Differing elevation gain/loss in Gaia vs TrainingPeaks using same GPX file

Answered

Alp Runner

25 July 2020 15:57

I often record my routes via the Gaia GPS app on my phone; then take that GPX file and upload it to my TrainingPeaks account. There is often a discrepancy, especially in the total elevation gain/loss, between the Gaia GPS stats and TrainingPeaks stats (even though it is the same GPX file). Using the same GPX file, the elevation gain/loss totals on the TrainingPeaks dashboard are much higher than on Gaia GPS.

Here are some examples where the same route using the same GPX file end up showing different stats in TrainingPeaks vs. Gaia GPS.

EXAMPLE 1
Distance: 10.2 mi (TrainingPeaks) vs. 9.7 mi (Gaia GPS)
Elevation gain/loss: 4,190 ft (TrainingPeaks) vs. 3,400 ft (Gaia GPS)
(Route involved some Class 3 scrambling across a ridge with minimal elevation differential.)

EXAMPLE 2
Distance: 22.8 mi (TrainingPeaks) vs. 22.7 mi (Gaia GPS)
Elevation gain/loss: 8,300 ft (TrainingPeaks) vs. 5,800 ft (Gaia GPS)
(Route involved multiple summits of mostly Class 2 with only a few sections of Class 3 scrambling.)

EXAMPLE 3
Distance: 20.8 mi (TrainingPeaks) vs. 20.3 mi (Gaia GPS)
Elevation gain/loss: 9,700 ft (TrainingPeaks) vs. 8,000 ft (Gaia GPS)
(Route involved substantial Class 3 scrambling up/down two substantial peaks.)

Can someone help me understand how these different apps are using the same data (same GPX file) to arrive at different elevation gain/loss results? Which one is more accurate (or is it somewhere in between)? Does the type of terrain impact how the raw data is rendered in these different apps? Thanks for any insights you can provide!

0

• 

  Chris Hillbee

  • Community moderator

  28 July 2020 23:43 Edited

  Hi there,
  
  This is likely a difference in how we and TrainingPeaks filter elevation data.
  
  An elevation algorithm can spit out vastly different results based on how often you sample the points. For a mountainous track that has continuously undulating ups and downs, it's possible to get widely different elevations calculated, depending on how frequently you want to sample the points. It will be much higher if you include every tiny up and down. If you under-sample - peaks and valleys may get cut off and the elevation will be too low.
  
  We think the trick is finding the middle ground between what people expect to see, and what is most accurate. In my opinion, I think we've done a really good job of this.
  
  Also, there's always a chance some GPS noise was introduced in your track recording. If you suspect this is the case, you can append '&elevationlookup' to the url of your track to pull elevation data from our servers instead of the GPX.
  
  Example:

  • https://www.gaiagps.com/datasummary/track/2aa7a214-dbf7-4bb3-a02f-3f860f245bb8/?layer=GaiaTopoRasterFeet
  • vs
  • https://www.gaiagps.com/datasummary/track/2aa7a214-dbf7-4bb3-a02f-3f860f245bb8/?layer=GaiaTopoRasterFeet&elevationlookup

  1
• 

  Ned_F

  03 August 2020 01:53

  Thanks for the explanation on the total ascent discrepancies. Is this also true for distance discrepancies?

  0
• 

  Jim

  06 August 2020 15:21

  @Ned_F
  
  This is also true for distance, but for slightly different reasons. When calculating distance, the app is fed all the raw location data gathered by the GPS chip on your device. From there, the algorithm filters out inaccurate points, applies a smoothing filter, and then averages the GPS points recorded. It is likely that other apps use a similar but different algorithm, so this might be part of the reason for discrepancies in distances.
  
  This article discusses some other reasons why you may see a discrepancy in the distance of recorded tracks: Why GPS Track Recording Can Be Inaccurate

  0

Please sign in to leave a comment.