When it comes to an earthquake, just a few seconds’ warning could make the difference between life and death. But many earthquake-prone countries lack the seismic networks that would give their citizens the lead time to find cover or shut down critical utilities. Now, a group of enterprising engineers is looking at a substitute network: smartphones. Using smartphones’ built-in accelerometers, researchers have invented an app—released today—that they say can detect strong earthquakes seconds before the damaging seismic waves arrive. MyShake, as the app is called, could become the basis for an earthquake warning system for the world’s most vulnerable regions.
Trying to harness the world’s 1 billion smartphones for earthquake warning isn’t new, says John Vidale, a seismologist at the University of Washington, Seattle, who was not involved in the study. “We all know the approach could work if the phones are properly equipped,” he says. “But [this study has] done a number of things to solidify arguments that it’s practical. [They’re] showing that it could work now, if people are willing to run the software.”
Previous efforts have struggled to tease out an earthquake’s shaking from everyday jostles and bumps. Most use sensors already built into cellphones, particularly accelerometers, three-axis motion sensors that keep track of orientation, or GPS, which gives the phone’s absolute position at any time. Last year, a separate team of scientists showed that smartphone GPS systems—although not as precise as sensitive scientific instruments—are still accurate enough to positively detect an earthquake of at least magnitude 7. But as anyone who has used a phone’s map function knows, GPS is a constant drain on battery power, making it somewhat impractical for a warning system that requires constant vigilance.
The smartphone’s ability to identify an earthquake in progress is just one part of the challenge. Another part is on the back end: Devising an algorithm to take data from tens of thousands of smartphones, determine a magnitude and epicenter for the quake, and then issue a timely warning.
In a new study, online today at Science Advances, researchers show that the MyShake app can detect earthquakes of at least magnitude 5 occurring within 10 kilometers of the phone. The team spent more than a year collecting data from phones placed on a “shake table,” a device that simulates the ground motions produced by an earthquake, and comparing them with data from everyday movements. “We’ve shown that we can identify earthquakes with a 93% success rate,” says Richard Allen, paper author and seismologist at the University of California, Berkeley. “[That] is remarkable because the kinds of motion a phone experiences from everyday movements are much larger.” The team has also been tweaking its network detection algorithm, which powers the centralized system that analyzes the smartphone data and relays warnings.
The initial release of the MyShake app is intended as a large-scale test of the algorithm, aimed at collecting tens of thousands of data points from cellphones. So this version won’t issue alerts just yet, Allen says. But if the app passes the test, the alert system might be functional as early as the middle of this year. So far, MyShake is available only for Android phones. “Globally, there are many more Android phones, and that’s why we focused on that first.”
“This is a really nice paper and a good approach,” Vidale says. “It’s quite a practical result. What he’s proposing, properly tuned, could be as good as a standard early warning system with established networks.”
Indeed, there are a lot of ways a smartphone early warning system could outperform a standard network, he says: A standard system doesn’t have an infinite number of seismometers, but more like one every 10 to 20 kilometers. A cellphone-based warning system could be much denser, and therefore more accurate. But timing is also an issue. How quickly the system can issue warnings is based on a number of factors, including the magnitude of a quake, the number of cellphones detecting it, and the distance of phones from the epicenter. For example, the researchers estimate that for the 7.8 quake that struck 80 km north of Kathmandu last April, the warning system would give people in the city 20 seconds warning.
But that doesn’t mean that such a system would replace a standard network in a well-instrumented country like the United States, Vidale adds. “Hard-wired systems are more immune to tampering and more predictable in behavior than a cellphone network could be. A lot of things are hard to control, including whether cellphone towers might go down in a quake. “It’s not a replacement for existing systems—but it has tremendous potential for places that can’t afford a standard seismic network.”