Fear seizes the TV presenter at the very moment her TV studio starts to shake. She stops, mid-monologue, and falls silent. The shaking worsens. Other presenters seated around the awkwardly large plastic table sit stock still, save for a few worried glances left and right. Then the shaking gets stronger. The rattling of equipment above them can be heard. One broadcaster turns his gaze upward to see. The main presenter gasps. It’s time to go.

As the South Korean live TV team hastily discarded body microphones and abandoned their set, the seismic ripples of a 5.5 magnitude earthquake continued to shudder across Pohang. It was a powerful jolting. Other footage shows people running from buildings as walls collapse behind them. An entire city of half a million residents was left in shock. But this quake wasn’t a freak natural event. It was started by people.

That’s the conclusion of a report published in March by a team of experts who tried to find out what caused the event in Pohang on 15 November 2017. It left 135 people injured and 1,700 had to be temporarily relocated to emergency housing. Thousands of buildings were damaged, costing $75m ($60m). Because a geothermal drilling project had been operational nearby at the time, a big question needed to be answered: Whodunit? Humans or nature? To find out if industrial activity had set off the quake, the South Koreans called on a new breed of seismologist: the earthquake detectives.

You might also enjoy:

They are the ones tasked with combing through seismic records and industry data to see if the shaking was natural or not. It is not an easy thing to prove either way. But these scientists are now coming up with surer methods of identifying the culprit. They are forensics for the Earth.

With more drilling and fracking occurring around the world, human-induced or anthropogenic earthquakes have become an increasingly common concern. About 100,000 oil wells are now drilled every year and the use of geothermal energy, which sometimes involves injecting fluid into hot rock in order to create steam, could increase six-fold by 2050. By removing large quantities of fossil fuel or by flooding fractured rock with liquid, it’s possible to upset the balance of stresses below and set an earthquake in motion.

While we like to use metaphors like “on solid ground” in English, on a geological scale the stuff below our feet is anything but. It’s full of shifting planes of material with varying densities. There are faults and fractures, often with ribbons of fluid running through them. There are sediments, clays and bedrock. Not to mention, on an even bigger scale, gigantic tectonic plates rubbing against or pulling apart from one another. In some places, the ground is like a tower of toy bricks just waiting to topple.

Bill Ellsworth remembers the first time he saw images of people fleeing buildings as the Pohang earthquake rattled the city.

“They were very fortunate that no-one was killed, having seen some of the security cam footage,” he says. Ellsworth, from Stanford University’s Center for Induced and Triggered Seismicity, was part of the international team that investigated what happened.

The stakes for these investigators were high. They knew at the outset that to label Pohang a human-induced earthquake would be a big deal. Earthquakes are measured on the Richter scale – which is “logarithmic” – meaning an increase in one point signifies a 10-fold increase in strength. A quake of about 3 on the Richter scale would be felt by inhabitants, and 4 would be enough to knock objects off shelves. A 5.5 or higher magnitude event caused by human activity is very rare, and although it is still considered moderate, it would be enough to damage buildings.

The day after Pohang was rocked by tremors, NexGeo – the company operating the experimental geothermal power plant – denied it had any responsibility for what happened. But as Ellsworth and his team began scouring the evidence, a different story began to emerge. He and his fellow experts considered seismic data from the area as well as information from NexGeo – which cooperated with the investigation – about the drilling activity.

Geothermal plants work by using heat from the ground to generate electricity. There are various ways to do this, some for example make use of steam released directly from geothermal reservoirs. In other cases, the rock may be hot but there is not enough fluid to bring heat to the surface in the form of steam. To fracture the rock and release that heat, NexGeo planned to inject fluid into the ground.

Before the firm’s team could get to that stage, they had to drill deep into the earth. It was during this process that things went wrong.

The South Korean drillers unexpectedly hit an area of cracked and fractured rock about 3.8km down

When subterranean rock is drilled, it gets smashed into tiny particles by the drill bit and has to be removed. This is done by flushing a relatively dense fluid nicknamed “mud” down through the centre of the drill, which then exits at the bottom and flushes pulverised rock around the drill bit up to the surface. But the South Korean drillers unexpectedly hit an area of cracked and fractured rock about 3.8km (2.4 miles) down. A large quantity of the mud escaped into those cracks instead of flowing upwards. This increased the pressure in that area.

“For whatever reason there was a pathway that allowed the fluid to escape the borehole,” explains Ellsworth. By flushing even more fluid down, the drillers secured their borehole. But the huge pressure now present caused what no-one wanted: seismicity.

“It triggered some very tiny events, events that were so small they were not noted at the time,” says Ellsworth.

What the drilling team don’t appear to have realised then, but which spatial analysis of these mini earthquakes later showed, was that the drilling had actually crossed a fault line – a boundary underground where two planes of earth meet. Movement of earth can happen along these faults. That’s what causes earthquakes.

Ideally, fault lines in areas subjected to drilling or fluid injection are known about and are usually avoided. In this case, partly because there had been no indication of a fault line at the surface, the South Korean team had no idea what they had drilled into. As Ellsworth puts it: “That was very unlucky.”

“This fault was what we call critically stressed – only a small change in conditions could cause that fault to move, which is ultimately what happened.”

Those first, smaller earthquakes, overlooked at the time, were a sign that something wasn’t right. It was only a few weeks later that the 5.5 magnitude earthquake hit.

Data collected by Ellsworth and his colleagues convinced them that the event was human-induced. While there has been some debate over the results, the findings have already been accepted by the South Korean government, which says it will now dismantle the geothermal plant.

Could the drilling team have noticed the early seismicity and stopped drilling just in time? It’s possible, says Ellsworth, but they were relying on a relatively simple traffic light system to help them judge whether drilling was safe. This involves monitoring seismicity and only ceasing to drill should a certain magnitude of quake be reached. Ellsworth points out that in this case the magnitude of those mini quakes was very small, but plotting where they occurred reveals the presence of a fault. That kind of more comprehensive analysis could, in theory, have alerted drilling operators to the gravity of the situation earlier. (BBC Future has contacted NextGeo to discuss the contents of the report, but they have not responded to our request.)

But listening to the ground and making sense of the various rumblings going on below is no easy task. How do we do it? Five and a half thousand miles away in the south of England, one scientist has found himself embroiled in another seismic detective case – this time involving an oil drilling operation nestled in the pleasant countryside of Surrey.

Cold case earthquakes

Besides investigating contemporary tremors, seismologists are even discovering possible anthropogenic causes behind earthquakes that occurred decades ago. Susan Hough at the US Geological Survey has examined records of a huge 7.5 magnitude earthquake that hit Kern County, California in 1952.

Hough discovered old data online that let her analyse where drilling was located and at what depths. By building a mathematical model of the situation and noting the presence of fault lines, she was able to show that the industrial activity could indeed have released pressure, causing the quake, because oil had been removed from a reservoir right above a large fault. With that pressure gone, the fault was free to move. “To get an earthquake that big, you have to have a fault sitting there with stored stress on it,” she says. Her model even explained how the quake could occur about 100 days after initial oil production – which is almost exactly what happened.

It’s a bright afternoon in early spring when Stephen Hicks cracks open the big black box by a huge solar panel at the edge of a field. Through the hedge, a few horses look at us bemusedly before trotting off. “We’ve got five of these in the area,” he says, enthusiastically digging through the cables and components in the box to check everything is in place as it should be.

Hicks is a seismologist at Imperial College London. He has found himself leading a local investigation that aims to find the cause of a series of small earthquakes in the area. It’s not something that Surrey, with its gentle rolling hills and babbling streams, is used to. But back on 27 February the location was hit by a 3.1 magnitude quake in the early hours of the morning. That was the strongest so far and while not hugely damaging, it was an unusual event. The UK only gets two or three such quakes a year.

Because a firm called UK Oil and Gas (UKOG) has been extracting oil nearby, many locals are worried that the activity is disturbing ancient fault lines and causing the quakes. There has been a series of heated protests at the drilling site. Many have turned to scientists to see if they can prove what is really going on, which is where Hicks comes in.

“That’s what we call the digitiser,” he says, enthusiastically pointing to a small box inside the black case. “That’s just turning analogue signal into some sort of digital form and then we can later convert it into velocity, metres per second or acceleration.”

After stamping on the ground he shows me the huge spikes that pop up a few minutes later on the near-real-time chart of tremors he can check online. Having multiple instruments in the field means that incidental noise, say from passing vehicles (or stamping scientists), can be discounted. Only when tremors appear uniformly on a number of the seismic monitors does that indicate an earthquake.

The kit costs about £10,000 ($12,500) and is owned by the British Geological Survey. Hicks and his colleagues helped to set the five units in place during the summer of 2018. He’s been monitoring the signals ever since. But unlike Pohang, there doesn’t seem to be a smoking gun here.

Almost all of the 90 or so quakes Hicks has detected in the last eight months are tiny, less than magnitude 1. And they are occurring at a relatively shallow depth, about 2.5km (1.5 miles) down, but not as shallow as the drilling, which is happening at about 700m or 800m (2,300-2,600ft). Not only that, the quakes have been distributed around the area – known as the Weald basin – not clustered near the drilling site.

“We think it’s coincidence,” he says. “The swarm – we call them swarms of earthquakes, you get bunches of them in time – it’s kind of random.”

In this case, the quake mechanism is “strike-slip”, where planes of earth are moving side-by-side. Kind of like two shoes rubbing together, says Hicks. And while he doesn’t think they are human-induced, they are still interesting, because quakes like this at such a shallow depth aren’t usually recorded with such a high resolution in the UK.

“Regardless of the cause, it’s still an interesting sequence,” he says.

As we stroll through the countryside, near to the epicentre of the biggest quake – an amusingly boring field – we bump into a lady named Jackie Wilson, who is walking her dog. When that 3.1 quake hit in February, her cat “leapt off the bed”, she says.

“Somebody did come round with a petition for the locals to sign against all of this drilling going on,” she adds, “I guess it has kicked off since then, hasn’t it?”

Hicks has fielded emails, calls and tweets from locals similarly curious about what’s going on. But for now he is sticking to his conclusion that the quakes are natural. A few weeks after we meet, he and colleagues publish preliminary results of their investigation online. “Overall, we find no indicators in the earthquake parameters that would strongly suggest an induced source,” they write.

But public concern that tremors might be man-made is becoming a more common story around the world. Especially when drilling or geothermal activity happens in the same place as earthquakes.

People are clearly getting used to the concept of anthropogenic seismicity, which can be either human-induced or triggered. The latter is a slightly different condition in which earthquakes are mostly caused by tectonic activity but human activity plays a role in how they occur.

It’s natural for the public to have concerns, says Francesco Grigoli at ETH Zurich. Grigoli has studied what tools are available to earthquake detectives when they try to work out the cause of seismic disturbances.

“There is not a standard recipe for identifying any event, discriminating any event,” he explains. He and colleagues studied the Pohang quake less than a year after it occurred, but they were limited to public seismicity data from a Japanese station, many miles from the epicentre, and did not at the time have access to NexGeo’s information on drilling.

He makes the point that more open data can make a huge difference when it comes to deciding whether a quake was anthropogenic or not.

Plus, the resolution of seismic monitoring has improved greatly. If motivated to do so, drilling companies can today use highly sensitive listening arrays that hear “every last pop and crack”, says James Verdon at the University of Bristol.

“This gives us thousands, or even hundreds of thousands, of data points with which to make a much more detailed assessment of seismic hazard,” he explains.

One paper published earlier this year described how microseismic monitoring apparently helped to keep fluid-induced seismicity to a minimum at a geothermal project in Finland. In that case, the team listened carefully to small-scale seismicity, which in turn motivated them to occasionally lower the rate of fluid injection or wait for longer periods between pumping. The authors of the paper think this kept any more serious shaking at bay.

Some will never be comfortable with the idea of drilling near populated places. The potential consequences, they argue, are just too great, even if the probability of setting off a big earthquake remains small. Earthquake detectives, though, can in theory help to assess the situation while drilling is going on, not just after the fact, and raise the alarm should that drilling become dangerous. By listening carefully to what the ground is telling us, companies and governments may be better equipped to react – before it is too late.

--

Join more than one million Future fans by liking us on Facebook, or follow us on Twitter or Instagram.

If you liked this story, sign up for the weekly bbc.com features newsletter, called “If You Only Read 6 Things This Week”. A handpicked selection of stories from BBC Future, Culture, Capital, and Travel, delivered to your inbox every Friday.