Seismologists locate an earthquake by comparing the arrival times and shapes of seismic waves recorded at several stations. An earthquake begins at the hypocentre, or focus, below the surface; the epicentre is the point directly above it. P waves usually arrive first because they travel faster than S waves. The growing time gap between them gives information about distance from a station, but one station cannot normally determine a unique direction. A network combines readings and uses a model of how wave speed varies through Earth to estimate latitude, longitude, depth and origin time. Modern location is an iterative calculation, not simply three perfect circles drawn on a map. Station distribution, noisy signals and imperfect velocity models create uncertainty, so catalogues may revise locations as additional data and specialist analysis become available.
Key takeaways
- The hypocentre is the underground rupture origin; the epicentre lies above it.
- P waves usually reach a station before S waves.
- Arrival-time differences constrain distance but not direction from one station.
- Multiple stations and a velocity model are required.
- Depth is often less certain than horizontal position.
- Locations and magnitudes are different calculations.
What a seismometer records
A seismometer measures ground motion at its location. A seismograph is the recording system, and the resulting time series is a seismogram. Modern stations commonly measure motion in three directions and attach precise timing, often derived from satellite navigation systems.
The trace includes local noise from weather, traffic and instruments as well as earthquake waves. Analysts or algorithms identify arrivals called phases. An arrival pick includes uncertainty; the first movement may be weak or obscured.
P waves and S waves
P waves are compressional body waves and generally travel faster. S waves involve shear motion and do not pass through liquid in the same manner. Their different speeds help reveal Earth’s interior, as discussed in the structure of Earth.
The P–S time difference tends to increase with distance, but the conversion is not one universal kilometres-per-second rule. Speeds depend on rock, temperature, pressure and path. Travel-time curves and three-dimensional models provide the relationship.
Why one station is not enough
A measured distance places the source somewhere on a circle around a station on a simplified surface map. A second station narrows the possibilities, and a third may identify an intersection. This classroom triangulation illustrates the geometry.
Real location is closer to trilateration than angle-based triangulation, and the source is three-dimensional. Stations and travel paths lie on a curved, layered Earth. More than three stations make the estimate more robust and allow residual errors to be evaluated.
Origin time, location and depth
Arrival time equals the earthquake’s unknown origin time plus the wave’s travel time. The inversion must solve these unknowns together. A computer starts with a trial hypocentre, predicts arrivals, compares them with observations and adjusts the solution to reduce differences.
Depth is difficult when all stations are far away or on one side. Waves reflected near the source, depth phases and dense local networks improve it. Some catalogues fix a poorly constrained depth to a standard value; readers should check flags and uncertainty.
The importance of a velocity model
A velocity model describes how fast seismic waves are expected to move through different regions and depths. If the model is too fast along a path, the calculated source may shift to compensate. Sedimentary basins, mountains and subduction zones create complex structures.
Local networks build regional models using known explosions, well-constrained events and tomography. Global catalogues use broader Earth models. Two agencies can report slightly different locations without either acting carelessly because their data and models differ.
Station geometry and uncertainty
Stations surrounding a source provide better horizontal constraint than stations arranged on one side. Oceanic earthquakes may be located mainly by land stations, producing larger uncertainty. Ocean-bottom instruments can improve coverage but are harder to deploy and maintain.
Location software estimates an error region, but formal errors may omit model inadequacy. Reporting more decimal places does not create accuracy. A coordinate should be read with its uncertainty and network geometry.
Epicentre is not the whole rupture
Large earthquakes rupture faults over tens or hundreds of kilometres. The hypocentre marks where rupture began, not where all energy was released or where shaking was greatest. Aftershocks outline parts of the fault and finite-fault models estimate how slip spread.
Damage depends on rupture direction, ground conditions, building vulnerability and distance. Linking every impact to a dot labelled epicentre oversimplifies the physical event described in what causes earthquakes.
Location, magnitude and intensity
Location uses timing and wave propagation. Magnitude estimates earthquake size from recorded amplitudes or seismic moment with corrections. Intensity describes observed shaking and effects at a place. One earthquake has one reported magnitude within a scale but many local intensities.
An early automatic magnitude and location can change as more stations report and saturated instruments are replaced with appropriate measurements. Revision is a normal part of rapid science, not evidence that an earthquake itself changed.
Worked example: a coastal earthquake
A coastal network detects a P wave at inland stations, then S waves. Initial geometry places the event offshore with broad depth uncertainty. Additional island and ocean-bottom data arrive, shifting the hypocentre several kilometres and showing a shallow subduction-interface rupture.
The location is combined with mechanism, sea-level observations and seafloor displacement models when assessing a tsunami. As the formation of tsunamis explains, magnitude and offshore location alone do not guarantee a damaging wave; the style and amount of displacement matter.
From individual events to tectonic patterns
Accurate catalogues reveal dipping earthquake zones, fault structures and migrating sequences. Patterns support plate tectonics, while unusual clusters can guide field investigation.
Location does not mean reliable short-term prediction. Networks can rapidly detect earthquakes and estimate evolving sequences, but identifying a fault does not provide an exact future date, time and magnitude.
Automatic systems and analyst review
Continuous networks detect abrupt changes in signal and associate arrivals across stations. An automatic system can publish a preliminary origin within minutes, supporting emergency response. Algorithms must distinguish earthquakes from quarry blasts, sonic booms, storms and unrelated events close in time.
Analysts review phase picks, add stations, remove poor readings and choose an appropriate velocity model. Relative-location methods compare similar earthquakes and can resolve fault structures more finely than ordinary absolute locations because shared path errors cancel.
Machine learning can improve detection of small events and phase picking, but training data may represent some regions and instruments better than others. Performance must be tested on independent data, and catalogues should record method changes because an apparent rise in earthquakes can result from improved detection.
Earthquake early warning is not prediction
Early-warning systems detect an earthquake after it begins. Stations near the source recognise fast P waves and send alerts before stronger shaking reaches more distant places. The available warning may be seconds, and areas close to the rupture can receive little or none.
Prediction would specify an event before rupture with useful time, place and magnitude. Seismology cannot currently provide reliable deterministic short-term predictions. Hazard maps instead estimate long-term shaking likelihood from faults, past events and ground conditions.
Clear communication separates these services. A rapid location supports response, an early warning exploits wave-travel time, and a forecast describes probabilities. Confusing them creates unrealistic expectations and can undermine trust when a real system performs within its stated limits.
Wider significance and responsible use
Dense aftershock surveys can refine a fault plane after a damaging event. Temporary instruments lower the detection threshold and reveal structures missed by the permanent network. Because coverage improves after deployment, an apparent rise in small earthquakes may reflect observation rather than an accelerating process.
For nuclear-test monitoring, analysts combine location with wave characteristics and infrasound or hydroacoustic data. Explosions and earthquakes can differ statistically, but classification uses multiple observations and context. Seismology supports verification without making every unusual signal automatically artificial.
Citizen smartphone sensors may supplement professional networks by showing shaking distribution, but device placement and timing vary. They are most useful when calibrated against established stations. Broad participation can improve awareness while professional metadata and quality control preserve scientific comparability.
Further questions for investigation
Calibration events with accurately known locations help reveal systematic network bias. Quarry blasts, controlled sources and well-located aftershocks compare predicted and observed arrivals. Corrections can then improve routine catalogues, although a model tuned to one area may not transfer across a geological boundary.
Location quality influences every later inference. A depth error can place an event on the wrong fault or make a crustal earthquake appear within a subducting slab. Tectonic interpretation should use quality thresholds and test whether patterns persist when uncertain events are excluded.
Conclusion
Earthquake location is an evidence-rich inverse problem. Networks turn timed ground motion into an estimate of origin time and three-dimensional position using models of seismic-wave travel. P and S arrivals provide the basic clues, while station coverage, phase identification and Earth structure control precision. Treating the epicentre as an uncertain starting point—and distinguishing it from depth, magnitude, intensity and rupture extent—leads to a much more accurate understanding of seismic reports.
Frequently asked questions
What is the difference between epicentre and hypocentre?
The hypocentre is the earthquake’s starting point underground; the epicentre is the point on the surface directly above it.
Why are P waves detected first?
They generally travel faster through Earth materials than S waves.
Do three stations always give an exact location?
No. The three-circle picture is simplified; real Earth structure, depth and picking uncertainty require more data and modelling.
Why can agencies publish different depths?
They may use different stations, phase picks and velocity models, and depth is often weakly constrained.
Can locating small earthquakes help assess faults?
Yes. Dense patterns can reveal active structures, although they do not provide exact short-term predictions.
