Android smartphones around the world have been actively helping provide earthquake alerts using their acceleration sensors. Between 2021 and 2024, these phones sent thousands of successful warnings to people living in quake-prone regions in 98 countries.
Since 2020, Google has included its Android Earthquake Alert (AEA) system in its phones to detect seismic P- and S-waves that precede an earthquake and provide alerts. The company and the University of California Berkeley’s Seismology Laboratory have released data in a new paper in Science, detailing the performance and methods of the algorithm over the last three years.
The AEA system was first rolled out in the U.S. in 2020, followed by New Zealand, and Greece in 2021, and then the rest of the world.
It issued alerts for 11,231 events, including ahead of the Turkey-Syria (magnitude 7.8), the Philippines (6.7), and Nepal (5.7) events, all in 2023, and the 6.2-magnitude event in Turkey in April 2025. In the Philippines, Turkey-Syria, and Nepal quakes, users received 10 to 60 seconds of advance warning.
The system has thus far detected more than 18,000 quakes of varying magnitudes, including more than 2,000 strong quakes (moderate shaking or more on the MMI scale) and issued a total of 79 crore alerts. The AEA team also surveyed users between February 2023 and April 2024. Some 79% of 1.5 lakh users said they found the alerts very helpful.
AEA crowdsources signals from a phone’s accelerometer, a sensor that measures acceleration. On stationary phones, accelerometers can sense the faster-moving P-waves from an earthquake, which stretch and squeeze the geological material they pass through.
When an accelerometer senses these waves, it sends a signal to Google’s earthquake detection server, which then analyses data from other phones nearby to confirm the wave is real.
This data is collected and processed rapidly together with the waves’ location and magnitude. The goal is to issue an alert before the slower and more destructive S-waves arrive. While P-waves can travel through both solids and liquids, S-waves move only through solids. They also move up and down, displacing the material they move through.
AEA doesn’t monitor surface waves, the third type of wave to arrive in an earthquake. Surface waves are much slower and traverse only the outermost layer of the crust, and are typically responsible for the most aboveground damage.
Algorithm at work
To detect an earthquake, all smartphones bank on their accelerometers to sense vibrations on surfaces and in air. Earthquakes produce many such vibrations before the energy that causes damage actually reaches a location. The extent of damage caused by a quake increases with each type of wave that reaches a location on the surface.
The algorithm estimates the distance to the epicentre in one of two ways. If it’s within 200 km of the surface, it’s equal to the time difference between the arrival of the S- and P-waves multiplied by 8 km/s. For deeper epicentres, the algorithm includes additional methods to account for the waves’ motion through the mantle.
The algorithm issues two kinds of alerts: ‘BeAware’ alerts for impending light shaking and ‘TakeAction’ alerts for stronger shaking; the latter are accompanied by an audio warning that overrides ‘do not disturb’ settings. Alerts can also be manually enabled and disabled.
According to the paper, the system has increased the number of people with access to earthquake early warning by nearly a factor of 10: from around 25 crore people in 2019 to 250 crore people in 98 countries.
The authors of the paper, who are also affiliated with Harvard University and Google Germany, also highlighted the AEA’s shortcomings — including its inability to accurately estimate the magnitude and tendency to relay warnings after shaking has begun.
They also documented these issues in detail and specified tweaks to fix many issues, including lowering the median absolute error in the first estimates of a quake’s magnitude from 0.5 to 0.25. They also traced three false positives in the 11,321 events to two hurricanes and one mass phone vibration from an unrelated alert, and included fixes for alerts to be issued faster.
“The proliferation of low-cost sensors in consumer products provides opportunities for new observations of our physical environment that are both dense and global in scale. AEA makes it possible to make global observations of earthquakes,” the authors wrote.
Sandhya Ramesh is a freelance science journalist.
Published – July 27, 2025 05:40 am IST
