When searching of gps routes in wikiloc, for the trailrunning race "Transvulcania" I have several routes to look. I downloaded a lot, but every route has a distance between 71 kilometers and 74 kilometers, of course someone in the race could be lost for a while, so that is not a big problem for now, but when I see the elevation gain, is very different between them. For example the route in wikiloc says: 74.460m +4.171m -3.874m The same route in eMapzone 71.923m +6.495m -6.132m

Another one wikiloc 72.100m +4.560m -4.210m eMapzone 72.045m +5.638m -5.286m

and the worst is that the official distance of the race is: 73.000m and +4.300m.

So I decide to normalize the elevation of the route with a Digital Elevation Model, I upload the original file to http://www.gpsvisualizer.com/elevation and then I look the information with eMapzone, and there is some consistency in the data.

The elevation DEM calculated of the same route with several and diferents gps hardware is something in between +5354 , +5598 and -5015, 5263.

So my question is? is better to take the elevation of a Digital Elevation Model?

  • @ChrisW You are completely correct, and I'm not sure how I misread it so completely. Original comment deleted.
    – mkennedy
    Jun 28 '15 at 0:29
  • @mkennedy No worries, I had to read it a few times to be sure myself and thought I might have missed something. The link is a good one and to preserve it I've incorporated it into my answer. I do think the formatting of the page (headings) and some of their wording does lead to some confusion - I usually see it as 'consumer' not 'commercial' (which to me implies at least differential).
    – Chris W
    Jun 28 '15 at 5:43

Short answer, use the DEM.

Handheld GPS units of the type you mention (consumer grade), can potentially use two different methods to figure elevation - the calculated GPS position or an internal barometric altimeter. Because of the way GPS works (explained at the page mkennedy originally linked to in a comment, and elsewhere), without high-grade, corrected systems, the barometric method is typically more accurate than GPS. The next step up from Consumer is Differential, which according to that page has a vertical accuracy of 2-3x horizontal, and above that is Survey Grade which can get down to mm level depending on mode and range factors. Differential and Survey do use GPS calculation rather than barometric.

However the barometric method still has issues - readings can be affected by weather as the air pressure and temperature changes. You're unlikely to get consistently detailed or accurate elevation readings. I use a Garmin 60CSx and could provide multiple GPS track examples of this; I have tracks that are loops yet show a 200ft vertical difference between the start and end points even though they're the same place, as well as multiple versions of the same track with completely different elevations throughout. Some documentation on their Edge units puts vertical accuracy with calibrated barometric readings at +/-50 feet.

A couple of related links from Garmin's FAQ:

And there are many other places on the web you can find information/discussions about GPS calculated vs barometric altimeter accuracy for elevation readings, particularly related to handheld consumer units. Bottom line, every unit will return different values, and even the same unit will on different trips.

On the other hand, the DEM is basically a mapped value you're comparing the route against. You will have inaccuracies based on both the horizontal and vertical resolution of the DEM (so higher res is better), but it will be consistent from track to track (assuming said tracks are within a given horizontal tolerance of course). A DEM or map with contours or any other such data source is basically fixed - the values are what they were when measured/interpolated. But each GPS track is going to be whatever the unit recorded at whatever time, so it will differ from the map value at the same point. This is readily apparent on GPS units which are capable of displaying map data; you can see the elevation the map says you are at along with what the altimeter and potentially even GPS calculated elevation values are (and all three will probably differ).

  • 1
    If you use a DEM, I suggest to take something better than SRTM data. Especially in mountain areas, SRTM has artefacts, and in city and forest areas you can not be sure if the data is top of building/tree or ground height.
    – AndreJ
    Jun 28 '15 at 5:55
  • @AndreJ Agreed, SRTM even at 30m is very low resolution and will have significant steps from cell to cell, even more pronounced in mountainous areas. Plus all the other artifacts and bare ground vs tree height issues. If a suitable/high resolution DEM can't be found or generated (could always resample to a higher res understanding the implications of that), your next best bet would be to average or examine all the track numbers. Possibly throw out the outliers and average the remaining.
    – Chris W
    Jun 28 '15 at 6:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.