How GPS Estimates Distance on a Running Watch

A GPS-enabled running watch is a wrist-worn receiver that estimates its own position on Earth by communicating with a constellation of orbiting satellites. The distance figure it reports after a run is not a tape-measure reading; it is a computed value assembled from a series of position fixes taken at intervals throughout the activity.

This piece covers the mechanics of how that estimate is built — from the moment a satellite signal reaches the antenna inside the watch to the final distance value stored in a completed session. The same principles apply whether the watch is a slim fitness band with GPS or a full-featured multisport device.

From Satellite Signal to Distance Figure: The Estimation Chain

Every GPS satellite broadcasts a continuous radio signal that carries two pieces of information: the satellite's precise orbital position and the exact time the signal was transmitted. The watch's receiver records the moment each signal arrives. Because radio waves travel at the speed of light, the difference between transmission time and arrival time — multiplied by that speed — gives the distance between the satellite and the watch. This single measurement is called a pseudorange, and it contains a small timing error because the watch's clock is far less precise than the atomic clocks aboard the satellites.

To resolve that timing error and produce a usable position, the receiver must hear from at least four satellites simultaneously. With four pseudoranges in hand, the firmware solves a set of simultaneous equations for four unknowns: latitude, longitude, altitude, and clock offset. The result is a single position fix — a coordinate pair with an associated accuracy estimate. Modern watches typically record a new fix every one to five seconds, depending on the GPS mode selected.

Distance is then computed by stringing those fixes together. The watch calculates the straight-line distance between consecutive coordinate pairs using spherical geometry — specifically a formula that accounts for Earth's curvature — and adds each small segment to a running total. At the end of an activity, the sum of all those segments is the reported distance. The watch never directly measures the ground underfoot; it measures the geometry between itself and objects in orbit, then infers ground distance from the resulting trail of coordinates.

Many running watches supplement raw GPS with an onboard accelerometer. The accelerometer detects wrist motion and estimates cadence and stride length, producing a dead-reckoning distance estimate during moments when satellite signals are blocked — inside a tunnel, beneath a dense canopy, or near tall buildings. The firmware blends the two streams, weighting GPS more heavily when signal quality is high and falling back to the accelerometer-derived estimate when it is not. This sensor fusion is why the distance figure does not simply freeze or drop to zero when GPS briefly loses lock.

Components Inside the Watch That Shape the Estimate

GPS antenna. A small patch or chip antenna embedded in the watch body receives L-band radio signals from satellites. Antenna size and placement affect sensitivity; a larger antenna with a clear view of the sky acquires signals faster and maintains lock more reliably than a compact antenna pressed against the wrist. Signal reception is passive — the watch transmits nothing.

GPS chipset. A dedicated low-power chip handles the computationally intensive work of decoding satellite signals, solving the position equations, and outputting coordinate fixes. The chipset also manages which satellite constellations the watch listens to. Beyond the original United States GPS constellation, many modern receivers can also process signals from the Russian GLONASS system, the European Galileo system, and others. Listening to more constellations increases the number of visible satellites at any moment, which generally improves fix accuracy and reduces the time to first fix after the watch is powered on.

Accelerometer. A micro-electromechanical (MEMS) accelerometer measures acceleration along multiple axes many times per second. During a run, the watch firmware interprets the rhythmic wrist-motion pattern as steps and derives a stride-based distance estimate. This is the same type of sensor that counts steps in a basic fitness tracker, though the algorithms applied in a GPS watch are calibrated specifically to running gait rather than general movement. Just as a mechanical specification only describes conditions at the time of testing, an accelerometer-derived distance estimate is only as accurate as the motion model it was built on.

Barometric altimeter. Many running watches include a pressure sensor that continuously measures air pressure to infer elevation change. Altitude data does not directly affect horizontal distance, but it informs the three-dimensional path length calculation and improves the accuracy of the position fix by constraining one of the four unknowns.

Onboard firmware and filtering algorithms. Raw GPS coordinates contain noise — small random errors that cause the plotted track to jitter slightly even when the watch is stationary. Firmware applies smoothing filters (commonly Kalman filters) to reduce that jitter. The filter blends incoming GPS fixes with predictions from the accelerometer and previous position history to produce a cleaner coordinate trail. The aggressiveness of this filtering is a design choice that trades responsiveness to real direction changes against resistance to noise spikes.

Where GPS Distance Estimates Break Down

Multipath interference. In urban environments, satellite signals reflect off building facades and reach the antenna after bouncing one or more times. The receiver cannot always distinguish a reflected signal from a direct one, so the calculated pseudorange is longer than the true distance to the satellite. The resulting position error can displace a fix by several meters or more, and a series of displaced fixes produces a distance reading that does not match the actual path taken.

Canopy and canyon effects. Dense tree cover absorbs and scatters L-band signals. Deep urban canyons block large portions of the sky, reducing the number of visible satellites below the four needed for a reliable fix. When fewer than four satellites are available, the firmware either holds the last known position, uses dead reckoning from the accelerometer, or interpolates — all of which introduce error into the cumulative distance total.

Fix-rate and straight-line approximation. Because the watch samples position at intervals rather than continuously, it draws a straight line between each consecutive pair of fixes. On a straight road this approximation is negligible. On a tight curve or a switchback trail, the chord between two fixes is measurably shorter than the arc actually traveled. A lower fix rate (one fix every five seconds versus one per second) amplifies this undercount on curved routes.

Satellite geometry (DOP). Even with four or more satellites visible, their geometric arrangement in the sky affects fix quality. When satellites are clustered in one part of the sky rather than spread across it, the intersection of their pseudorange spheres is elongated, and position uncertainty increases. This is expressed as Dilution of Precision (DOP) — a dimensionless number that scales the underlying measurement noise. A high DOP value means the same signal errors produce a larger position error on the ground.

Accelerometer calibration drift. The stride-based dead-reckoning estimate that fills GPS gaps depends on a motion model calibrated for a typical running gait. Unusual terrain, fatigue-altered gait, or carrying the watch in a hand rather than on the wrist can all shift the actual motion pattern away from the model, causing the accelerometer distance to diverge from true distance. This is a fundamentally different error source from GPS geometry, and the two can compound each other during a single session.

What the Reported Distance Specification Does and Does Not Capture

A watch specification sheet may list GPS accuracy as something like "±5 meters CEP" (Circular Error Probable). CEP means that in ideal open-sky conditions, 50 percent of individual position fixes fall within that radius of the true location. The other 50 percent fall outside it, and the specification says nothing about the distribution of those outlying fixes. It is also a single-fix accuracy figure, not a cumulative distance accuracy figure — the error in a reported 10-kilometer distance is not simply ±5 meters, because each fix in the chain carries its own independent error that may add or partially cancel with adjacent fixes over the course of a session.

Manufacturer accuracy claims are typically measured under controlled open-sky conditions with strong satellite geometry. Real-world sessions in urban environments, forested trails, or mountainous terrain routinely produce larger errors. The specification does not describe worst-case performance, and it does not account for the firmware filtering choices that vary between devices and firmware versions.

The distance figure stored in a completed session also does not reflect elevation-adjusted path length unless the watch specifically offers a "3D distance" mode. Standard GPS distance is the sum of horizontal segments projected onto a flat plane. A hilly route will show a shorter distance than the actual ground covered, because the vertical component of each segment is discarded. This is a systematic undercount on any route with significant elevation change, not a random error, and it does not cancel out over time.

Understanding what a GPS specification actually measures is similar in spirit to understanding what any other equipment rating captures at the moment of its test. Just as the mechanics of a treadmill belt and motor describe performance under controlled lab conditions, GPS accuracy ratings reflect a specific test environment that may differ considerably from the conditions of ordinary use.

A GPS running watch is a compact radio receiver solving a geometry problem dozens of times per minute, and the distance it reports is the accumulated result of that ongoing calculation — shaped at every step by satellite visibility, antenna placement, firmware filtering, and the physical environment the watch moves through. The number on the screen is an estimate with a known class of limitations, not a measurement of the ground itself.

Note: This explains how fitness equipment and sensors work. It is not a workout program, it is not personal training, and it is not a substitute for a trainer or physician. Check the cited sources for current guidance.