An earthquake is a sudden shaking of the ground caused by huge pieces of Earth’s outer shell moving against each other. These pieces are called tectonic plates. They move very slowly, but sometimes they get stuck and then suddenly slip. That sudden slip is what makes the ground shake. Scientists count about 500,000 earthquakes around the world every year, but most are too small to feel.
Why earthquakes are tricky to understand
You cannot see the part of an earthquake that matters most. The rocks that break and slip during an earthquake are usually buried miles below your feet. The first you notice on the surface is when the ground starts to shake.
Earthquakes do not happen everywhere. They cluster along the edges of tectonic plates. About 90 percent of the world’s earthquakes happen along a giant horseshoe-shaped zone around the Pacific Ocean called the Ring of Fire. It runs through Japan, Alaska, the West Coast of the United States, and the Andes Mountains in South America.
The size of an earthquake is measured with a number called magnitude. The scale is not like inches or pounds. Each step up is about 32 times more energy. A magnitude 7 quake is about 32 times more powerful than a magnitude 6, and roughly 1,000 times more powerful than a magnitude 5. The largest earthquake ever measured, in Chile in 1960, was magnitude 9.5.
Key facts about earthquakes
The ground breaks along cracks called faults. The famous San Andreas Fault in California is about 750 miles (1,210 km) long.
Most earthquakes are tiny. Of the roughly 500,000 detectable earthquakes per year, about 100,000 are strong enough for people to feel. Only around 100 cause damage.
The biggest earthquake ever measured was in Chile in 1960. It had a magnitude of 9.5 and made a tsunami that crossed the entire Pacific Ocean.
The biggest earthquake in US history was in Alaska in 1964. It was magnitude 9.2 and shook the ground for about four and a half minutes.
An earthquake in Japan in 2011 nudged Earth. The magnitude 9.1 Tōhoku quake shifted Earth’s axis by about 6.5 inches (17 cm) and made the day shorter by about 1.8 millionths of a second.
The first earthquake-detecting machine was Chinese. Around 132 AD, Zhang Heng built a device with eight bronze dragons. When the ground shook, a ball dropped from one dragon’s mouth to show which direction the shaking came from.
Most tsunamis come from undersea earthquakes. When the seafloor jumps up or drops down during a quake, it pushes the water above into a wave that can cross an entire ocean.
California has a warning system.ShakeAlert started sending public alerts in October 2019. It cannot tell the future, but it can warn faraway places that shaking is on the way once a quake has started.
Doorways are not the safest spot. Today, scientists say to drop to the floor, take cover under a sturdy table, and hold on until the shaking stops.
Common myths about earthquakes
Myth: Earthquakes can drop a state into the ocean. Along the San Andreas, the two sides slide past each other rather than pulling apart. No state can fall into the sea.
Myth: Animals can predict earthquakes days ahead. Some pets and farm animals act strange just before a quake, but only because they feel the small first shake a few seconds before people do. There is no proof that animals sense earthquakes hours or days in advance.
Myth: Earthquakes happen mostly in hot weather. Earthquakes happen in every kind of weather, all over the year. There is no such thing as “earthquake weather.”
Myth: Small earthquakes prevent big ones. It would take about 32 magnitude 6 earthquakes to release as much energy as one magnitude 7. Tiny daily quakes do not stop the big ones from coming.
Frequently asked questions about earthquakes
What causes an earthquake?
Earthquakes happen when two blocks of rock that have been pushing on each other for a long time suddenly slip. The slip releases energy that travels outward as shaking. The blocks are part of giant tectonic plates.
What is a fault?
A fault is a crack between two blocks of rock that can move against each other. Most earthquakes happen along faults.
Where do most earthquakes happen?
Most earthquakes happen along the edges of tectonic plates. The Ring of Fire around the Pacific Ocean has about 90 percent of them. California, Alaska, Japan, Chile, and Indonesia are all in this zone.
What should I do if an earthquake happens?
If you are inside, drop to your hands and knees, take cover under a sturdy desk or table, and hold on. Stay there until the shaking stops. If you are outside, move away from buildings, trees, and power lines, then drop to the ground.
Can scientists predict earthquakes?
Not yet. Scientists can map places where earthquakes are likely over many years, and they can warn faraway places that shaking is on the way once a quake has already started. But no one can yet tell exactly when the next earthquake will strike.
What is a tsunami?
A tsunami is a giant ocean wave caused by a big undersea earthquake. When the seafloor jumps up or down, the water above is pushed into a wave that can travel across an entire ocean.
Trivia question references throughout this topic’s Rookie, Curious, Sharp, and Expert quiz sets each cite a primary source for the specific fact tested.
An earthquake is the shaking of the ground when stored stress in Earth’s crust suddenly releases along a fault. A fault is a crack between two blocks of rock that can slip past each other. The energy released travels outward as seismic waves. Most earthquakes happen along the boundaries of tectonic plates, where the giant slabs of Earth’s outer shell push, pull, or grind against each other. About 500,000 earthquakes are detected globally every year, but only around 100 cause damage.
Why earthquakes are tricky to understand
The action that creates an earthquake happens deep underground. The point inside Earth where the rupture begins is called the hypocenter, or the focus. The point on the surface directly above is called the epicenter. News reports usually give the epicenter, but the actual rupture is somewhere below it, sometimes a few miles down, sometimes hundreds of miles down inside a sinking tectonic slab.
Magnitude works in a way that surprises most people. The scale is logarithmic, which means each whole-number step is much bigger than the one before. A magnitude 7 earthquake releases about 32 times more energy than a magnitude 6 and roughly 1,000 times more than a magnitude 5. That is why a magnitude 5 quake might break dishes while a magnitude 8 can flatten a city. The largest earthquake ever measured was magnitude 9.5 in Chile in 1960.
Damage in a major earthquake is rarely just from shaking. Liquefaction can turn loose, water-soaked sand into a slurry that sinks buildings. Broken gas mains can start fires that spread for days. Undersea earthquakes can launch tsunamis that travel across whole oceans. The 1906 San Francisco earthquake had a magnitude of 7.9, but most of the damage came from the fires that burned out of control for three days afterward, destroying about 25,000 buildings.
Earthquakes cannot be predicted in the way the weather is. Scientists know which regions face the highest risk, and they can issue forecasts that estimate probabilities over decades. ShakeAlert in California, Oregon, and Washington can give people seconds of warning before strong shaking arrives, but only after the earthquake has already started. The exact time, place, and size of the next big quake remain unknown.
Key facts about earthquakes
Plate boundaries control where earthquakes happen. The Pacific Plate slides past the North American Plate along the San Andreas Fault. The Nazca Plate dives beneath South America. The Pacific Plate dives beneath Japan. Almost every large earthquake is at one of these plate boundaries.
The Ring of Fire dominates. About 90 percent of the world’s earthquakes happen along the horseshoe-shaped Ring of Fire that circles the Pacific Ocean. The same zone has roughly three quarters of Earth’s active volcanoes.
The 1960 Chile earthquake was magnitude 9.5. It is still the strongest earthquake ever measured. It triggered tsunami waves that killed people in Hawaii and Japan thousands of miles away.
The 1964 Alaska earthquake was magnitude 9.2. It is the largest US earthquake on record and the second largest worldwide. The ground shook for more than four minutes.
Tsunamis come from undersea earthquakes. When the seafloor jumps up or down during a quake, the water above is shoved into a wave. The wave can be only a few feet high in the open ocean but pile up to dozens of feet near a coast.
The 2004 Indian Ocean earthquake killed about 228,000 people. Most deaths came from the tsunami, not from shaking. The fault rupture lasted close to ten minutes and stretched for about 750 miles (1,200 km).
The 2011 Japan earthquake nudged Earth’s axis. NASA scientists calculated that the magnitude 9.1 Tōhoku quake shifted Earth’s figure axis by about 6.5 inches (17 cm) and shortened the day by about 1.8 microseconds.
California’s ShakeAlert system started statewide in October 2019. The system detects fast P-waves near the epicenter and rushes a warning to phones and machinery before the slower, stronger shaking arrives.
Drop, Cover, and Hold On. When the ground starts to shake, drop to your hands and knees, take cover under a sturdy desk or table, and hold on. Doorways are not safer than other parts of a modern building.
Common myths about earthquakes
Myth: Earthquakes happen during certain kinds of weather. Earthquakes happen in every kind of weather, season, and climate. There is no real “earthquake weather.”
Myth: California will fall into the ocean. The San Andreas Fault is a transform boundary, where the Pacific Plate slides north past the North American Plate at about 1 to 1.4 inches (2 to 3.5 cm) per year. The plates are not pulling apart.
Myth: Animals know hours in advance. USGS finds no reproducible evidence that animals predict earthquakes hours or days ahead. Some animals do react seconds before humans because they detect the faster but weaker P-wave that arrives before the larger S-wave.
Myth: Doorways are the safest spot. Doorways in modern houses are no stronger than other walls and leave you exposed to falling objects. The current advice from emergency planners is Drop, Cover, and Hold On under a sturdy table.
Myth: Small earthquakes prevent a big one. Each whole-number jump in magnitude is about 32 times more energy. It would take about 32 magnitude 6 earthquakes, or about 1,000 magnitude 5 earthquakes, to release the energy of one magnitude 7.
Myth: Aftershocks are on a different, distant fault. Aftershocks happen near the original fault and come from local rock adjusting to its new stress state. They can continue for weeks, months, or years.
Frequently asked questions about earthquakes
What is the difference between magnitude and intensity?
Magnitude is one number per earthquake describing the energy released by the rupture. Intensity describes how strong the shaking felt at a specific location and uses Roman numerals from I to XII on the Modified Mercalli scale. One earthquake has only one magnitude, but its intensity changes from place to place depending on distance, soil, and building type.
What are P-waves and S-waves?
P-waves and S-waves are the two main types of seismic body waves. P-waves (primary) are compression waves that travel fastest, around 4 miles per second (6 km/s) in the crust, and arrive first. S-waves (secondary) are shear waves that travel at about 60 percent of P-wave speed and arrive later. Surface waves come last and usually do the most damage.
Why do some big earthquakes cause tsunamis but others do not?
Tsunamis form when an undersea earthquake lifts or drops the seafloor vertically. Subduction zones, where one tectonic plate dives under another, do this efficiently. Strike-slip earthquakes, where the rock slides side to side, move the seafloor only a little vertically, so they generate weaker tsunamis even at high magnitude.
How do scientists know when an old earthquake happened?
For events before modern instruments, geologists piece together evidence from coastal sediments, tree rings, and historical records. The 1700 Cascadia earthquake off the Pacific Northwest is dated to the night of January 26, 1700 thanks to Japanese village records of an “orphan tsunami” that arrived without local shaking, plus tribal oral histories and buried marsh sediments.
Can people make earthquakes happen?
Yes, in some cases. Deep wastewater injection from oil and gas operations can raise pressure on faults and trigger earthquakes. Oklahoma’s earthquake rate jumped sharply between 2009 and 2015 because of wastewater disposal. The largest such induced event so far was magnitude 5.8 near Pawnee, Oklahoma in September 2016.
What was the deadliest earthquake in modern history?
The 2004 Indian Ocean earthquake and tsunami killed about 228,000 people across 14 countries. Most deaths were caused by tsunami waves up to about 100 feet (30 m) tall. The earthquake itself had a moment magnitude of 9.1 and was the most powerful ever recorded in Asia.
Trivia question references throughout this topic’s Rookie, Curious, Sharp, and Expert quiz sets each cite a primary source for the specific fact tested.
An earthquake is the sudden release of stored elastic strain in Earth’s crust or upper mantle, radiating outward as seismic waves that shake the ground. Most earthquakes occur where tectonic plates grind past, pull apart from, or dive beneath one another. About 90 percent of the world’s earthquakes happen along the Ring of Fire, the horseshoe of plate boundaries circling the Pacific Ocean. USGS estimates that around 500,000 earthquakes are detectable each year, of which roughly 100,000 can be felt and about 100 cause damage.
What is often misunderstood about earthquakes
The Richter scale you may have heard about is no longer the scale USGS uses for moderate and large events. Charles Richter and Beno Gutenberg published the original scale at Caltech in 1935 for use with local southern-California seismograms. It saturates above about magnitude 7, meaning the formula cannot distinguish a magnitude 8 from a magnitude 9. Modern reports use moment magnitude (Mw), introduced by Hanks and Kanamori in 1979 and based on seismic moment, the product of fault slip, rupture area, and rock rigidity. Moment magnitude is calibrated to match Richter values where the older scale still works, which is why headline numbers look familiar.
Magnitude and intensity are different. Magnitude is one number per earthquake describing the energy released at the source. Intensity describes felt shaking at a specific place and varies across a map. The Modified Mercalli Intensity scale, developed by Wood and Neumann in 1931, runs from I (not felt) to XII (total destruction) and is reported in Roman numerals. A magnitude 7 earthquake produces high intensities near the epicenter and very low intensities a few hundred miles away.
The magnitude scale is logarithmic. Each whole-number step is about a tenfold jump in ground motion and roughly 32 times more energy released. So magnitude 7 is about 32 times the energy of magnitude 6 and roughly 1,000 times the energy of magnitude 5. The Gutenberg-Richter law, published in 1944, captures the inverse relationship in earthquake counts: for every magnitude 7 event, a typical region has about 10 magnitude 6 events and 100 magnitude 5 events.
Most damage in big earthquakes is not from shaking alone. Liquefaction of saturated sandy soil, fires from broken gas mains, and tsunamis from undersea ruptures often dominate the casualty list. The 1906 San Francisco earthquake killed more than 3,000 people, and the fires that followed for three days destroyed roughly 25,000 buildings, accounting for the great majority of property losses. The 2004 Sumatra-Andaman earthquake killed about 228,000 people, almost all by tsunami rather than by ground shaking.
Predicting earthquakes in the strict sense, time, place, and magnitude with high confidence, is not currently possible. What seismology offers instead is forecasting: long-term hazard maps that estimate probabilities over years and decades, and short-term aftershock-sequence forecasts in the days after a main shock. ShakeAlert, which began statewide public alerting in California in October 2019, is a warning system rather than a prediction system. It detects an earthquake that has already started, then races a digital warning to more distant places before the slower S-waves and surface waves arrive.
Key facts about earthquakes
The largest earthquake on record. The 1960 Valdivia earthquake in southern Chile, magnitude 9.5, remains the strongest earthquake ever measured with modern instruments. It ruptured a roughly 500-mile (800 km) segment of the Nazca-South American subduction zone and triggered a tsunami that killed 61 people in Hilo, Hawaii.
The largest US earthquake. The 1964 Great Alaska earthquake, magnitude 9.2 in the Prince William Sound region, is the largest US earthquake on record and the second largest worldwide. The shaking lasted about four and a half minutes.
The 2011 Tōhoku quake nudged the planet. The March 11, 2011 magnitude 9.1 megathrust off northeastern Japan shifted Earth’s figure axis by about 6.5 inches (17 cm) and shortened the day by about 1.8 microseconds, according to NASA JPL calculations by Richard Gross.
Body waves arrive first. P-waves are compressional and travel at about 4 miles per second (6 km/s) in the crust. S-waves are shear waves and move at roughly 60 percent of P-wave speed. Surface waves, Love and Rayleigh, arrive last and generally cause the most building damage because their longer periods can match the resonance of mid-rise buildings.
The Ring of Fire dominates. About 90 percent of earthquakes occur in the Pacific-rim plate boundaries that include the Cascadia, Aleutian, Japan, Philippines, Indonesia, New Zealand, and Andean subduction zones, and the San Andreas transform system in California.
Hypocenter and epicenter. The hypocenter (or focus) is where rupture begins inside Earth. The epicenter is the point on the surface directly above. Most earthquakes are shallow (less than 70 km), but deep-focus events extend down to about 435 miles (700 km) inside subducting slabs.
The 1700 Cascadia earthquake. A roughly magnitude 9 megathrust ruptured off Washington, Oregon, and northern California on the night of January 26, 1700. The date was reconstructed from a Japanese “orphan tsunami” that struck Honshu, plus Pacific Northwest tribal oral histories and coastal sediment cores.
Foreshocks are only known after the fact. USGS notes that no real-time signal distinguishes a small earthquake from one that will be followed by a larger event. A foreshock is identified only after a larger main shock arrives.
Liquefaction destroyed San Francisco’s Marina District in 1989. During the magnitude 6.9 Loma Prieta earthquake, water-saturated fill briefly behaved like a liquid, sinking buildings and rupturing utilities. The shock arrived 31 minutes before Game 3 of the World Series at Candlestick Park.
Induced seismicity in Oklahoma. Wastewater injection from oil and gas operations drove a sharp rise in Oklahoma earthquake rates between 2009 and 2015, with annual M 3+ counts surpassing California from 2014 through 2017. The largest documented injection-induced event was a magnitude 5.8 near Pawnee on September 3, 2016.
Common myths about earthquakes
Myth: Doorways are the safest place during an earthquake. Modern building codes make doorways no stronger than other parts of a house, and standing in one leaves you exposed to flying objects. Safety experts recommend Drop, Cover, and Hold On: drop to your hands and knees, take cover under a sturdy desk or table, and hold on until the shaking stops.
Myth: Earthquakes happen in “earthquake weather.” Aristotle proposed that hot, calm air trapped underground caused earthquakes. Statistically, earthquakes are spread evenly across hot, cold, wet, and dry conditions. There is no surface weather pattern that reliably precedes a quake.
Myth: California will fall into the ocean. The San Andreas Fault is a transform boundary where the Pacific Plate slides horizontally past the North American Plate at about 1 to 1.4 inches (2 to 3.5 cm) per year. The two plates are not pulling apart, so no large block of California can slide off into the Pacific.
Myth: Animals can predict earthquakes hours or days ahead. USGS finds no reproducible mechanism that lets animals foresee earthquakes hours or days in advance. The clearest documented effect is short-range: many animals detect the faster P-wave and react seconds before humans feel the slower S-wave.
Myth: Small earthquakes “release pressure” and prevent big ones. Each whole-number step on the magnitude scale is about 32 times more energy. It would take roughly 32 magnitude 6 earthquakes, or about 1,000 magnitude 5 earthquakes, to release the energy of one magnitude 7. Background seismicity does not meaningfully relieve a locked megathrust.
Myth: The Ring of Fire is a single fault. The Ring of Fire is a chain of plate boundaries that includes subduction zones, transform faults, and back-arc rifts. The Cascadia, Japan, and Andean subduction zones are separate plate boundaries that happen to ring the same ocean.
Frequently asked questions about earthquakes
What is the difference between magnitude and intensity?
Magnitude is one number per earthquake that describes the energy released at the source. Intensity describes felt shaking at a particular location and is reported in Roman numerals on the Modified Mercalli scale from I to XII. One earthquake has only one magnitude but many different intensities depending on distance from the epicenter, local geology, and building type.
How is an earthquake’s location determined?
Seismologists time the arrival of P-waves and S-waves at multiple stations. The longer the gap between P and S arrivals, the farther the station is from the source. Triangulating from at least three stations places the epicenter on a map and the hypocenter at depth. Modern catalogs use thousands of stations and pick arrivals automatically, locating most earthquakes within minutes.
Why do tsunamis follow some earthquakes but not others?
Tsunamis form mainly from undersea earthquakes that displace the seafloor vertically. Subduction-zone megathrusts, where the upper plate pops upward when locked-up strain releases, are the main tsunami sources. Strike-slip earthquakes move rock horizontally and produce little vertical seafloor change, so they generate weaker tsunamis even at high magnitude.
What happens during liquefaction?
In water-saturated, loosely packed sandy soil, strong shaking briefly disconnects grains from one another. Pore pressure rises, the soil temporarily behaves like a liquid, and structures can sink, tilt, or break apart. Liquefaction was responsible for much of the damage in the Marina District during the 1989 Loma Prieta earthquake, on land that had been built on bay fill.
Are induced earthquakes a real thing?
Yes. Deep wastewater injection associated with oil and gas extraction can raise pore-fluid pressure on pre-existing faults and trigger earthquakes. Oklahoma’s M 3+ rate jumped sharply after 2009 and exceeded California’s between 2014 and 2017. Reservoir filling, mining, and geothermal projects have also been linked to induced events, though hydraulic fracturing itself is responsible for only a small share.
How does ShakeAlert give warnings if prediction is impossible?
ShakeAlert detects fast P-waves at seismic stations near the epicenter and then sends a digital alert to phones, transit systems, and utilities ahead of the slower S-waves and surface waves. The system does not predict that an earthquake is about to happen. It detects one that has already started and rushes the warning ahead of the destructive shaking.
What is the deadliest earthquake in modern history?
The 2004 Indian Ocean (Sumatra-Andaman) magnitude 9.1 earthquake and tsunami killed roughly 228,000 people across 14 countries. Most deaths were caused by tsunami waves up to about 100 feet (30 m) tall. The fault rupture extended for about 750 to 800 miles (1,200 to 1,300 km) and lasted close to ten minutes.
Trivia question references throughout this topic’s Rookie, Curious, Sharp, and Expert quiz sets each cite a primary source for the specific fact tested.
An earthquake is the radiation of seismic waves from the sudden release of stored elastic strain along a fault, usually within Earth’s brittle crust or in subducting oceanic lithosphere extending into the mantle. The mechanism is captured by Harry Fielding Reid’s elastic rebound theory, formulated after the 1906 San Francisco earthquake and published in 1910: rocks across a fault deform elastically over years to centuries under tectonic loading, then snap back when fault strength is exceeded, releasing the accumulated strain as seismic waves. Modern measurements use moment magnitude (Mw), introduced by Hanks and Kanamori in 1979, which is computed from seismic moment, the product of average fault slip, ruptured fault area, and the shear modulus of the host rock. Moment magnitude does not saturate at the largest events, unlike the original 1935 Richter scale, and is the magnitude reported by USGS for moderate and large earthquakes worldwide.
Why seismology is non-intuitive
The Richter number quoted on a magnitude graphic is no longer Charles Richter’s 1935 quantity. Richter and Beno Gutenberg defined the original local magnitude (ML) for southern-California earthquakes recorded on Wood-Anderson torsion seismometers. ML saturates above about magnitude 7 because the seismometer’s response and the formula’s distance term cannot capture the long-period radiation of giant earthquakes. Moment magnitude resolves this by working from physical rupture quantities. The two scales are calibrated to agree where ML still works, so headline magnitudes look continuous across the historical catalog.
Magnitude and intensity are different physical quantities. Magnitude is a single number per earthquake, derived from waveform data and intended to characterize the source. Intensity, reported on the Modified Mercalli Intensity scale developed by Wood and Neumann in 1931, runs from I to XII and describes felt shaking and damage at specific locations. A single earthquake produces one magnitude but a continuous intensity field; a magnitude 7 event in a soft-sediment basin can yield higher local intensities than a magnitude 8 event in stable shield crust.
The Gutenberg-Richter frequency-magnitude relation, published by Gutenberg and Richter in 1944, expresses the cumulative count N of earthquakes at magnitude greater than or equal to M as a log-linear decrease, log N equals a minus bM. The slope b, called the b-value, is close to 1 in many tectonic regions, meaning that for every magnitude 7 event, the same region has roughly 10 magnitude 6 events and 100 magnitude 5 events. Departures from b near 1 are diagnostic: low b-values often appear in stress-loaded crust before large events; high b-values appear in volcanic and induced-seismicity settings. The relation underlies modern probabilistic seismic hazard analysis and is used in operational aftershock forecasts at USGS.
Aftershocks decay according to Omori’s law, formulated by Fusakichi Omori at Tokyo Imperial University in 1894 from the 1891 Nobi earthquake aftershock sequence. The rate of aftershocks declines as roughly 1 over time since the main shock, generalized in the modified Omori law with an exponent p typically between 0.7 and 1.5. Combined with the Gutenberg-Richter relation, this gives the empirical Reasenberg-Jones model and its modern descendants used to issue aftershock probability statements. Båth’s law, a separate empirical observation, states that the largest aftershock is on average about 1.2 magnitude units smaller than the main shock, a useful rule of thumb but not a strict bound.
The deepest earthquakes recorded reach about 435 miles (700 km), inside subducting oceanic slabs that have descended hundreds of miles into the mantle. The 2013 Sea of Okhotsk earthquake, magnitude 8.3 at a focal depth near 380 miles (609 km), is the largest deep-focus event on record, narrowly surpassing the 1994 Bolivia event (magnitude 8.2 at 631 km). Below 700 km the slab material is too ductile to fail brittlely under realistic stress conditions. The mechanism that allows brittle failure at high pressure and temperature is still debated, with current proposals invoking transformational faulting in metastable olivine, dehydration embrittlement of hydrous minerals, and shear melting along weak zones.
The interior of Earth is itself read out from earthquake records. S-waves cannot propagate through the liquid outer core, producing the S-wave shadow zone at angular distances beyond about 105 degrees from the source. P-waves are refracted at the core-mantle boundary, producing a P-wave shadow zone between about 103 and 142 degrees. In 1936 Inge Lehmann analyzed P-wave arrivals deep within the supposed shadow and concluded they had reflected from a solid sphere within the liquid outer core. The boundary she identified is now called the inner-core boundary (the term “Lehmann discontinuity” is sometimes applied loosely to it but is more often reserved for a distinct seismic-velocity feature near 220 km in the upper mantle). Her interpretation has been confirmed by decades of subsequent seismology, including the routine identification of PKIKP arrivals through the inner core.
Not every fault slip is a conventional earthquake. Episodic tremor and slip (ETS), discovered by Rogers and Dragert in 2003 from joint GPS and seismometer data, refers to recurring aseismic slip on the deeper plate interface beneath subduction zones, accompanied by a low-frequency tremor signal at 1 to 10 Hz. In northern Cascadia the events recur about every 14 months, last about two weeks, and release strain equivalent to a magnitude 6 to 7 earthquake spread over the duration of the episode rather than concentrated in seconds. ETS is now recognized in Cascadia, southwest Japan, Mexico, New Zealand, and elsewhere, and is one of several slow-slip phenomena that occupy the spectrum between locked faults and freely creeping ones.
Operational earthquake forecasting and warning are distinct from prediction. Long-term hazard maps, such as the USGS National Seismic Hazard Model, give probabilities of exceedance for various ground-motion levels over decades. Operational aftershock-sequence forecasts are issued in the days after a main shock. ShakeAlert, which began statewide public alerting in California in October 2019, detects an earthquake that has already started and races a digital warning ahead of the slower S-waves and surface waves. None of these tools is a deterministic prediction of time, place, and magnitude. USGS states plainly that prediction in that strict sense is not currently possible.
Key facts
Moment magnitude. Mw is computed from seismic moment, defined as the shear modulus times the rupture area times the average slip. The largest recorded event is the 1960 Valdivia, Chile earthquake at Mw 9.5, with a rupture length of approximately 500 miles (800 km). The 1964 Great Alaska earthquake at Mw 9.2 is the second largest and the largest in US history.
Body and surface waves. P-waves are compressional and travel at about 4 miles per second (6 km/s) in the crust. S-waves are shear and travel at roughly 60 percent of P-wave speed; they cannot propagate through liquid. Love waves are horizontally polarized surface waves; Rayleigh waves are retrograde elliptical surface waves. Surface waves arrive last, carry larger amplitudes than body waves at long periods, and dominate building damage at mid-rise to tall structures whose natural periods match.
Plate-tectonic settings. Divergent boundaries, including mid-ocean ridges and continental rifts, host shallow basaltic magmatism and shallow extensional earthquakes. Convergent boundaries host subduction-zone megathrusts capable of Mw 9 ruptures along strike lengths exceeding 600 miles (1,000 km). Transform boundaries such as the San Andreas Fault accommodate horizontal plate motion through strike-slip earthquakes. Intraplate earthquakes, including the 1811 to 1812 New Madrid sequence in the central US, occur far from active plate boundaries and are driven by reactivation of older fault zones.
The 1700 Cascadia earthquake. Reconstructed at approximately Mw 8.7 to 9.2 on the night of January 26, 1700, from a Japanese “orphan tsunami” that struck Honshu without local shaking, Pacific Northwest tribal oral histories, and coastal subsidence and tsunami sediment records. The Cascadia subduction zone has produced great earthquakes on average every 200 to 600 years over the Holocene.
The 2004 Sumatra-Andaman event. Mw 9.1 on December 26, 2004, with a fault rupture length of about 750 to 800 miles (1,200 to 1,300 km) and rupture duration close to ten minutes. The tsunami killed roughly 228,000 people across 14 countries, the deadliest earthquake-driven disaster in modern history.
The 2011 Tōhoku earthquake. Mw 9.1 on March 11, 2011, with seafloor displacements of more than 160 feet (50 m) at the trench. NASA JPL calculations attribute about 6.5 inches (17 cm) of figure-axis shift and roughly 1.8 microseconds of day-length shortening to the redistribution of mass.
The 2013 Sea of Okhotsk deep-focus event. Mw 8.3 at about 609 km depth, the largest deep-focus earthquake on record (surpassing the 1994 Mw 8.2 Bolivia event). Its rupture surprised seismologists who expected deep events to remain small. Mechanism candidates include transformational faulting in metastable olivine and shear melting along narrow zones.
The 1989 Loma Prieta earthquake. Mw 6.9 on October 17, 1989 on the San Andreas fault system, 31 minutes before Game 3 of the World Series at Candlestick Park. Liquefaction in San Francisco’s Marina District, built on early-20th-century bay fill, produced striking damage at large epicentral distance and motivated decades of soil-mechanics research.
Induced seismicity in Oklahoma. USGS attributes the post-2009 surge in central Oklahoma seismicity primarily to deep wastewater injection from oil and gas operations, which raises pore-fluid pressure on pre-existing faults. The Mw 5.8 event near Pawnee on September 3, 2016 is the largest documented injection-induced earthquake to date.
Did You Feel It? USGS DYFI launched in 1999 and has accumulated over 5 million citizen intensity reports. Its community internet intensity maps complement instrumental ShakeMap products, particularly in regions with sparse seismic instrumentation.
Frequency-size statistics. USGS estimates 500,000 detectable earthquakes per year globally, with roughly 100,000 felt and about 100 damaging. The Gutenberg-Richter relation predicts these counts to within factors of a few in stable seismic regions.
Common misconceptions at expert level
Misconception: The Richter scale is still in use as defined. The original 1935 Richter local magnitude (ML) is still computed for small to moderate earthquakes recorded at short distances, but USGS reports moment magnitude for moderate and large events. ML saturates above about magnitude 7. Moment magnitude is the operational standard.
Misconception: Megathrust earthquakes have unbounded magnitude. The largest plausible megathrust earthquake is constrained by along-strike rupture length, down-dip width, and average slip. Synthesis of historical and paleoseismic data places the practical upper limit near Mw 9.5, consistent with the 1960 Chile event. No Mw 10 earthquake has been recorded, and there is no known fault geometry on Earth that would readily host one.
Misconception: All earthquakes are brittle failures of cold crust. Deep-focus earthquakes occur within subducting slabs at depths of 300 to 700 km, well into the mantle. The mechanism is not standard brittle failure under classical Mohr-Coulomb conditions; it requires transformational faulting, dehydration embrittlement, or thermal-runaway shear, all of which are active research areas.
Misconception: P-waves and S-waves both pass through Earth’s outer core. S-waves are shear waves and require a medium that supports shear stress, which a liquid does not. They are blocked by the liquid outer core and produce the S-wave shadow zone beyond about 105 degrees from the source. This observation is what established that the outer core is liquid.
Misconception: Foreshocks can be identified in real time. USGS and modern operational forecasting find no real-time signal that distinguishes a small earthquake from one that will be followed by a larger event. A foreshock is reclassified as such only after a larger main shock occurs. Most small earthquakes are not foreshocks.
Misconception: Animals reliably predict earthquakes hours or days in advance. USGS finds no reproducible mechanism for animal prediction. The clearest documented effect is short-range: many animals detect the faster, weaker P-wave a few seconds before humans feel the slower S-wave.
Misconception: A few small earthquakes prevent the big one. Energy scales as roughly 32 to the magnitude difference per whole-number step. Releasing the energy of a Mw 7 event by Mw 5 background seismicity would require about 1,000 such events. Small-earthquake occurrence does not measurably relieve a locked megathrust or transform fault.
Misconception: Cascadia recurrence is a regular cycle. Paleoseismic records from coastal subsidence horizons, turbidite sequences, and tsunami sediments indicate Cascadia great earthquakes recur on a 200 to 600 year envelope, with substantial variability and clustering. Treating the recurrence as a fixed period is unsupported.
Frequently asked questions
Why does moment magnitude not saturate?
Moment magnitude is computed from seismic moment M0, equal to the shear modulus times rupture area times average slip. As an earthquake gets bigger, both rupture area and slip continue to grow without a fixed instrumental ceiling, so M0 grows without saturation. The local magnitude ML, by contrast, is computed from peak ground motion in a fixed period band on a specific seismometer response. Above about Mw 7, additional seismic moment goes mostly into long-period radiation that ML’s bandpass cannot record, so the ML number stops increasing.
What controls whether an earthquake generates a damaging tsunami?
Tsunami generation requires sudden vertical displacement of a substantial volume of seawater, which in turn requires vertical slip on a fault under the ocean. Subduction-zone megathrust earthquakes are efficient tsunami sources because the upper plate snaps seaward and upward over rupture lengths of hundreds of miles. Strike-slip earthquakes, even at high magnitude, displace the seafloor mostly horizontally and produce weaker tsunamis. Slow-rupture “tsunami earthquakes” can generate tsunamis disproportionate to their magnitudes by rupturing slowly through soft sediment near the trench.
What is episodic tremor and slip and why does it matter?
Episodic tremor and slip is a coupled phenomenon discovered in northern Cascadia in 2003: aseismic slow slip on the deeper plate interface, accompanied by non-impulsive tremor at 1 to 10 Hz. In Cascadia it recurs about every 14 months and lasts about two weeks. ETS matters because the slow-slip patch lies down-dip of the locked seismogenic zone, so each ETS cycle transfers stress upward onto the part of the megathrust capable of generating a great earthquake. Tracking ETS is one input to long-term hazard estimates for the Pacific Northwest.
What does the b-value of the Gutenberg-Richter law tell us?
The b-value sets the relative number of large to small earthquakes in a region. b = 1 means a tenfold drop in count for each unit of magnitude. Lower b-values, around 0.7, indicate proportionally more large events relative to small ones, often associated with stress-loaded crust before large earthquakes. Higher b-values, sometimes 1.5 to 2 or higher, are typical of volcanic systems and some induced-seismicity settings, where small fracturing dominates. Probabilistic hazard models are sensitive to b at the few-percent level, so b is one of the most carefully estimated parameters in operational seismology.
How accurately can earthquakes be located?
Modern multistation arrivals, often picked automatically and refined with seismic-velocity models, locate moderate earthquakes to within a kilometer in well-instrumented regions. Depth is harder than horizontal location, especially for shallow events recorded only at distance, where teleseismic depth phases (pP, sP) are needed. Special techniques such as double-difference relocation can resolve fault structure to sub-kilometer precision in regions with dense seismicity.
What does the future of earthquake science look like?
Several research directions are active: dense urban seismic and geodetic arrays, real-time machine-learning models for phase picking and rupture imaging, distributed acoustic sensing using existing fiber-optic cables, and improved physics-based simulations of fault-system behavior. Earthquake early warning is being expanded and refined globally, with seconds-to-tens-of-seconds of warning available across the US west coast and parts of Japan, Mexico, and Taiwan. Deterministic prediction at useful lead times remains an open problem.
What is the role of citizen science in modern seismology?
The USGS Did You Feel It tool collects structured felt-shaking reports from the public and aggregates them into community internet intensity maps that supplement instrumental ShakeMap output. Smartphone-based projects such as MyShake use the accelerometers in millions of phones as a distributed seismic network. Crowdsourced photo and damage reports complement official damage assessments after major events. The integration of citizen science with instrumental data has measurably improved post-event response in the last two decades.
Trivia question references throughout this topic’s Rookie, Curious, Sharp, and Expert quiz sets each cite a primary source for the specific fact tested.
