Why two apps disagree on GPX elevation gain

Add the rises one at a time

Start with five illustrative heights in metres, in order: 100, 112, 110, 114 and 108. The rises are 12 m and 4 m, so the raw ascent is 16 m. The falls are 2 m and 6 m, so the descent is 8 m. The walk finishes 8 m higher than it began. The numbers balance because every step appears once on one side of that ledger.

Now suppose the 2 m dip was a brief error in an otherwise rising trace. A method that smooths it out can report 14 m to the high point, then 6 m down. The original five points have not changed. The calculation applied to them has.

Find out where the heights came from

The GPX format can carry heights without explaining whether a device recorded them or a map service added them. Position, point spacing and how an app processes a recorded track can all affect the trace. The same path may come back with a different number of points on a second recording.

When an app offers elevation correction, record whether you applied it. A different elevation source can make broad terrain easier to see but can misplace narrow features, a bridge above a river or a route carved into a steep side slope. A different number is a reason to check the profile and the map, not enough to identify an error by itself.

Leave holes in incomplete data

Open the file in the GPX route profile. The coverage line says how many points carry heights. The view stops drawing through missing values and between recorded segments. That matters when an uphill section goes missing. Joining its ends would assume a slope and miss any intermediate dips. See how to handle a recording break before combining separate sections.

The inspector reports raw rises when every point has an elevation. A small staircase of up and down on a level towpath is a clue to inspect the original file. Read the displayed raw total as a description of the sample, then decide whether to use a map estimate for the plan.

Check the difference at the scale of the decision

A climb estimate of 600 m and another of 720 m differ by 120 m. With Naismith’s default allowance of ten minutes per 100 m, the change is 12 minutes before stops. That may matter if a bus leaves close to your estimated arrival.

Use contours to verify the large climbs that drive that difference. In Trail time enter the more conservative supported estimate, then compare it with the lower one. Leave the time reserve separate so you can still see it on the timetable.

On a short descent with close switchbacks, inspect the point density as well as the heights. Too few points can flatten a dip just as a noisy position can scatter samples across the hillside.