TREMR

When a large earthquake struck Osaka in 2018, many residents received a warning alert on their phones a few seconds before the shaking began. The time was short — barely enough to step away from a window or brace under a desk — but it was enough for hospitals to pause surgeries, bullet trains to begin braking, and factories to halt dangerous processes. A few seconds, it turns out, is not nothing.

Earthquake early warning systems are one of the more elegant applications of basic physics to public safety. They do not predict earthquakes — that remains impossible. What they do is detect the first, less-destructive waves of an earthquake and send an alert that travels faster than the more dangerous shaking that follows.

Two Kinds of Seismic Waves

Every earthquake produces multiple types of seismic waves. The two most important for early warning are P-waves (primary waves) and S-waves (secondary waves).

P-waves are compressional waves — they push and pull the rock in the direction they travel, like sound waves moving through air. They are fast: they travel through the crust at roughly 6 kilometres per second. They are also relatively weak. If you have ever heard a faint rumble before a large earthquake hit, you likely felt a P-wave. They cause little damage on their own.

S-waves are shear waves — they move the ground perpendicular to their direction of travel, making the earth roll and sway. They travel slower than P-waves, at about 3.5 kilometres per second. But they carry far more energy. The violent shaking that collapses buildings, throws people off their feet, and shatters glass is almost entirely caused by S-waves, sometimes followed by slower surface waves.

How the Warning Is Generated

An earthquake early warning system works by detecting the P-wave at sensors near the earthquake source, rapidly estimating the quake's magnitude and location, and transmitting a warning to distant areas before the S-wave arrives.

The chain of events is extremely fast. A seismic sensor detects the P-wave. Within one to two seconds, the signal is transmitted to a processing center. Algorithms analyze the P-wave's characteristics — its frequency, amplitude, and duration — to estimate how large the earthquake is likely to be. If the predicted magnitude exceeds a threshold, an alert is broadcast via radio, internet, cellular networks, and public address systems. The entire process, from P-wave detection to public alert, takes as little as 2–5 seconds.

For communities near the earthquake's epicenter, there may be no usable warning — the S-wave arrives too quickly. But for cities tens or hundreds of kilometres away, those seconds are valuable. The farther you are from the source, the more warning time you receive.

Japan's System: The World Standard

Japan's earthquake early warning system, operated by the Japan Meteorological Agency, is the most advanced in the world. It uses a nationwide network of over 4,000 seismometers. When triggered, warnings are broadcast via television and radio interrupts, the cell broadcast system (which causes every mobile phone in the affected area to emit a distinctive alarm sound regardless of whether it is on vibrate or silent), and the "J-Alert" public address system that activates loudspeakers in towns and cities.

The system has been operational in its current form since 2007 and has been tested by numerous major earthquakes, including the 2011 Tōhoku event. During Tōhoku, the warning arrived at Tokyo about 80 seconds before the strongest shaking, giving the capital's residents and infrastructure operators a critical window to react. High-speed bullet trains, which operate at speeds up to 320 km/h, have automatic braking systems that trigger on early warning alerts — no significant shinkansen derailments have occurred during earthquakes in the warning system's operational history.

ShakeAlert: The US System

The United States operates ShakeAlert, an early warning system covering California, Oregon, and Washington — the three states most exposed to significant earthquake risk. Developed by the USGS in partnership with state geological surveys, ShakeAlert uses a network of over 1,600 seismic sensors along the west coast.

Public alerts from ShakeAlert are distributed through the same Wireless Emergency Alert (WEA) system used for AMBER alerts and severe weather warnings — the familiar loud alarm that appears on mobile phones. Since its public launch, ShakeAlert has successfully issued warnings for several California earthquakes, including events in the M4–5 range that gave residents in Los Angeles a few seconds of notice.

The system's coverage and reliability continue to improve. Its main limitation is sensor density: the more sensors you have, and the closer they are to likely fault rupture zones, the faster and more accurate the initial estimates become. Areas of the Pacific Northwest, where a major Cascadia event would originate offshore, present particular challenges because ocean-bottom seismometers are needed to maximize warning time.

What You Can Do in the Warning Window

The effectiveness of any warning system depends on people knowing how to respond. A few seconds of warning is more useful than it sounds, provided you react instinctively rather than trying to think through options.

• Drop, cover, and hold on. Get under a sturdy table or desk, or against an interior wall away from windows. Cover your neck and head.
• If driving, slow down safely and pull over away from overpasses and bridges.
• If in a high-rise building, expect the shaking to feel amplified on upper floors. Do not use elevators afterward.
• If near the coast, the earthquake itself is your warning to move to high ground immediately — do not wait for an official tsunami alert.

The Limits of the Technology

Early warning systems have real limitations. The initial magnitude estimate, made within seconds from P-wave data alone, can be inaccurate — sometimes underestimating a quake's eventual size (leading to insufficient warning) or overestimating it (producing false alarms). Both types of errors erode public trust over time, which is why system designers work hard to calibrate the alert thresholds carefully.

Some areas near the fault — within 10 to 20 kilometres of the rupture zone — will receive no useful warning at all. The S-wave arrives too quickly. And the systems themselves require investment: dense sensor networks, robust telecommunications, and ongoing maintenance. Not all earthquake-prone countries have the resources to deploy them fully.

Despite these limitations, early warning systems represent one of the most cost-effective investments in earthquake preparedness available. The infrastructure is already built in the places that need it most. And every second of warning, used well, is a second that changes outcomes.