Even for the Midwest, the earth beneath Ashtabula, Ohio, was notable for its utter stillness. No one in the blue-collar coal-port town had ever felt an earthquake, nor had the older generations passed down any stories of fleeting tremors. That’s why John Armbruster, a seismologist from the Lamont-Doherty Earth Observatory, left Palisades, New York, in 1987 with a load of seismographs and steered for this town on Lake Erie’s southern bank. A 3.8 had rattled Ashtabula’s industrial area and its residents’ nerves. The tremors came one after another, swarming over a countryside once thought immune to them. This kind of thing simply hadn’t happened here until recently. Armbruster wanted to know why.

As soon as he arrived with his team, Armbruster arranged a seismic network near the epicenters. The highly sensitive equipment, however, was picking up confounding levels of meaningless background noise usually attributable to a construction site. “I asked around, and they said it must be the pumps at a well where they’re injecting waste into the ground. And I thought, ‘Oh, somebody is injecting waste in the ground right near the earthquakes? Interesting!’ ”

His suspicions sharpened when he learned that the disposal well was not only 700 yards from the epicenters, but it had also started injecting a year before the first tremors. Nearly 16 million gallons had seeped into a sandstone formation more than a mile underground. And there was a fault system less than a mile away. “They kept pumping, and they kept denying that they’re doing anything,” he says. “Of course, if they stop pumping, they’re admitting it.”

Alan-Branchett SECOND THOUGHTS: As an Azle city councilman, Alan Brundrett welcomed the oil and gas industry. As mayor, he’s not sure it should stay.

As Armbruster continued to monitor the area, Ashtabula gradually quieted down. A burst of seismic activity was reported in 1995, but otherwise it looked like the worst had passed. The disposal well was shut down in 1994. But the legacy of the fluid pumped below was lasting. A massive underground plume had been on the move those last few years. Imagine the pressure exerted by nearly 90 million gallons of wastewater, pressing it downward and outward. Its movement through the porous rock formation may be imperceptibly slow, but it can travel miles if volumes are large enough.

In 2001, some 15 years after the first drop entered the disposal well in Ashtabula, and seven years after the last, the area experienced a 4.3 earthquake—its biggest yet. The sprawling slug of waste had traveled nearly four miles to a parallel fault. When it got there, the wastewater reduced the friction keeping the critically stressed fault stable. It was akin to placing a little extra pressure on the trigger of a loaded gun.  Armbruster and a colleague wrote a peer-reviewed paper about the tremors for the Bulletin of the Seismological Society of America in 2004. That’s why Ohio reached out to him again in 2011 to report another suspicious sequence of earthquakes.

An hour south of Ashtabula, in Youngstown, bewildered residents were reporting unprecedented seismic activity. Armbruster was technically retired but was curious to see if it was a repeat episode of induced seismicity. The epicenters, he found, were all within a mile of a disposal well, which was pumping straight into a fault system. That explained why the tremors began within weeks, not months, of its first injection. “If someone gave me $3 or $4 million and said, ‘Make some earthquakes,’ I would do what they did in Youngstown,” Armbruster says.

Pennsylvania’s Marcellus Shale was in the midst of a natural gas bonanza. But it hosted few wastewater disposal wells because a less-permissive EPA had oversight in the state. As a result, much of the waste was crossing the border to off-load in wells near Youngstown.

When Armbruster told state officials about how close the epicenters were to the disposal well, it was shut down three days later. But on New Year’s Eve, 24 hours after its closure was announced, a 3.9 hit Youngstown. Governor John Kasich ordered three other disposal wells in the area to suspend operations indefinitely. 

The story was much the same in Arkansas’ Fayetteville Shale, near the towns of Greenbrier and Guy. A disposal well intersected a known fault and set off a swarm of tremors, culminating in a magnitude 4.7. Arkansas now maintains a 1,000-square-mile moratorium zone for disposal wells around the seismically active area, encompassing nearly half of the shale play.

Cleburne, just south of Fort Worth, saw 50 small earthquakes over a six-month span in 2009, all within about a mile of two disposal wells. According to a draft internal EPA report obtained by EnergyWire, the Railroad Commission worked with Chesapeake Energy, which voluntarily shut one well down. The commission allegedly weighed in again when another earthquake swarm touched off near Dallas–Fort Worth International Airport. Residents called 911, reporting “loud booming noises and shaking of the walls and furniture,” and Chesapeake once more plugged a nearby well sited next to a fault. In both cases, the company denied that the evidence against it was conclusive.

Far to the east, the town of Timpson experienced a 4.8 on May 17, 2012, the most powerful earthquake on record for the region. Residents reported white-tailed-deer busts falling from mantels and collapsing brick walls. The epicenters were less than two miles from disposal wells that had injected more than a billion gallons of wastewater since 2006. 

Cliff Frohlich, a University of Texas researcher, identified a fault system nearby but didn’t have access to enough granular geologic mapping to identify a specific culprit. That’s often the problem in defining what constitutes, as the Railroad Commission puts it, “definitive” proof of man-made earthquakes. Oil and gas companies may map these areas carefully. But geologic knowledge is the competitive edge in the oil and gas fields, and the price of gathering it is dear. Those who possess it aren’t inclined to share. 

This is doubly true when liability issues arise. Class-action lawsuits have already been filed against well operators in Cleburne and Arkansas. “It’s not going to happen,” Armbruster says. “If the operator of a well starts cooperating with me, he’s admitting that he’s causing earthquakes. His lawyers are telling him not to do it.” 

That means researchers don’t get their hands on data that could help them identify perfectly situated faults. They make do with what is publicly available, and the result is often an earthquake swarm tied to a disposal well by proximity, timing, and the historical absence of seismicity. Bolstering the connection is the grand sweep of seismic activity in the central and eastern United States. Between 1967 and 2000, the regions averaged 21 earthquakes per year greater than a magnitude 3. From 2010 to 2013, as the shale play spread from North Texas, there were 450, and 188 in 2011 alone. It’s all compelling circumstantial evidence, yet it leaves room for plausible deniability. 

And there’s no shortage of disagreement about some of the most basic questions surrounding man-made earthquakes. How destructive can they get? Is it possible to determine how much stress exists on a particular fault? Can they be brought under control? Starting in 1969, the answer looked as if it could be “yes.” USGS partnered with Chevron for a first-of-its-kind experiment. Chevron’s oil-field water-injection operation was triggering earthquakes in Rangely, Colorado. Researchers discovered they could cycle the tremors off and on by tweaking injection rates. A USGS researcher recently suggested that this might be one way to handle them. But his peers feared the unforeseen.

“Some people thought this was a dangerous conclusion that could encourage behavior that might not be safe to the public,” Armbruster says.

The danger lies in the large earthquake accidentally set off, and its seismic waves triggering an even bigger one. The events of November 6, 2011, upended much of what the scientific community thought it knew and posed another troubling question: how much time can elapse between the first injection and a seismic event?

A 5.7 struck near Prague, Oklahoma, a small town west of Oklahoma City. It destroyed 14 homes, injured two people, buckled a federal highway, and, at St. Gregory’s University, toppled one of Benedictine Hall’s nearly century-old turrets. The students were having a homecoming dance inside when it came down at 10:53 that night. As they ran from the building, bricks rained down. Governor Mary Fallin sought a federal disaster declaration. Felt as far away as Chicago, it was the strongest earthquake in Oklahoma’s recorded history, and the most powerful ever linked to a disposal well. It was presaged by a series of smaller tremors that began in 2010, with epicenters some 200 meters from active disposal wells. 

In complete contrast to other cases in which disposal operations were followed closely by seismic activity, the wells here had started up in 1993. It had taken nearly two decades of consistent injection to raise the subsurface pressure high enough to induce an earthquake. This changed everything. In dry but chilling academic language, researchers from the University of Oklahoma and Columbia University wrote that the incident should cause the scientific community to reconsider “the maximum possible size of injection-induced earthquakes … and the time scale considered diagnostic of induced seismicity.”

Before Japan’s Fukushima Daiichi nuclear plant went into meltdown in 2011, European countries like Germany and Belgium relied heavily on nuclear power. In the wake of that disaster, some countries began phasing out their nuclear programs and mothballing reactors. It’s a cautionary tale, Armbruster says, that should be heeded. “People in the industry should worry more than they do now,” he says. “There could be a case where it’s so obvious that fracking and disposing of waste causes an earthquake that kills 100 people. And overnight, the rules change.”

The most sophisticated investigation of an earthquake swarm the field has ever seen is taking place in Reno and Azle right now. Seismographs have been installed across the area, and have transmitted precise locations and depths of the epicenters. Before, the closest seismic station was nearly 60 miles to the south. The data it produced was useful, but the locations were inexact, with margins of error measured in kilometers. In the early maps, the earthquakes were sprawled across the area, south near Azle, north of Reno, and west near Springtown. But once Heather DeShon and Brian Stump set up their sensors, the buckshot spread collapsed into a tight little pattern just northwest of Reno, not far from Georgianna McKee and the Yins. It was all a mile or so from XTO Energy’s disposal well.

DeShon and Stump are careful about pointing fingers, and their speech is littered with caveats. They have correlation, but what they need is specific, subsurface data. Stump says they need it to create a model that helps them understand how fluids get from one place to another, critical information for addressing induced seismicity and triggering. But it isn’t clear they’ll get that data from the companies drilling in the area.

“They might not collect something we think is important,” DeShon says, “so maybe the data isn’t even there.” Stump adds that there is economic benefit for companies that track the data he and DeShon need. “Faults can be pathways for fluids, so you don’t want to drill near faults.” 

The industry has remained mum on the issue and the growing evidence that its practices have caused damaging earthquakes. Major operators like Chesapeake Energy declined to comment. Others didn’t respond at all. XTO Energy, an ExxonMobil subsidiary, issued a statement expressing support for SMU’s research. It did not respond to questions about a possible role for its disposal well in the tremors.

The ability to say with forensic certitude that a specific well triggered a particular fault may, in many cases, be impossible. Scientists know that the rate of felt earthquakes has risen by multiples. They know that in nearly every instance, shutting down a disposal well has resulted in a sharp drop or complete cessation of the tremors. And they know that Earth’s crust is a geologic collage, assembled by the subduction of land deep beneath the surface, by the rise of mountains, and by the volcanism welding it all together. The boundaries of these crustal plates were buried and ancient stresses accreted. The rocks became dead and cold, and their energy was stored away. Maybe wastewater found the old boundaries and something finally gave. The fault would have moved only a centimeter each time, but it would have done so at a speed of several kilometers per second and sent waves through the earth like ripples through water.

• • •

North of Reno, a tanker truck down shifts and rolls onto the gravel entryway of the XTO disposal well. Its load sits low and heavy on the rear axle, and it kicks up a scrim of caliche dust as it crosses the site and pulls up behind another tanker, next to a battery of seven 40-foot tanks. A hose is attached to a nozzle at the rear of the tanker truck. The air is filled with the thrum of an engine that drowns out the idling semis. The driver climbs out of the cab and speaks with a man in a hard hat as thousands of gallons of fracking fluid and water sluice through the hose and down nearly two miles underground. 

In 15 minutes, it’s empty. The tanker pulls onto Knob Hill Road, and another takes its place. In the first nine months of 2013, some 3 million barrels of wastewater were dumped into the well. According to Railroad Commission records, this pace has been maintained since the well was drilled in 2009. And XTO says the well continues to operate normally.

A few miles south, Georgianna McKee putters around her kitchen, putting groceries away, talking about earthquakes as though they are some form of inclement weather. She hasn’t felt many in the last month. It’s quiet again here. The shaking earth doesn’t wake her anymore. SMU’s sensors still detect seismicity, but at a level too faint to be felt by humans. But no one—not the scientists, not the oil and gas companies—can tell McKee it’s over. No one can say this isn’t the lull other small towns like Ashtabula have felt before. 

 When they kept pumping, the quakes returned. And when they stopped pumping, the quakes sometimes got stronger. The waste moved implacably through the earth years later, following paths science can’t yet track, releasing pent-up stresses whose magnitudes it can’t yet predict, slowly squeezing the trigger of a gun they can’t see.