Why does elevation gain look high on a walk that felt completely flat?

The short answer

GPS altitude wanders up and down by a metre or two even when you are standing still on level ground, because vertical position is the least precise thing a phone's GPS measures. Elevation gain is not the difference between your highest and lowest point — it is the sum of every single upward step along the way, and on a flat walk, roughly half of those tiny random wobbles happen to be upward. Add enough of them together over a long, flat, unhurried walk and they stop looking like noise and start looking like a hill that was never there.

Flat to your legs is not flat to a GPS reading

Walking is a good judge of grade because your effort tells you immediately when the ground tips even slightly. A phone has no such sense. It works out altitude from satellite geometry that brackets you only from above — every satellite in view is somewhere overhead, never below the horizon — so height is inherently harder to pin down than east-west or north-south position. The result is a raw altitude reading that jitters by a metre or two from one moment to the next, on ground that is, by any measure that matters to a walker, dead level.

That jitter is symmetric around the true height on average — it is not a slow drift in one direction — but symmetric noise is exactly the ingredient that produces a lopsided total, for a reason that has nothing to do with the phone being wrong.

Why summing the noise inflates the number, even though the noise itself does not favour up or down

Elevation gain is built by walking through a track point by point and adding the difference to a running total every time altitude goes up, while ignoring every time it goes down. That asymmetry is the whole mechanism. A coin that comes up heads exactly half the time still produces a growing pile of heads if you only ever count the heads and throw away every tally of tails — and that is precisely what a naive elevation-gain calculation does with altitude noise. Every metre of downward jitter is thrown away rather than cancelling out the upward jitter that came before or after it, so the running total only ever grows.

A single wobble is invisible. A few hundred of them, one after another over a long flat walk, is not — even though the ground never moved and the average altitude the phone recorded barely changed at all.

Why a longer or slower flat walk makes it worse, not better

Because there are more points to sum. A ten-minute stroll along a level path might only pick up a handful of noisy readings and a few metres of spurious gain. The same flat ground covered over three unhurried hours, with a phone logging altitude many times a minute the whole way, accumulates far more of those small upward steps — not because the walk got any hillier, but because there were simply more opportunities for the random jitter to land on the "up" side of the running total. Distance covered barely matters here; time spent recording on flat ground is what drives the total up.

Why a smoothing threshold helps but does not make the number zero

Every app that reports elevation gain applies some kind of noise filter, usually by ignoring changes smaller than a set threshold — commonly a few metres — before adding anything to the running total. That is why the number on a flat walk is rarely as large as the raw altitude jitter alone would produce. It is also why it is rarely exactly zero: set the threshold too low and genuine noise gets counted as climbing; set it too high and small real rises on a gently rolling path get thrown away along with the noise. There is no threshold that perfectly separates the two, only a compromise, which is part of why elevation gain is the number most worth treating as approximate rather than exact.

When it is worth a second look

• A flat walk shows tens of metres of gain, scaling roughly with how long the recording ran rather than with anything you remember climbing — this is the ordinary pattern described above, not a fault in the recording.
• The elevation profile on the map is a flat, dense scribble rather than a smooth line with occasional real bumps — visually, this is the same noise the running total is built from, just seen directly instead of summed.
• The gain total is genuinely large — hundreds of metres — on ground you are certain never rose at all, which is worth checking against a map of the route rather than assumed away, since it can also point at a GPS altitude fix that briefly jumped, rather than ordinary jitter.
• Two recordings of the same flat loop, one quick and one leisurely, show noticeably different gain totals — expected, for the same reason a longer recording accumulates more of the noise, and a sign to compare walks by distance and moving time rather than by elevation gain when the ground is this level.

Where TrailCam fits in

TrailCam draws elevation gain as a profile under the map for every recorded route, both while you are out and again on replay, and includes it in a GPX 1.1 or CSV export alongside distance, moving time and pace. A flat towpath finishing with a modest, nonzero gain figure is the ordinary texture of GPS altitude described above, not a sign that anything was recorded wrong. The app is free to download.