Why can’t we predict earthquakes?
Because faults give no reliable warning. A real prediction needs a date, a place and a size, and no scientist has ever predicted a major quake. Scientists forecast the odds instead.
Quick Answer
According to the U.S. Geological Survey, neither the USGS nor any other scientists have ever predicted a major earthquake. A true prediction has to name three things: when, where and how big. Faults don’t give a reliable warning sign before they slip. The most famous test was at Parkfield, California, where magnitude-6 quakes seemed to arrive about every 22 years. In 1985, scientists forecast the next one would hit by 1993. It didn’t come until September 2004, and it arrived with no obvious warning signs. What scientists can do is calculate probabilities. The USGS estimates a 72% chance of at least one magnitude 6.7 or larger earthquake in the San Francisco Bay Area before 2043. And systems like ShakeAlert on the U.S. West Coast can detect a quake that has already started and send alerts seconds before strong shaking arrives.
By the Numbers
Key Facts
- Short answer
- No reliable warning sign has ever been found
- A real prediction needs
- A date and time, a location and a magnitude
- What we can do
- Forecast probabilities, and send early warnings once a quake starts
- Bay Area odds
- 72% chance of a magnitude 6.7+ quake before 2043
Prediction vs. forecast vs. early warning
| What it tells you | Possible today? | |
|---|---|---|
| Prediction | Exact date, place and size of a future quake | No |
| Forecast | The probability of a quake in an area over years or decades | Yes |
| Early warning | Shaking is on its way from a quake that has already started | Yes, in some areas |
What a prediction would take
The USGS says a real earthquake prediction must name the date and time, the location and the magnitude. People who claim to predict quakes usually leave one of those out, or make predictions so broad that some earthquake will always fit.
The Parkfield experiment
A stretch of the San Andreas Fault near Parkfield, California, had magnitude-6 earthquakes roughly every 22 years. In 1985, scientists forecast the next one would come within about five years of 1988, and they filled the area with instruments to catch the warning signs. The quake didn’t arrive until September 2004, and it showed no obvious warning signs.
Forecasts and warnings instead
Scientists study past quakes and how fast faults move to estimate the odds over decades, like the 72% chance for the Bay Area before 2043. Early warning systems such as ShakeAlert detect a quake as it begins and send alerts that can arrive seconds before strong shaking, enough time to drop, cover and hold on or slow a train.
Myth vs. Fact
MythThere’s such a thing as “earthquake weather.”
FactThe USGS says earthquakes happen in all kinds of weather.
MythAnimals can predict earthquakes.
FactSome animals may sense the first, faster waves seconds before people feel shaking, but the USGS says consistent, reliable behavior days or weeks before a quake hasn’t been shown.
The Bottom Line
We can’t predict earthquakes because faults give no reliable warning. Scientists can forecast the odds over decades and, in some places, send alerts seconds before the shaking arrives.
Sources & Further Reading
- USGS — Can you predict earthquakes?
- USGS — What is the difference between earthquake early warning, earthquake forecasts, earthquake probabilities, and earthquake prediction?
- USGS Fact Sheet 2016-3020 — Earthquake Outlook for the San Francisco Bay Region 2014–2043
- Nature — Bakun et al., Implications for prediction and hazard assessment from the 2004 Parkfield earthquake (2005)
- USGS — Can animals predict earthquakes?
- USGS — ShakeAlert