What causes earthquakes? Faults, stress and sudden slips

Most earthquakes are caused by rock slipping suddenly along a fault, a crack in Earth's crust. The plates on either side keep moving, but friction holds the fault locked, so the rock bends and stores energy, sometimes for centuries. When the stress grows greater than the fault's grip, the rock snaps past itself in seconds and the stored energy spreads out as seismic waves, which are the shaking. Volcanoes, collapsing caves, explosions and human activity cause smaller ones.

In short

• Plates move steadily, a few centimeters a year, but faults move in jumps: years of stored bending released in seconds.
• Each whole step on the magnitude scale means about 32 times more energy, so a magnitude 8 releases about 1,000 times the energy of a 6.
• No one can say when the next big earthquake will strike, but scientists can estimate the odds over decades and warn people seconds before the shaking arrives.

How does a fault cause an earthquake?

The explanation is called elastic rebound. Harry Fielding Reid worked it out from the 1906 San Francisco earthquake, after surveyors found that land on either side of the San Andreas Fault had been bending for decades before the rupture, then jumped. Step by step:

1. The plates move. Along the San Andreas, the Pacific Plate slides northwest past the North American Plate at about 5 cm a year on average, according to the U.S. Geological Survey.
2. The fault sticks. The two sides are pressed together, and friction stops the fault surface from sliding. The rock farther away keeps moving, so the rock near the fault bends, like a ruler flexed between your hands.
3. Strain builds. The bent rock stores elastic energy. Over 100 years at 5 cm a year, the offset waiting to be released is about 5 m.
4. The fault breaks. When the stress overcomes friction at one weak spot, a rupture starts there and races along the fault at a few kilometers a second, unzipping it.
5. The rock rebounds. Both sides spring back to an unbent shape in their new positions. The ground surface across the fault is offset, by meters in a large earthquake.
6. The waves go out. The energy released travels as seismic waves through and around the planet.
7. Then it locks again. Smaller aftershocks follow as the rock nearby adjusts, and the cycle starts over.

Faults move in different ways depending on how they are stressed. Where the crust is pulled apart, one block drops down a sloping fault, a normal fault. Where it is squeezed, one block is pushed up and over the other, a thrust or reverse fault, which is how the largest earthquakes on Earth happen. Where blocks slide sideways past each other, as on the San Andreas, it is a strike-slip fault. Almost all of this happens at plate edges, which is why earthquakes trace the boundaries described in the note on plate tectonics; fewer than 10% of earthquakes strike inside plates, the USGS says.

What is the difference between the focus and the epicenter?

The focus, also called the hypocenter, is the point underground where the rupture starts; the epicenter is the spot on the surface directly above it. News reports name the epicenter because it is a place on a map. Most earthquakes start in the top 70 km, which seismologists class as shallow, and shallow ones do the most damage. In subduction zones, where one plate sinks under another, earthquakes can start more than 700 km down, inside the sinking slab, as in the Pacific's Ring of Fire.

The rupture itself isn't a point. In a great earthquake, the fault can break over hundreds of kilometers. The focus is only where it began.

What kinds of seismic wave are there?

An earthquake sends out three main kinds of wave, which arrive in order:

Wave	How it moves the ground	Speed in the crust, roughly	What you feel
P wave, primary	Pushes and pulls in the direction it travels, like a sound wave	About 6 km a second	A sharp jolt or thud, first
S wave, secondary	Shakes side to side or up and down, across its direction of travel	About 3.5 km a second	Stronger shaking, a few seconds later
Surface waves	Roll and sway along the surface	A little slower than S waves	The long, rolling motion that does most damage

The gap between the P and S waves tells you how far away the earthquake is. P waves cover a kilometer in about 1/6 of a second and S waves in about 1/3.5 of a second, so for every kilometer of distance the S wave falls behind by 0.29 − 0.17 = 0.12 seconds. Turn that around: each second of gap means about 8 km. A 10-second gap between the first jolt and the strong shaking puts the earthquake about 80 km away. Three stations' distances, drawn as circles on a map, cross at the epicenter. The same waves, passing through the whole planet, are how scientists mapped what the Earth is made of.

How is an earthquake's size measured?

Two different things are measured. Magnitude is the size of the earthquake itself: one number for the whole event. Intensity is how strongly the shaking was felt at a given place, rated on the Modified Mercalli scale from I to XII, and it varies with distance and with the ground underneath. A moderate earthquake under a city on soft soil can do more harm than a huge one far offshore.

The Richter scale, devised by Charles Richter in 1935, has been replaced for large earthquakes by the moment magnitude scale, which is based on the area of fault that slipped, how far it slipped and how stiff the rock is. Both are logarithmic, as the note on logarithms explains: each whole number is 10 times more ground motion on a seismogram and about 32 times more energy released. The energy ratio comes from the formula: energy grows as 10 to the power 1.5 times the magnitude, and 10^1.5 ≈ 31.6.

Here is what that does to the numbers:

1. From magnitude 6 to 7: 31.6 times the energy.
2. From 6 to 8: 31.6 × 31.6 ≈ 1,000 times the energy.
3. From 6 to 9: 31.6 × 31.6 × 31.6 ≈ 31,600 times.
4. Even 0.1 of a step matters: from 7.0 to 7.1 is about 1.4 times the energy.
5. The largest earthquake ever recorded, in Valdivia, Chile, in 1960, was magnitude 9.5: about 178,000 times the energy of a magnitude 6.

So "a 7 isn't much bigger than a 6" is the most common misreading of the scale. A 9 is not 50% larger than a 6; it is tens of thousands of times larger in energy.

Why can't earthquakes be predicted?

A real prediction would give the place, the time and the size, closely enough to act on. No one has ever done that reliably for a large earthquake. The problem is that a fault's breaking point depends on stresses and frictions kilometers underground that can't be measured directly, and the small events that might come before a big one look the same as the thousands that come before nothing.

The best test was at Parkfield, California. Magnitude 6 earthquakes had struck there six times between 1857 and 1966, about every 22 years. The USGS forecast the next for about 1988, almost certainly by 1993, and filled the area with instruments to catch the run-up. It came on 28 September 2004, 16 years after the expected date and 11 years after the window closed, and the instruments recorded no clear warning sign.

What does work is forecasting, in the sense of a weather forecast: the probability of a large earthquake in a region over the coming decades, from the history of past earthquakes and how fast strain is building. That is enough to set building codes and insurance, even though it can't say which year.

The other thing that works is early warning, which starts after an earthquake has begun. Sensors near the fault detect the fast, weak P waves, computers estimate the size within seconds, and an alert travels at the speed of electronics, faster than the damaging S and surface waves. Using the figures above, the S waves reach a city 80 km away about 10 seconds after the P waves, so once detection has taken its few seconds, there is time to slow trains, stop surgery and open fire-station doors. Places close to the epicenter get little or no warning. Japan runs a national system, and in the western United States the USGS-managed ShakeAlert covers California, Oregon and Washington. What to do in an earthquake is set out by national agencies such as these.

Do people cause earthquakes?

Some, yes. Pumping large volumes of wastewater from oil and gas production deep underground raises the fluid pressure in rock, which can unclamp a fault that was already close to failing. A magnitude 5.7 earthquake in Oklahoma in 2011 is widely thought to have been caused this way. Filling large reservoirs, mining and underground nuclear tests can also trigger earthquakes. These are mostly small, and they need a stressed fault to be there already: human activity can bring an earthquake forward, not create the strain behind it.

Check yourself

• Without looking back: why does a fault stay still for years if the plates on either side keep moving?
• How much more energy does a magnitude 8 release than a magnitude 6, and why?
• What is the difference between predicting an earthquake, forecasting one and warning of one?

Sources

• Earthquakes, a general interest publication — Shedlock and Pakiser, U.S. Geological Survey, 1995
• Understanding plate motions, in This Dynamic Earth — U.S. Geological Survey
• ShakeAlert earthquake early warning — U.S. Geological Survey and partners
• Elastic-rebound theory — Wikipedia
• Parkfield earthquake — Wikipedia