Seismic activity refers to any vibration or movement within the Earth’s crust, and most of it stems from the slow, relentless motion of tectonic plates grinding past, pulling apart, or diving beneath one another. But plate tectonics is only the headline act. Earthquakes also originate hundreds of kilometers deep in the mantle, inside glaciers, beneath volcanoes, and even as a consequence of human industry. The science of seismicity stretches from everyday tremors too faint to feel to rare catastrophic ruptures that reshape coastlines, and it connects to questions people are genuinely curious about: why some places shake more than others, whether quakes can be predicted, and what the seconds before strong shaking can buy you.
What Happens at Plate Boundaries
The outer shell of the Earth is broken into roughly a dozen major plates, and most seismic activity clusters where those plates meet. When two plates converge, one often slides beneath the other in a process called subduction. Stress accumulates along the contact zone until the locked fault gives way in a sudden slip, releasing energy as seismic waves. The same basic story plays out at transform boundaries, where plates slide horizontally past each other, and at divergent boundaries, where plates pull apart and magma wells up to fill the gap.
A common simplification is that faults snap and then fully relax, like a rubber band returning to its resting shape. In reality, that elastic rebound is incomplete. Modeling of thrust faults shows that while the motions during and between earthquakes are opposite in direction, they do not perfectly cancel over a full seismic cycle. Gravity inhibits complete strain release on dip-slip faults, so permanent deformation gradually accumulates in the surrounding rock over many earthquake cycles.1Journal of Geophysical Research: Solid Earth. Accumulation of permanent deformation during earthquake cycles on reverse faults That leftover strain is part of what builds mountain ranges and reshapes landscapes over geological time.
Deep-Focus Earthquakes
Not all earthquakes happen near the surface. Some originate at depths between 400 and 700 kilometers, well inside the upper mantle, where temperatures and pressures are so extreme that rock should flow rather than fracture. These deep-focus earthquakes puzzled scientists for decades because brittle fracturing, the mechanism behind shallow quakes, should be physically impossible at those depths.
The leading explanation involves a mineral transformation. When a cold slab of oceanic plate sinks deep into the mantle, pockets of the mineral olivine survive longer than they should, persisting in a form that is unstable at those pressures. As the slab warms to a critical temperature, the olivine begins to transform into a denser crystal structure. That transformation can trigger sudden faulting.2Geological Society, London, Special Publications. The failure mechanism for deep-focus earthquakes Laboratory experiments have confirmed that when olivine-analog materials are squeezed at high pressure and heated through the right temperature window, fractures nucleate right at the onset of the phase transition. Those fractures propagate fast enough to generate intense acoustic emissions, the lab equivalent of seismic waves, and the emissions follow the same statistical patterns that real deep-focus earthquakes do.3PubMed. Deep-focus earthquake analogs recorded at high pressure and temperature in the laboratory
One counterintuitive finding is that colder, faster-sinking slabs actually produce more deep-focus earthquakes, not fewer. Experiments on olivine-analog materials show that transformational faulting can keep propagating through a process involving shear-enhanced melting of the newly formed high-pressure phase. Because the balance between strain rate and transformation rate shifts in cold, fast slabs, those slabs end up hosting transformational faulting at higher temperatures, which generates more frequent deep earthquakes.4Geology. Deep-focus earthquakes: From high-temperature experiments to cold slabs This helps explain why the deepest earthquakes cluster in specific subduction zones where old, cold oceanic plates are plunging rapidly into the mantle.
Slow Slip and Silent Earthquakes
Some fault movement releases energy so gradually that no one feels a thing. These slow slip events, sometimes called silent earthquakes, unfold over days or weeks rather than seconds. They were essentially invisible until sensitive GPS instruments began detecting the centimeter-scale ground shifts they cause. Most slow slip events occur just beyond the deeper edge of the zone that produces regular earthquakes, in a transition region where fault behavior shifts from unstable, stick-slip motion to steady creep.5Reviews of Geophysics. Slow slip events and seismic tremor at circum‐Pacific subduction zones
The distinction matters because slow slip events can redistribute stress onto the locked portion of a fault, potentially nudging it closer to failure. In subduction zones around the Pacific, slow slip has been documented repeatedly in the years and months preceding large earthquakes, though the relationship is not straightforward enough to use for prediction. Researchers have also found that slow slip is often accompanied by a faint seismic hum called tectonic tremor, a signal so weak it blends into background noise without specialized processing.
Earthquakes Triggered by Human Activity
People have been accidentally causing earthquakes for longer than most realize, and the problem has grown sharply in the past two decades. The clearest cases involve injecting large volumes of wastewater deep underground, a common byproduct of oil and gas production. When fluid is pumped into porous rock formations, the rising pore pressure reduces the friction holding nearby faults in place. If those faults were already close to their breaking point, even a tiny pressure increase can set them off.
Oklahoma became the poster child for this phenomenon. Through the early 2010s, the state went from experiencing a handful of earthquakes per year to hundreds, some large enough to damage buildings. Research concluded that stress on pre-existing faults, amplified by pore-pressure diffusion and poroelastic effects from massive wastewater injection, was the primary driver.6PubMed Central. Pore-pressure diffusion, enhanced by poroelastic stresses, controls induced seismicity in Oklahoma Similar patterns have been documented elsewhere. Modeling of the 2014 Milan, Kansas earthquake showed that pore pressure at the quake’s depth had risen past a minimum triggering threshold before the event occurred, on a fault that was already critically stressed before injection began.7Geochemistry, Geophysics, Geosystems. Numerical Models of Pore Pressure and Stress Changes Along Basement Faults Due to Wastewater Injection: Applications to the 2014 Milan, Kansas Earthquake In Colorado’s Raton Basin, the timing and location of early earthquake clusters lined up with modeled pore pressure fronts advancing outward from wastewater injection wells.8Journal of Geophysical Research: Solid Earth. Pore Pressure Diffusion and Onset of Induced Seismicity
Wastewater disposal is not the only culprit. Filling large reservoirs behind hydropower dams loads the crust with billions of tons of water, which can deform the ground and alter pore pressures in underlying rock. Studies of large hydropower stations in China have documented both crustal deformation and triggered seismic activity following reservoir impoundment and discharge cycles.9The Innovation Geoscience. Crustal response to water loads and expansion of triggered seismicity around the Xiluodu Reservoir, Southwest China Geothermal energy extraction and underground mining can also produce induced seismicity, though typically at smaller magnitudes than what wastewater injection has caused.
How Aftershock Sequences Behave
After a large earthquake, smaller quakes follow in a pattern that is remarkably consistent worldwide. The rate of aftershocks decays roughly as the inverse of time since the mainshock, a relationship first described in the late 1800s and refined into what seismologists know as the modified Omori law. At the same time, the size distribution of earthquakes in any region follows a predictable curve: many small events, fewer moderate ones, and rare large ones. Researchers have shown that these two patterns, along with the general relationship between a mainshock’s magnitude and its largest aftershock, can be unified into a single mathematical framework.10Geophysical Journal International. Generalized Omori–Utsu law for aftershock sequences in southern California
In practice, this means scientists can make useful short-term forecasts of aftershock rates within hours of a mainshock. Analysis of a decade of seismicity in West Sumatra, for example, found that the aftershock sequence from a major 2009 earthquake followed the expected decay pattern, with a moderately slow falloff in activity over time and a size distribution dominated by small-to-moderate events.11KONSTAN – JURNAL FISIKA DAN PENDIDIKAN FISIKA. Statistical Analysis of Seismicity in West Sumatra (2015–2025) and Aftershock Decay of the 2009 Padang Earthquake Using Gutenberg–Richter and Omori–Utsu Laws These forecasts are not predictions of specific events but probability statements: the chance of a magnitude-5 aftershock in the next week, for instance. That information helps emergency managers decide how long to keep damaged buildings off-limits.
Volcano Monitoring Through Seismic Signals
Volcanoes generate their own characteristic seismic signals, and reading them correctly can save lives. One of the most distinctive is harmonic tremor, a sustained, rhythmic vibration quite different from the sharp snap of a tectonic earthquake. Harmonic tremor often signals that magma or volcanic fluids are moving underground. At one Japanese volcano, researchers traced harmonic tremor to the migration of magmatic fluid deep within the volcanic system.12Earth, Planets and Space. Harmonic tremor from the deep part of Hakone volcano
Automated detection is becoming essential as monitoring networks expand. During a five-month period of heightened eruptions at one volcano, an algorithm detected over 1,400 minutes of harmonic tremor, with the tremor episodes generally preceding bursts of explosive activity.13Geophysical Research Letters. Automated detection and characterization of harmonic tremor in continuous seismic data That kind of lead time, even if measured in minutes to hours, can trigger evacuation warnings. Seismic monitoring remains the backbone of eruption forecasting, often more reliable than visual observation or gas measurements, especially at remote or cloud-covered volcanoes.
Seismic Signals From Ice
Glaciers and ice sheets are surprisingly noisy. A whole subfield called cryoseismology has emerged around the seismic signals generated by ice-related processes, from the cracking of glaciers to the calving of icebergs to the grinding of ice sheets against bedrock. These signals span a wide range of frequencies and magnitudes, with some glacial events producing seismic waves equivalent to a magnitude-7 earthquake.14Reviews of Geophysics. Cryoseismology
At the calving front of Bowdoin Glacier in Greenland, researchers found that microseismic activity tracked closely with the glacier’s flow speed. Tidal cycles and surface melt rates both influence how fast the ice moves, and when the glacier stretches longitudinally, it generates more seismic events. Falling tides and higher melt rates both accelerated ice motion and increased the stretching that drives this seismicity.15Geophysical Research Letters. Tide‐modulated ice flow variations drive seismicity near the calving front of Bowdoin Glacier, Greenland This kind of monitoring gives glaciologists a way to track ice dynamics in real time, which is increasingly valuable as ice sheets respond to warming temperatures. A seismometer planted on or near a glacier can reveal processes that are invisible from satellite images.
Earthquake Early Warning Systems
You cannot predict earthquakes days in advance with any useful reliability, but you can detect them seconds after they start and beat the most damaging waves to distant cities. Earthquake early warning systems exploit the fact that the first waves to arrive from an earthquake, called P-waves, travel faster but carry less destructive energy than the S-waves and surface waves that follow. If a network of sensors can identify an earthquake from its P-waves, it can send an alert to people farther away before the strong shaking reaches them.
The race is for speed. Machine learning systems have pushed detection windows down to remarkably short intervals. One system, called E3WS, can detect, locate, and estimate the magnitude of an earthquake from just three seconds of P-wave data recorded at a single station, offering faster estimates than traditional multi-station systems and providing additional seconds for protective actions.16Journal of Geophysical Research: Solid Earth. Earthquake Early Warning Starting From 3 s of Records on a Single Station With Machine Learning Other approaches use gradient-boosted machine learning models to predict ground shaking intensity from P-wave features extracted in one-to-three-second windows.17Geophysical Journal International. Real-time prediction of distance and PGA from P-wave features using Gradient Boosting Regressor for on-site earthquake early warning applications
Even low-cost sensors deployed by ordinary citizens can contribute. A study of a community-engaged early warning network in New Zealand tested four different detection algorithms against data from a magnitude-5.8 earthquake and found that a wavelet-based approach gave the best balance of catching real events while avoiding false alarms.18New Zealand Journal of Geology and Geophysics. Performance analysis of P‐wave detection algorithms for a community‐engaged earthquake early warning system – a case study of the 2022 M5.8 Cook Strait earthquake The seconds these systems buy you may sound trivial, but they are enough to stop trains, open fire station doors, pause surgeries, and send millions of phone alerts telling people to drop, cover, and hold on.
What Paleoseismology Reveals About Earthquake Timing
Since instrumental earthquake records only go back about a century, understanding long-term earthquake patterns requires digging, literally. Paleoseismology involves excavating trenches across fault lines to find physical evidence of past ruptures: displaced soil layers, buried fissures, folded sediment, and sand layers injected upward by shaking. Combined with radiocarbon dating, these features let researchers reconstruct earthquake histories stretching back thousands of years.
One persistent finding from this work is that earthquakes on a given fault do not recur on a neat schedule. Along the Garlock Fault in California, trench studies identified six well-resolved earthquakes spanning roughly the last 7,000 years. The intervals between events ranged from about 215 years to as many as 3,300 years, a variation of more than tenfold.19Journal of Geophysical Research: Solid Earth. Irregular recurrence of paleoearthquakes along the central Garlock fault near El Paso Peaks, California In southern Italy, trenching along the Irpinia Fault found evidence of four pre-1980 earthquakes over the past 8,600 years, giving an average recurrence interval of about 2,150 years, though individual gaps varied by a factor of two.20Journal of Geophysical Research: Solid Earth. Paleoseismology along the 1980 surface rupture of the Irpinia Fault: Implications for earthquake recurrence in the southern Apennines, Italy Nearby, the Cittanova Fault showed a possible return time of 1,350 to 1,800 years for events comparable to the devastating 1783 Calabrian earthquakes.21Journal of Geophysical Research: Solid Earth. Paleoseismology along the Cittanova fault: Implications for seismotectonics and earthquake recurrence in Calabria (southern Italy)
The irregularity matters enormously for hazard assessment. If someone tells you a fault has an “average recurrence interval” of 2,000 years, that does not mean you are safe for 2,000 years after the last quake. The next event could come centuries sooner or later. Paleoseismic data is the best tool available for characterizing this variability, but it depends on finding sites where sediment has recorded each event cleanly, which limits its application to certain geological settings.
Tsunamis and the Role of Upper-Plate Structure
Large undersea earthquakes can displace the ocean floor abruptly enough to generate tsunamis, and the size of the resulting wave depends on more than just the earthquake’s magnitude. The stiffness of the rock in the overriding plate plays a significant role. Modeling has shown that when the upper plate is more compliant, meaning softer and more deformable near the trench, a rupture that extends to shallow depths produces larger seafloor deformation and bigger tsunami waves than the same slip on a stiffer plate would.22Journal of Geophysical Research: Solid Earth. The Influence of Depth‐Varying Elastic Properties of the Upper Plate on Megathrust Earthquake Rupture Dynamics and Tsunamigenesis This is one reason why two megathrust earthquakes of similar magnitude can produce very different tsunami hazards depending on local geology.
The Earthquake Prediction Problem
Reliable short-term earthquake prediction, telling a specific community that a damaging earthquake will strike within a specific window of days, remains beyond current science. It is not for lack of trying. Researchers have investigated a wide array of potential precursors, from changes in electromagnetic radiation across a broad frequency range to fluctuations in radon gas concentrations in soil, groundwater, and air.23Geosciences. Electromagnetic and Radon Earthquake Precursors Satellite-based observations of ground deformation and thermal anomalies have been examined too. The problem is consistency: a signal that appears before one earthquake may be absent before the next, and similar signals show up without any earthquake following.
Animal behavior has been another popular avenue. Reports of unusual animal behavior before earthquakes go back centuries, and there are plausible sensory mechanisms that could explain some of them. Certain fish can detect electric field changes as small as a hundred-thousandth of a volt per meter, and many animals are far more sensitive than humans to low-frequency sounds and ground vibrations below 50 Hz. Gas releases from stressed rock could also provide olfactory cues that some animals detect before humans notice anything.24Reviews of Geophysics. Unusual animal behavior before earthquakes: A review of possible sensory mechanisms The sensory capabilities are real, but converting anecdotal sightings of agitated animals into a reliable warning system has proven impossible so far. Animals behave strangely for many reasons, and filtering earthquake-related behavior from everyday restlessness requires a predictive framework that does not yet exist.
Seismicity on the Moon and Mars
Earth is not the only world that shakes. NASA’s Apollo missions left seismometers on the Moon that recorded moonquakes for years, and the InSight lander did the same on Mars from 2018 to 2022. These datasets have revealed how different planetary interiors respond to seismic waves. One striking contrast is that seismic waves scatter much more intensely on the Moon than on Earth, roughly ten times more, because the lunar surface is shattered by billions of years of meteorite impacts with no wind or water to smooth out the damage.25Journal of Geophysical Research: Planets. Quantitative Evaluation of the Lunar Seismic Scattering and Comparison Between the Earth, Mars, and the Moon This scattering makes lunar seismograms look dramatically different from Earth’s: instead of sharp arrivals that decay quickly, moonquakes ring for an hour or more as energy bounces around in the fractured crust.
Mars sits somewhere between the two. Its seismograms show more scattering than Earth’s but less than the Moon’s, consistent with a surface that has been weathered but not pulverized to the same degree. Marsquakes tend to be modest in magnitude, and the planet lacks active plate tectonics in the way Earth has them, so its seismicity likely comes from cooling and contraction of the crust, volcanic stresses, and possibly tidal forces from its moons. Studying seismic waves on other worlds is not just a curiosity; it is the primary way scientists map the interiors of planets where they cannot drill.