Data from thousands of GPS devices detects an earlier phase that heralds major earthquakes

It is the holy grail of seismologists and geologists: finding a reliable clue as to when, where, and with what magnitude an earthquake will strike. So far this century, more than one million people have died as a result of earthquakes, not counting the astronomical cost to infrastructure and the economy, particularly in developing countries. Now, French scientists have detected a precursor phase that begins hours before a major earthquake occurs. As detailed in the journal Science, they have achieved this by analyzing miniscule displacements recorded by GPS. These researchers believe that deploying detection networks around major faults could help find that holy grail.

In the 1970s, seismologists were euphoric. The accumulation of data on earthquakes, new theoretical models, and laboratory experiments led to the dream of detecting phenomena and mechanisms that heralded an earthquake. As University of California professor Roland Bürgmann says, everything indicated that “earthquakes are often preceded by precursor processes.” But enthusiasm waned: “As scientists looked harder and had better observations of these precursors, they discovered that, while they sometimes happen, they could not really be distinguished from similar processes that often occur at other times and places.” Julián García Mayordomo, an expert in earthquake geology at Spain’s Geological and Mining Institute (IGME) also recalls the complexity: “Large earthquakes occur 10 to 15 kilometers (6 to 9 miles) deep in the earth’s crust, where we have never been able to look. In addition, a major fault that produces earthquakes on the order of 6.5 or 7 is a plane that can be tens of kilometers long by 15 kilometers deep. It is a huge area where many geological processes occur. It is absolutely impossible to control. There are too many variables, which makes the phenomenon highly unpredictable.”

“Large earthquakes occur 15 to 10 kilometers deep in the Earth’s crust, where we’ve never been able to look.”

Julián García Mayordomo, expert in earthquake geology at Spain’s Geological and Mining Institute

But scientists Quentin Bletery, from the French Côte d’Azur University, and Jean-Mathieu Nocquet, from the Institut de Physique du Planète in Paris, have found a way to detect the signal of a future earthquake amid all the noise. Their idea makes use of global navigation satellite systems (GNSS), such as the United States’ GPS or the European Galileo. The entire planet is dotted with geodetic stations that include a number of sensors of interest to geologists. One of them is a GNSS module that relies on triangulation with GPS or Galileo satellites (and with the Russian GLONASS or Chinese Beidou networks) to determine its location. Fixed to the ground, the position of these stations is measured to the millimeter and is essential to map making. But the stations move and are not always in the same place: their position changes throughout the year due to global phenomena, such as continental drift, or local phenomena, such as the construction of a reservoir, land prospecting, or fracking. A large earthquake can also move them out of place and this is registered by the GPS.

What the French scientists have done is analyze the positioning data from more than 3,000 geodetic stations as the earth shook with the 90 earthquakes of magnitude 7 or greater so far this century (those in Turkey were as high as 7.8 and 7.5). More importantly, they also collected and analyzed GPS data from the 48 hours prior to each of these major tremors. Their starting hypothesis was that earthquakes begin with a precursor phase which features a slow displacement, without tremors, at the point of the fault where the hypocenter of the coming earthquake will be.

“The precursor phase is the window of time during which the tectonic blocks begin to move relative to each other, first slowly and then accelerating progressively,”

Jean-Mathieu Nocquet of the Institut de Physique du Planète de Paris

“Earthquakes are sudden slips along faults that separate two tectonic blocks,” says Nocquet, co-author of this research. Prior to the quake, the two rock masses are stuck together. “The precursor phase is the time window during which the tectonic blocks begin to move relative to each other, first slowly and then accelerating progressively to finally reach a rapid sliding velocity. Rapid sliding produces the seismic waves that cause the damage observed during major earthquakes,” the French scientist explains. Although there is some consensus on the existence of this precursor phase, there is no consensus on its key characteristics, such as its duration. For some, it lasts only a few seconds, for others it can be seen as a succession of micro-earthquakes over weeks or months. “In fact, our study suggests that the slip accelerates progressively over a few hours, around two,” he adds.

To make sure that the signal they detected was correct, they repeated their analysis, supported by artificial intelligence, for another 100,000 time windows, but after which there was no earthquake. They did not detect a slow but exponential growth signal as observed in the precursor phase of a large earthquake.

So much for the good news. As the authors themselves admit, they did not find this precursor phase in almost half of the earthquakes. That does not mean they do not have one: it could have occurred before the time frame they analyzed. For various reasons, such as the cost of computational calculations, they did not extend their analysis beyond the hours prior to each major earthquake. Another reason could be that the earthquake occurred too far from one of these geodetic stations. Nocquet is convinced that “developing systematic, precise, and accurate fault monitoring could potentially detect such precursor slip in the future for individual events.”

Víctor Puente is a researcher in geodesy applied to seismology at the National Geographic Institute. Puente, who values the importance of a study that has relied on information from the last 90 major earthquakes, recalls that the French scientists based their analysis on the Geodesy Laboratory of Nevada (United States) database. This database has records from 17,000 stations, rather than 3,000. If all of them were used, the capillarity of the analysis would be much greater. “But this lab offers the data with a two-hour latency,” Puente notes. If a detection system were to support them, the alert would arrive when the earthquake had already occurred. In any case, Puente stresses that if the results achieved by the French researchers are confirmed, the latency should be reduced until real-time data is available. “It would be difficult, but not impossible.”

“The more we know about faults, the more we will be able to know what the maximum magnitude of the earthquakes will be, then the intensity at the surface and then, in a second step, try to predict when they will occur.”

Jesús Galindo of the Geodynamics Department of the University of Granada

Another key to the functioning of a system such as the one suggested in this research is the need for in-depth knowledge of all the faults that could potentially be the origin of a major earthquake. Jesús Galindo, from the Geodynamics Department of the University of Granada, points out that this is a future research field to follow. “As has happened with meteorology, with more stations and better mathematical models, we are now able to predict what the weather is going to be like, the temperature, heat waves, or when it is going to rain. The same for faults; what is also needed is to have knowledge of how the ground moves and other physical parameters, such as the deep structure. The more we know about faults, the more we will be able to know what the maximum magnitude of the earthquakes will be, then the intensity at the surface and then, in a second step, try to predict when they will occur,” he explains.

“Absolutely key is the data you get near the hypocenter of the earthquake,” stresses the professor at the Polytechnic University of Madrid, an expert in geodesy applied to seismic hazards. About the research of French scientists, who are at the forefront in this field, he opines that it is a great contribution and is heading in the right direction. “But to be able to say that in two hours there will be an earthquake, we still have a long way to go.”

Sign up for our weekly newsletter to get more English-language news coverage from EL PAÍS USA Edition