Gravitational Time Dilation and GPS: Why Satellite Clocks Run Fast

What gravitational time dilation actually is

Gravitational time dilation is a direct consequence of Einstein's general relativity (1915): the flow of time itself depends on where you are in a gravitational field. A clock sitting deeper in a gravitational potential well — closer to a mass — ticks measurably slower than an identical clock higher up where the field is weaker. This is not an illusion, a Doppler artifact, or a mechanical effect on the clock's works; it is the local rate of proper time that changes.

The effect is governed by the metric of curved spacetime. For a weak, static field the fractional slowing between two heights is:

• Δτ/τ ≈ ΔΦ / c², where ΔΦ is the difference in gravitational potential and c is the speed of light.
• Near Earth's surface, ΔΦ/c² ≈ g·Δh / c² ≈ 1.09 × 10⁻¹⁶ per meter of height — about one part in 10¹⁶ for every meter you climb.

Because a GPS satellite sits far above the surface where the potential Φ is higher (less negative), its clock runs faster than one on the ground. The same physics is why clocks on a mountaintop beat clocks in a valley — a difference now measurable with a tabletop optical clock.

The mechanism: deriving the satellite clock rate

Start from the Schwarzschild metric, which describes spacetime outside a spherical mass M. For a clock at radius r moving with speed v, the elapsed proper time dτ relative to coordinate time dt (a clock infinitely far away) is, to first order:

• dτ/dt ≈ 1 + Φ/c² − v²/(2c²), where Φ = −GM/r is the (negative) gravitational potential.

Two terms compete. The gravitational term Φ/c² makes deeper clocks slower — so raising the satellite makes its clock faster. The kinetic term −v²/2c² is the special-relativistic effect: any moving clock runs slow, so the satellite's orbital motion drags its rate back down.

Comparing a satellite clock to a ground clock, you subtract the two proper-time rates. The gravitational difference between r = 20,200 km altitude and Earth's surface gives +45.8 μs/day. The velocity term, with orbital speed v ≈ 3.87 km/s, gives −7.2 μs/day. The Earth's surface also contributes a rotation-velocity term, all folded into the geoid reference. The net is a satellite clock that gains +38.6 μs/day — a fractional rate of about +4.46 × 10⁻¹⁰.

Key numbers and a worked example

Let's make the +38 μs/day concrete. A microsecond is 10⁻⁶ s; light travels c ≈ 299,792,458 m/s, so one microsecond of timing error corresponds to about 300 meters of range error.

• Daily timing drift: 38.6 μs/day.
• Range error per day: 38.6 μs × c ≈ 11.6 km/day (commonly quoted as ~11 km/day).
• Per hour that's roughly 1,600 ns, translating to nearly half a kilometer of positional error building up every hour if uncorrected.

The engineering fix is elegant: before launch, each satellite's cesium or rubidium clock is deliberately detuned. The nominal frequency is 10.23 MHz; the onboard oscillator is set to 10.22999999543 MHz — about 4.47 parts in 10¹⁰ low — so that once it is up in weaker gravity and moving at orbital speed, relativity speeds it back up to exactly 10.23 MHz as seen from the ground. The ground segment also applies a small periodic correction for orbital eccentricity, since a slightly elliptical orbit makes Φ and v vary around each revolution.

How it is observed and where it appears

GPS is the most famous demonstration, but gravitational time dilation was measured long before and has been confirmed across an enormous range of precision:

• Pound–Rebka experiment (1959–1960): Robert Pound and Glen Rebka measured the gravitational redshift of gamma rays climbing just 22.5 m up a Harvard tower — a fractional shift of only 2.5 × 10⁻¹⁵ — using the Mössbauer effect.
• Gravity Probe A (1976): A hydrogen-maser clock launched to 10,000 km on a rocket confirmed the GR prediction to about 70 parts per million — still one of the sharpest tests.
• GNSS constellations: GPS, Europe's Galileo, Russia's GLONASS and China's BeiDou all bake in the correction. Galileo's accidentally elliptical satellites (2014) even served as a bonus redshift test.
• Optical lattice clocks: Today's best clocks (fractional stability ~10⁻¹⁸) resolve the time dilation from raising a clock just 1 centimeter, turning relativity into a practical geodesy tool.

The effect appears anywhere clocks sit at different potentials: mountains vs valleys, the top vs bottom of a skyscraper, and dramatically near neutron stars and black holes.

How it differs from its close cousins

Gravitational time dilation is easy to confuse with several related effects — the distinctions matter:

• vs. special-relativistic (velocity) time dilation: The velocity effect (−7.2 μs/day for GPS) comes from motion through space and always slows a clock; the gravitational effect comes from position in a potential and, for a raised clock, speeds it up. In GPS the gravitational term is ~6× larger and wins.
• vs. gravitational redshift: These are two views of one phenomenon. A photon climbing out of a well loses energy and reddens (frequency drops) precisely because the emitter's clock ran slow relative to the receiver's — redshift is time dilation seen in the frequency domain.
• vs. cosmological redshift: That stretching comes from the expansion of space over cosmic time, not a local potential difference.
• vs. Doppler shift: Doppler is a classical relative-velocity effect; gravitational shift persists even between two mutually stationary clocks.

In strong fields the linear ΔΦ/c² approximation fails and you need the full metric factor √(1 − 2GM/rc²), which diverges at a black hole's event horizon.

Significance, famous cases, and open questions

GPS proves that relativity is not an abstraction: a global infrastructure used by billions — aviation, banking timestamps, power-grid synchronization, smartphones — would fail within hours without the general-relativistic correction. It is often called relativity's most-used practical application.

Beyond navigation, gravitational time dilation is the physics behind some of the most dramatic scenarios in astrophysics. Near a black hole, a clock hovering just outside the event horizon runs arbitrarily slow as seen from afar (the basis of the extreme time distortion dramatized in Interstellar). On a neutron star surface, where 2GM/rc² can reach ~0.3–0.4, clocks run roughly 20–30% slow relative to infinity — a shift observed in the redshifted spectral lines and pulse profiles of these objects.

Open frontiers: Optical clocks now test whether the redshift is truly universal — searching for violations of the Einstein Equivalence Principle that would betray new physics or a variation of fundamental constants. Proposals like a clock network in space and clock comparisons across the geoid aim to test GR to 10⁻¹⁹ and use time dilation to map Earth's gravity field (relativistic geodesy). No violation has yet been found — general relativity keeps passing.

The two relativistic effects on a GPS satellite clock (altitude 20,200 km, orbital speed ~3.87 km/s) relative to a clock at Earth's surface

Effect	Cause	Direction	Magnitude
Gravitational time dilation	Weaker gravity at orbit (higher potential Φ)	Clock runs FAST	+45.8 μs/day
Special-relativistic (velocity) dilation	Satellite moving at ~3.87 km/s	Clock runs SLOW	−7.2 μs/day
Net combined offset	Gravitational wins over velocity	Clock runs FAST	+38.6 μs/day
Fractional frequency shift	Fractional rate difference dτ/dt − 1	Positive	≈ +4.46 × 10⁻¹⁰
Positioning error if ignored	38 μs × c per day	Accumulates	≈ 11.4 km/day
Onboard clock pre-set frequency	Factory offset to compensate	Detuned below 10.23 MHz	10.22999999543 MHz