ahead of time?”

If uplift was found to regularly happen before an induced earthquake, it would be a significant finding, suggesting a method for predicting earthquakes. It would also mark a change from earthquakes caused by tectonic faults, where uplift mostly tends to follow major temblors, not predate them.

“This whole business of earthquake prediction—when I was beginning my career in the 1970s, seismologists said, ‘We can crack this. Just give us funding for 20 years and we can predict earthquakes.’ Thirty-five years later, they’re not saying that as confidently. We can sometimes say where and how big, but we can’t say when,” Frolich told me. “But part of my positive feeling about this paper on the Timpson is you can’t be too critical of scientists at the beginning if they don’t cure cancer.”

Shirzaei told me that the team next intends to conduct a comprehensive InSAR study of other disposal wells to see if uplift really does correlate to seismicity across a broader area. A finding like this won’t matter as much for the people of Timpson: Shirzaei’s model suggests that when the two western wells are shut off sometime this year, the water pressure beneath the town will quickly fall. But it will matter for the hundreds of other disposal wells that dot middle America.

Earthquakes aren’t fracking’s only environmental harm: It’s unclear whether fracking wells have worsened drinking water quality across the country, and many climate scientists are increasingly worried about the greenhouse effects of natural gas itself.  But when it comes to earthquakes, satellite observation could be one tool among many. Earlier this year, Oklahoma began to closely regulate how these wells are created, and early reports indicate the number of quakes there is falling.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.