An earthquake alert can appear on a phone before the ground begins to move. That may look like a prediction, but it is something more practical—and more limited. Earthquake early warning systems detect a quake that has already started, calculate where damaging shaking may travel and send an alert faster than the seismic waves can reach nearby communities.
The warning may provide tens of seconds in favourable conditions. Close to the epicentre, it may provide only a few seconds or no advance notice at all. The system is a race between information travelling through communications networks and shaking travelling through the Earth.
Earthquake warning in one sentence
Sensors near a fault detect the first, faster seismic waves; computers estimate the quake's location and size; alerts then race ahead of the slower, stronger shaking to people and automated systems farther away.
It is called early warning, not prediction, because the rupture has already begun before an alert is generated.
P-waves and S-waves create the warning window
Earthquakes release several kinds of seismic energy. Two are especially important for early warning:
- P-waves, or primary waves, travel faster and usually cause weaker initial motion.
- S-waves, or secondary waves, travel more slowly but are generally responsible for stronger side-to-side shaking.
Seismometers positioned near the source can detect the first P-waves. Data from multiple stations is sent to a processing centre, which rapidly estimates the earthquake's origin, magnitude and likely shaking intensity. If the expected motion crosses an alert threshold, a warning is distributed to areas that may soon feel the stronger waves.
Electronic signals move through fibre, radio and mobile networks far faster than seismic waves move through rock. That difference creates the brief period in which an alert can arrive before the most dangerous shaking.
What happens between detection and an alert?
The process is automatic and usually follows four stages.
1. Sensors detect unusual ground motion
A single sensor is not normally enough. Systems compare readings from several stations so that traffic, construction or a local mechanical vibration is less likely to trigger a false event.
2. Software estimates the earthquake
Algorithms calculate an initial location, depth and magnitude from the first seconds of data. These estimates can change as the fault continues to rupture and more stations report measurements.
3. The system forecasts expected shaking
Magnitude alone does not tell every community what it will experience. Distance from the rupture, soil conditions, fault geometry and local geology can all affect shaking. The system therefore estimates intensity for different areas rather than sending the same message everywhere.
4. Alerts reach people and machines
Warnings may appear through mobile operating systems, dedicated apps, sirens, television or radio. The same data can trigger automated actions: trains may slow, elevators may stop at a floor, factory equipment may shut down and utility operators may protect sensitive systems.
How much warning time can people get?
There is no fixed answer. Warning time depends mainly on distance from the source, sensor density, communications speed and how quickly the system can identify the event.
People farther from the epicentre may receive several seconds or, in some cases, tens of seconds. Mexico City can sometimes receive longer notice from large offshore earthquakes because major faults are hundreds of kilometres away. Japan uses a dense national sensor network to issue public warnings and support systems such as high-speed rail.
People very close to the rupture can be inside the blind zone. Strong shaking may reach them before the system has collected enough data, completed its calculation and delivered an alert. This limitation cannot be solved by a faster phone alone; the first seismic information must still travel from the fault to a sensor.
Why early warning is not earthquake prediction
A prediction would identify a future earthquake's time, place and magnitude before it begins. Scientists cannot currently make that kind of reliable short-term prediction.
Early warning starts after the quake begins. The U.S. Geological Survey describes its ShakeAlert system as “detect, deliver, protect”: detect the event, deliver information and enable protective action before peak shaking arrives.
Forecasts and probabilities are different again. Scientists can estimate the long-term chance of earthquakes on particular faults, and aftershock forecasts can describe how risk may change after a major event. Neither tells a person the exact minute when the next earthquake will occur.
Why alerts sometimes change or arrive late
An earthquake can grow over several seconds as more of a fault ruptures. The first estimate may therefore be smaller than the final magnitude. A system must balance two competing goals: alert quickly with limited data, or wait for more confidence and lose valuable time.
Other limits include:
- gaps between seismic stations;
- interrupted power or communications;
- uncertainty about local ground conditions;
- a phone being offline, muted or outside the alert area;
- thresholds designed to avoid sending alerts for weak shaking.
A warning that arrives after mild motion has begun may still be useful if stronger waves are approaching. An alert does not guarantee severe shaking at every device that receives it, because estimates are produced before the full earthquake is known.
What should you do when an alert arrives?
The seconds are meant for immediate protection, not for checking social media or trying to leave a building.
The standard advice is Drop, Cover and Hold On:
- Drop to your hands and knees so shaking does not knock you down.
- Cover your head and neck under sturdy furniture if possible.
- Hold on until the shaking stops.
Move away from windows if that can be done immediately and safely. Do not use lifts. Drivers should slow and stop in a safe place away from bridges, power lines and overhead hazards. People near a coast should follow official tsunami instructions after strong or prolonged shaking rather than waiting for an informal online warning.
Which countries use earthquake early warning?
Operational systems exist in several earthquake-prone regions. The USGS lists countries including Japan, Mexico, China, Taiwan, Turkey, Romania and Italy among those with earthquake early warning capabilities. Coverage, delivery methods and public alert thresholds differ by location.
Japan's public service has operated since 2007. The Japan Meteorological Agency says its warnings estimate seismic intensity and the arrival time of principal motion, allowing people and infrastructure to take protective action.
The bottom line
Earthquake early warning works because information can travel faster than destructive shaking. Sensors detect the first waves, computers estimate what is happening and alerts are sent to places the stronger motion has not yet reached.
It does not predict earthquakes, and it cannot create a useful warning window everywhere. Even a few seconds, however, can help someone take cover, stop a train or pause hazardous machinery. The alert is valuable only when people understand it and act immediately.
For another guide to rapid hazard communication, read how tsunami warning systems work. Our article on why the heat index feels hotter explains a different kind of public warning. Browse more background guides in the Explainers section.
