Sep 28, 2017

Why we're decades away from predicting earthquakes

Illustration: Rebecca Zisser / Axios

Almost as long as scientists have been studying earthquakes, they've been trying to predict them. Despite over 150 years of seismology, most scientists agree that no earthquake has ever truly been predicted.

The bottom line: "We're no closer to earthquake prediction than we ever were, and are perhaps farther in that we now understand the difficulties better," says Robert Geller, a seismologist at the University of Tokyo.

What's needed: A reliable and accurate predictions of the time, location and strength of a temblor.

At its heart, it seems simple: earthquakes are just a release of stress. Shouldn't it be basic physics to understand when the fault has reached a breaking point? That's what geologist Harry Fielding Reid thought in 1910 when he proposed stretching strings across a fault, and measuring their movement until an earthquake came.

But that was 100 years ago. We've moved beyond string. We have seismometers and electromagnetic sensors and devices that measure stress deep below the ground.

The problem: The picture that emerged is much more complex than Reid imagined:

  • There are slow-slip earthquakes, which release fault stress without shaking.
  • Sometimes only a small part of the fault ruptures, while other parts remain locked in place.
  • Faults behave differently depending on the material composition of the crust around them.
  • The more we image faults, it seems, the more questions we have.

What to watch: In order to predict earthquakes, we need to find some kind of reliable pre-earthquake signal, or precursor. But we haven't found one yet.

It's not for lack of trying. The U.S. Geological Survey invested millions of dollars into an attempt to predict an earthquake in Parkfield, California. They covered the fault in sensors of every type, but when the quake came — ten years later than predicted — no pre-seismic signs were detected, though valuable data were gathered. Other such experiments have also been met with failure.

Some have suggested that some slippage might occur along a fault before the temblor lets loose — but USGS geophysicist Andrew Michael says it's currently impossible to measure such gradual change from the surface.

Others have proposed grinding rocks might cause electric charges before the quake, but Michael says it's unlikely those signals make it up through the ground and reach our surface sensors. "Consider how hard it is to get radio stations in a tunnel," says Michael.

People have been looking for precursors like this for decades, and some claim to have found them. "There are thousands of papers written on this," says Geller, "but it's highly unscientific."

What's next: Michael thinks there's hope, but not for earthquake predictions as traditionally defined. He emphasizes operational earthquake forecasting, which tells the probability that a quake of a certain strength might occur in a given region over a given period of time. The more we understand faults, the more accurate forecasts become.

These sorts of forecasts don't tell people when to evacuate, but they do help inform building codes, train emergency response teams and ultimately save lives.

Go deeper