How Earthquakes Are Measured

An earthquake is a physical event happening kilometres underground, often beneath the ocean. Everything we know about it is inferred from the waves it sends to the surface.

How a seismometer works

The core problem in measuring ground motion is that any instrument you build is sitting on the very ground that is moving. You need a stationary reference — and there isn't one.

The classical solution exploits inertia. A heavy mass is suspended from a spring or pendulum inside a frame anchored to the ground. When the ground shakes, the frame moves with it, but the inertia of the mass resists that motion, so it stays comparatively still. The relative displacement between frame and mass is a proxy for ground motion.

Modern instruments are force-balance seismometers. Rather than letting the mass move freely, an electronic feedback loop applies a force to hold it in place, and the current required to do so is recorded. This delivers vastly better linearity and a much wider dynamic range, letting a single instrument record both a distant magnitude 8 and a local magnitude 1.

A complete station records three orthogonal components — two horizontal and one vertical — because ground motion is three-dimensional and different wave types dominate different components. Broadband instruments capture periods from fractions of a second to hundreds of seconds. Strong-motion accelerometers sit alongside them to record violent nearby shaking that would drive a sensitive broadband sensor off scale.

Locating an earthquake

Location relies on the fact that P-waves and S-waves travel at different speeds. P-waves move at roughly 6 km/s in the crust, S-waves at roughly 3.5 km/s. They leave the focus at the same instant, so the further a station is from the earthquake, the larger the gap between the two arrivals.

That gap converts directly to distance. One station therefore establishes that the earthquake lies somewhere on a circle around it. A second station narrows this to two intersection points. A third resolves the ambiguity. This is triangulation, and it is the principle taught in every introductory course.

Real networks do considerably better. Hundreds of stations may contribute, and rather than drawing circles, software solves an inverse problem: find the latitude, longitude, depth and origin time that best explain all observed arrival times given a velocity model of the Earth's interior. The velocity model matters — the Earth is not uniform, and errors in assumed wave speeds translate directly into location errors.

Depth is consistently the least well constrained parameter, particularly for offshore events where all stations lie on one side. Analysts refine it using depth phases such as pP, which reflect off the surface above the focus.

Determining magnitude

Once the location is known, magnitude follows from the amplitude of recorded waves corrected for distance. Different magnitude types use different waves and different frequency bands:

• ML (local magnitude) — Richter's original, from peak amplitude on a nearby station. Used for small local events.
• mb (body-wave magnitude) — from P-wave amplitude. Fast to compute, but saturates above about magnitude 6.5.
• Ms (surface-wave magnitude) — from 20-second surface waves. Saturates above about magnitude 8.
• Mw (moment magnitude) — from seismic moment, computed by modelling the full waveform. Does not saturate, and is the standard for any significant earthquake.

Automatic systems produce a first magnitude within a minute or two using the fastest available method. Analysts then compute Mw properly, which is why reported magnitudes are frequently revised in the hours after an event.

Intensity: what people actually experienced

Magnitude describes the source. It says nothing about what happened in any particular town. For that, seismologists use intensity, and the standard is the Modified Mercalli Intensity (MMI) scale.

MMI is observational rather than instrumental. It is assigned from reports of what people felt and what was damaged:

• I–II — Not felt, or felt only by a few people at rest, especially on upper floors.
• III — Felt indoors, often mistaken for a passing truck. Hanging objects swing.
• IV–V — Felt by nearly everyone. Dishes and windows rattle, sleepers wake, unstable objects overturn.
• VI–VII — Felt by all. Plaster cracks, chimneys break, poorly built structures are damaged.
• VIII–IX — Substantial damage to ordinary buildings; partial collapse. Well-designed structures damaged.
• X–XII — Most masonry and frame structures destroyed. Ground visibly cracks, rails bend, near-total destruction.

A single earthquake produces a whole map of intensities, highest near the epicentre and decreasing outwards, but heavily modified by local geology. Agencies now crowdsource this directly — the USGS "Did You Feel It?" system converts public reports into intensity maps within minutes, often producing better coverage than instruments alone.

From detection to publication

The sequence after a rupture is tightly choreographed. Seismic waves reach the nearest stations within seconds and data streams continuously to processing centres. Automatic detection algorithms identify coincident arrivals across multiple stations, associate them into a single event, and compute a preliminary location and magnitude — typically within two to five minutes.

That preliminary solution is published immediately, which is why the earliest reports carry a status of "automatic". Human analysts then review the picks, correct errors, add stations and recompute, after which the status changes to "reviewed". For significant events, finite-fault models describing how slip was distributed across the rupture surface follow over the subsequent hours and days.