An analysis of a string of three deadly earthquakes that struck Italy in 2016 suggests that they occurred in a sequence constrained by their geology.

The conclusion, which has provoked some excitement among earthquake researchers, raises the tantalizing possibility that seismologists could produce useful forecasts of the quakes that follow this type of quake — called a sequence quake — potentially saving lives.

But many challenges remain, including how to helpfully communicate risk to people who might be affected.

Currently, seismologists can forecast earthquakes only in the vaguest of terms — say, a 30% chance over a large region in the next 50 years.

Stop-start shocks

Most earthquakes take the form of a single large quake followed by aftershocks of decreasing size. But in an event such as the 2016 Italian quake — known as a sequence quake — energy is released in a stop-start manner, in which several large quakes are interspersed by smaller aftershocks. Scientists aren’t sure why this happens.

The latest research — which was published in August ¹ and will be presented at the American Geophysical Union Fall Meeting in Washington DC in December — describes an underlying arrangement of cross-cutting faults, which act as barriers and prevent the earthquake strain in the larger, major fault, from being released in one go.

The results suggest that the arrival of underground water and gas, driven by the build-up of strain, release these barriers — and that tracking the movement of underground fluids could help to provide a warning of subsequent quakes.

Forecasting where and when subsequent quakes might happen could allow people living in the path of any expected, follow-up earthquakes to evacuate in time. This could be invaluable, given that subsequent quakes can be big and deadly: in a sequence quake that hit the Indonesian island of Lombok in July and August, the second quake killed some 460 people.

Sequence quakes occur in all tectonically active areas of the world, but they are thought to be more prevalent in geologically young fault systems. In Italy’s Apennine mountains, which run the length of the country, they occur every few decades, most recently in 2016, 1997 and 1979.

A deadly trio

More than 300 people died between 24 August and 30 October 2016, when three earthquakes hit the region, each larger than magnitude 6. The small, historic town of Amatrice was badly damaged by the first quake: more than three-quarters of its buildings were flattened and 299 people died.

“Essentially, we can consider sequence quakes as ‘failed’ big earthquakes,” says Richard Walters, a geophysicist at Durham University, UK, who led the research. “The initial stress conditions are the same, but the cascading rupture of multiple segments takes place over days to weeks instead of over seconds.”

To find out why, Walters and his colleagues took advantage of the wealth of satellite data that captured the Italian quakes.

The satellites — part of Europe’s Sentinel Earth-observing constellation — provided images of the shape of the ground surface. Because the data were gathered roughly every 1.5 days, the scientists were able to compare images from before and after each quake, and calculate exactly how the ground had moved.

On the basis of satellite pictures — and seismological and ground-based measurements — Walters’ team reports that a network of smaller, cross-cutting faults underlies the Apennine region. The researchers say that these small faults act as barriers to the rupture process, preventing the major faults from being ‘unzipped’ in one go.

Had the faults all failed in one go, the region would have experienced a single earthquake with a magnitude of about 6.7 — some 50% bigger than the largest of the three quakes that struck.

Instead, the energy was released in a sequence of three quakes over the course of a few weeks.

Under pressure

In between the larger quakes, the scientists observed a wave of thousands of small aftershocks, which crept northwards at a rate of around 100 metres a day. The team found that this matched the expected speed at which naturally occurring underground water and gas would move, driven by unreleased strain, suggesting that the pressure changes associated with the fluid movement are generating the small aftershocks.

“The pattern of small aftershocks suggests that each subsequent quake is triggered by the increased pressure associated with fluids being pumped through the network of minor faults,” says Walters.

The second earthquake, two months after the first, occurred exactly when the aftershocks — and fluid, as predicted by the researchers’ models — arrived at the next major fault.

Walters and his colleagues suspect that the whole of the region’s major fault was ready to give way, but the cross-cutting smaller faults held the energy back until the underground fluids arrived, triggering the next movement. “The fluids are being driven by pressure changes. When they reach a fault, the increased pressure ‘unclamps’ the fault and allows it to move,” says Walters.

Nicola D’Agostino, a geoscientist at the National Institute of Geophysics and Volcanology in Rome, thinks it plausible that such a mechanism could explain why the quakes occurred in a sequence.

Data challenge

If underground fluids are the trigger for sequence quakes, then monitoring their movement after the first quake in a predicted sequence could give clues about the timing and location of subsequent quakes, says Walters, potentially allowing a forecast.

Stephen Hicks, an earthquake scientist at the University of Southampton, UK, agrees. “The challenge will be to monitor and interpret the data quickly enough to provide a meaningful forecast,” he says. “But it is possible that in the future we can exploit machine-learning technology, to help us rapidly process lots of different earthquake scenarios.”

The underground fault network would need to be well mapped, and a good seismometer network put in place. The Apennine region would be a prime candidate for this kind of detailed monitoring system, says Walters, but many other earthquake-prone parts of the world could ultimately benefit, too.

Seismologists would also need high-powered computer simulations of fluid movement in the fault network. Walters thinks that with appropriate investment and political will, such a system could be established within a decade.

Communicating risk

Hicks thinks that the findings will change the way geoscientists work. “Normally we don’t try and interpret aftershocks until later, but I think this will spur scientists into analysing more subtle features in real time,” he says. He also thinks that the work could apply to other types of earthquake. “Similar barrier mechanisms could be influencing big earthquakes in subduction zones.”

Roland Bürgmann, a geoscientist at the University of California, Berkeley, says that such detailed understanding of the fault system and movement of fluids could make a big difference to improving short-term forecasts of earthquake hazards.

D’Agostino agrees that it’s theoretically possible to forecast the later ruptures in sequence quakes, but thinks it could be tricky to predict whether a quake is a one-off event or whether it is the start of a sequence.

And even if scientists could produce a ‘quake forecast’, communicating the risk is fraught with difficulty, as Italy’s 2009 L’Aquila quake tragedy — which killed some 300 people — showed. This was not a sequence quake, but it does highlight the challenges of communicating earthquake risk.

In the months preceding that quake, the region experienced hundreds of minor earthquakes, sparking rumours of an imminent, larger quake. In line with their assessment of the scientific evidence, geoscientists downplayed the rumours, saying minor earthquakes weren’t sure predictors of major ones.

But days later, a deadly magnitude-6.3 quake struck. The geoscientists ended up in court , accused of providing “incomplete, imprecise and contradictory information” information to the public — although the initial guilty verdict was ultimately overturned .

“The L’Aquila quake showed just how challenging it is to communicate uncertainty,” says Hicks. “If we can start to forecast earthquakes over the timescale of weeks then we will have to be careful not to cause too much panic or give false alarms.”