Why is my running distance wrong?

Why do two GPS watches disagree on the same run?

Because every GPS fix carries a few metres of error, and no two devices round that noise the same way. Consumer GPS updates about once per second and lands within roughly three to ten metres in the open, worse under trees or between buildings. Run side by side and each device draws a slightly different wiggly line, so the totals never quite match.

This is expected behavior for any GPS receiver. A receiver does not know exactly where you are. It estimates your position from satellite timing signals, and that estimate jitters from one second to the next even when you run a dead-straight line. Two watches see the same satellites but weigh them differently, smooth them differently, and report distances that can differ by a percent or two over an hour. Neither is lying. They are both guessing, carefully, and guessing is the best any receiver can do.

Why does GPS over-count distance?

Because summing noisy points adds length that was never run. If each fix sits a few metres off the true path, the straight line between two of them is always longer than the real straight line. Add up thousands of those slightly-too-long segments and the total creeps upward. Statisticians call it a position random walk, and it inflates almost every naive GPS distance.

The effect gets worse wherever the signal degrades. Tree canopy, tall buildings, and narrow valleys bounce the satellite signals and scatter your fixes, so the wiggle grows and the phantom distance grows with it. Switchbacks and tight turns add their own error, because the receiver rounds each corner as a cloud of scattered points instead of a clean arc. Stand still at a traffic light with a cheap tracker running and watch the distance tick up while you are not moving. That is position noise, counted as running.

Why does GPS sometimes under-count distance?

Because when the signal drops, the track cuts the corner. Go through a tunnel, drop into a deep valley, or bury the phone in a loaded pack, and the receiver loses its fix. When it recovers, most apps draw a straight line from the last good point to the next one, skipping every metre of the curve you ran between them. On a bending trail that shortcut can erase real distance.

Slow movement causes the opposite of over-counting too. Walk a steep climb slowly enough and your true pace can shrink below the size of the GPS noise itself. Some trackers read that as standing still and stop adding distance, so a grinding uphill hike undercounts even though you covered every step. Between corner-cutting on dropouts and stalling at low speed, an unfiltered app can lose distance as easily as it invents it.

How do you make GPS distance accurate?

You stop trusting raw position alone and fuse several measurements together. Steady Pace blends GPS position with Doppler velocity and the phone's accelerometer through a Kalman filter, rejects fixes that fall outside what physics allows, and integrates distance from filtered velocity instead of summing noisy points. A barometer carries the shape of climbs while GPS corrects its slow drift. The result holds up where naive tracking wanders.

Each piece fixes a specific failure. Doppler velocity comes straight from the satellites' frequency shift and is far steadier than differencing two shaky positions, so integrating it kills most of the random-walk inflation. Outlier rejection throws out the fix that claims you teleported thirty metres sideways. The accelerometer keeps distance moving through short signal dropouts instead of cutting the corner, and it tells the difference between a genuine stop and a slow grind uphill. The barometer measures altitude from air pressure, which does not care about tree cover, so climbs keep their true shape. Put together, this is why Steady Pace stays honest on long, slow, and mountainous runs, the exact conditions that make simpler tracking drift.

How can I get a more accurate run distance?

A few habits help any app. Give the sky a clear view, wait for a solid fix before you start, keep the phone out of the deepest part of a bag when you can, and choose an app that filters properly. The first three reduce the noise going in; the last one decides how well that noise gets cleaned up.

• Let it lock on first. Step outside, give it fifteen or twenty seconds to settle, then start. A run begun on a half-acquired fix starts with a wild scatter of points and inflated early distance.
• Find open sky. The more satellites the receiver can see, the tighter its fixes. Open roads and fields beat forest, canyons, and streets walled in by towers.
• Mind where the phone rides. A pocket, an armband, or a hand keeps the antenna closer to the sky than the bottom of a stuffed pack. It will not fix a tunnel, but it helps in ordinary cover.
• Use an app that does the maths. Filtering, outlier rejection, and velocity-based distance matter more than the hardware. Two apps on the same phone can disagree by a noticeable margin because one cleans the signal and the other adds up dots.

You will never get a phone to match a surveyor's wheel to the metre, and you do not need to. What you want is a distance that is consistent from run to run and close to the true course. Good sky, a clean start, and honest filtering get you most of the way there.