T qupkene

qupkene / Tech / Altitude measurement

Tech

Altitude measurement

One device measures pressure, the other measures geometry. They are answering different questions.

Two devices on the same summit will report different heights, and people reasonably conclude one of them is broken. Usually neither is.

Barometric altimeters

A barometric altimeter measures air pressure and converts it to a height using a model of how pressure falls with altitude in a standard atmosphere. The real atmosphere is not standard, and its pressure changes with the weather. A front passing through can shift the indicated altitude by fifty to a hundred metres over a day without you moving at all. Calibrate at a known height, at a trailhead sign or a lake, and recalibrate if the weather turns.

GNSS altitude

Satellite positioning gives height geometrically, with no weather dependence, but vertical accuracy is typically two to three times worse than horizontal because of the satellite geometry: they are all above you, none below. There is also a reference problem, since the raw solution is relative to a mathematical ellipsoid rather than to sea level, and the correction between the two varies by location. Most devices apply it silently.

Which to trust

For instantaneous height in stable weather with a fresh calibration, barometric is usually better. For height after several hours, or in changeable weather, GNSS drifts less. Many watches blend the two, using GNSS to slowly correct barometric drift, which is generally the right answer and occasionally produces a visible jump when the correction applies.

Total ascent is the worst number

Cumulative ascent over a day is far less reliable than either instantaneous reading, because it accumulates noise. A barometric sensor twitching by a metre while you stand still adds metres of phantom climb. This is why the same walk gives different totals on different devices, and why figures on route pages, including the ones on this site, should be read as approximate. Post-processing a track against a terrain model usually gives a more consistent number than any device does live.