Glossary
GPS accuracy
How close a watch gets to the real route. A certified marathon usually records 1-1.5% long, which reads as 26.4-26.6 miles on the wrist.
What it is
How closely the track your watch draws matches the ground you covered. Two separate effects push the recorded distance up, and neither pushes it down.
Ranacher and colleagues showed the first in 2016. Every position carries random error, and adding up noisy points inflates the total.
The second is you. Races are measured on the shortest legal line. You will not run it.
| What you see | Typical | Why |
|---|---|---|
| Marathon reads 42.6-42.9 km | 1-1.5% long | noise plus your line |
| Marathon reads 26.4-26.6 mi | same thing | in miles |
| Tunnel or tall buildings | can be far worse | reflected signal |
| Track intervals | often short | tight turns get cut |
What it is not
It is not the race being long. Certified courses carry a short-course prevention factor, so they are, if anything, a fraction over the nominal distance on the measured line.
It is also not a fault you can fix in settings. Better antennas and dual frequency help, but the effect does not go to zero.
What moves it
- Sky view. Trees, buildings and tunnels all hurt.
- Sampling rate. Smart sampling saves battery and loses fidelity.
- Dual-band reception. Measurably better in cities.
- How you run the tangents. The part that is actually yours.
When to ignore it
Ignore small differences between friends' watches on the same run. A 1% spread is normal and means nothing.
Ignore watch distance for pacing a certified race. Pace off the course markers, and set your target from the clock rather than the wrist.
Sources
- Ranacher P et al. Why GPS makes distances bigger than they are. Int J Geogr Inf Sci 30(2):316-333, 2016 primary
- Gilgen-Ammann R, Schweizer T, Wyss T. Accuracy of distance recordings in eight positioning-enabled sport watches. Sensors 20(19):5493, 2020 primary
- World Athletics Competition Rules, Book C2.1 — course measurement and the short course prevention factor reference