Can Earthquakes Be Predicted? What Scientists Can and Cannot Do
Earthquake prediction is still out of reach, but probability maps, aftershock forecasts and the ShakeAlert system already save time and lives. Here is how each works and which popular signs are myths.
Earthquakes cannot be predicted. The US Geological Survey says neither it nor any other scientists have ever predicted a major earthquake, which would mean naming the date, the location and the magnitude in advance. What scientists can do is calculate the odds of a strong quake in a region over decades, forecast aftershocks after a big one, and send a few seconds of warning once shaking has already begun.
Those three tools sound modest, but they shape building codes, insurance, emergency plans and the alerts that now reach phones on the US West Coast.
Why can’t scientists predict earthquakes?
Scientists cannot predict earthquakes because they have found no reliable signal that a fault is about to break. Faults lie miles underground, under huge pressure, and stress builds on them for decades or centuries. Researchers can measure that buildup with GPS and satellites, but not the exact moment when friction on a fault gives way.
Many possible warning signs have been studied: small earthquakes, changes in groundwater, radon gas, electrical signals and odd animal behavior. The USGS notes that such things often happen without any earthquake following. A sign that appears before some earthquakes but also at many other times cannot be used to warn people.
The USGS FAQ on prediction puts it plainly: scientists can only calculate the probability that a significant earthquake will occur in a given area within a given number of years.
What is the difference between a prediction and a forecast?
A prediction says when, where and how big; a forecast gives the chance of an earthquake over a period of time, much like a climate outlook gives the chance of a wet winter.
| Tool | What it says | Time scale | Available now? |
|---|---|---|---|
| Prediction | Exact date, place and magnitude | Days to hours ahead | No |
| Long-term forecast | Probability of a quake of a given size in a region | Decades | Yes |
| Aftershock forecast | Chance and expected number of aftershocks | Next day, week, month, year | Yes, after significant US quakes |
| Early warning | Shaking is on its way to you | Seconds | Yes, in California, Oregon and Washington |
How do long-term earthquake forecasts work?
Long-term forecasts combine the history of past earthquakes, the slip rates of known faults measured in the field and by GPS, and statistics on how often quakes of each size happen. The result is a probability, not a date.
The best-known US example is for the San Francisco Bay Area. A USGS earthquake outlook for 2014 to 2043, based on the Uniform California Earthquake Rupture Forecast, gives a 72% chance of at least one magnitude 6.7 or larger earthquake in the region in that 30-year window. The same fact sheet gives a 51% chance of magnitude 7.0 or larger and a 98% chance of magnitude 6.0 or larger. The Hayward, Rodgers Creek, Calaveras and San Andreas faults have the highest odds.
These numbers do not say a quake is “overdue” or will strike in a particular year. They tell engineers and planners how strong to build and how much to prepare. The magnitudes involved are explained in our guide to how earthquakes are measured.
What are aftershock forecasts?
Aftershock forecasts are the closest thing to short-term earthquake forecasting that works. After a large quake, the chance of more quakes nearby is much higher than normal for days to months, and it declines in a fairly predictable way.
The USGS posts automatic aftershock forecasts after earthquakes of magnitude 4 and larger in the contiguous US and magnitude 5 and larger in other US states and territories. The first comes at least 20 minutes after the main shock and is updated many times over the following year. Each forecast gives the chance of at least one aftershock of magnitude 3, 4, 5 or larger, and how many to expect, over the next day, week, month and year.
The USGS also notes the opposite case: a small fraction of earthquakes are followed by a larger one, and the first quake is then called a foreshock. So a moderate quake is usually not a foreshock, but sometimes it is, and nobody can tell which until later.
How does ShakeAlert give warning before shaking arrives?
ShakeAlert works because an earthquake sends out two kinds of waves at different speeds. The USGS says the first, the P wave, travels at about 3.7 miles (6 km) per second and usually causes little damage. The slower S waves and surface waves, at about 2.5 miles (4 km) per second, do most of the shaking.
- Sensors near the fault detect the first P waves.
- Computers estimate the location, magnitude and expected shaking within seconds.
- The USGS issues an alert, and partners deliver it through phone emergency alerts, apps, Android phones, radio and automated systems.
- Trains slow down, elevators stop at a floor, and people drop, cover and hold on before strong shaking arrives.
ShakeAlert covers California, Oregon and Washington. Public alerting began in California in October 2019, and the whole West Coast had access by May 2021, according to the USGS early warning program. Phone apps and Android can alert people expected to feel weak shaking or more from quakes of magnitude 4.5 and up.
The USGS stresses that ShakeAlert is not prediction: it detects an earthquake that has already started. Warning time is usually only seconds, and it grows with distance from the epicenter. People very close to the start of a quake may feel shaking before any alert arrives.
Do earthquake weather and animal behavior predict quakes?
No. “Earthquake weather” is an old idea; the USGS traces it back to the ancient Greek belief that earthquakes came from air trapped underground. Earthquakes start miles down, where weather has no meaningful effect, and they happen in every kind of weather and at every time of day.
Stories of animals acting strangely before earthquakes go back to at least 373 BC in Greece, the USGS says. The trouble is that animals act strangely for many reasons, and odd behavior is noticed and remembered mostly after an earthquake. Some animals may sense the first, weaker P waves seconds before people feel the stronger shaking, which is a kind of natural early warning, not prediction. We look at the evidence in detail in can animals sense earthquakes.
Other claims fail the same test, including predictions tied to full moons, planetary alignments or “seismic forecasts” posted online. None has a track record better than chance.
Bottom line
No one can predict the time, place and size of an earthquake, and the USGS says no scientist ever has. What exists today is useful: decades-long probabilities, such as a 72% chance of a magnitude 6.7 or larger in the Bay Area from 2014 to 2043, aftershock forecasts after significant quakes, and ShakeAlert’s seconds of warning on the West Coast.
FAQ
Is California “overdue” for a big earthquake?
Scientists avoid the word. Faults do not break on a schedule, so forecasts give probabilities over decades instead. The Bay Area and Southern California both have high odds of a strong quake within 30 years.
Can small earthquakes relieve pressure and prevent a big one?
Not in a meaningful way. It takes roughly a thousand magnitude 5 quakes to release the energy of one magnitude 7, so small quakes cannot drain a fault’s stored stress.
How do I get earthquake early warnings on my phone?
In California, Oregon and Washington, government emergency alerts are delivered automatically to compatible phones. Android phones also have built-in alerts, and ShakeAlert-powered apps such as MyShake can deliver them too.
Does a cluster of small quakes mean a big one is coming?
Usually not, but it can. The USGS says a small fraction of earthquakes are followed by a larger one, so scientists watch swarms closely and issue aftershock-style forecasts.
