Incorrect elevation change in profile view

Justin's Avatar

Justin

18 May, 2021 05:55 AM

I am plotting the elevation profile on some line tracks I have used to plan prior routes and the elevation gain reported in the profile view appears to be a fair bit greater than previously.

For example, on one route the elevation gain was previously about 8,000 ft and now it is shown as 10,000 ft. On another, my Strava track estimated an elevation gain of 5,200 ft; however, CalTopo now shows the route as having 6,100 ft of elevation gain. Various websites report the elevation gain at about 5,500 ft.

In the attached example, I previously wrote down that this route has an elevation gain of 5,894 ft; however, CalTopo now reports it at 7,535 ft which cannot possibly be correct given the route starts at 3,800 ft and climbs to 7,600 ft before descending. Is anyone else noticing this? Could someone check this gpx file and tell me what they get?

  1. Support Staff 1 Posted by Julie on 18 May, 2021 03:02 PM

    Julie's Avatar

    HI Justin,

    For digital maps or even recorded tracks, the location of each waypoint is usually compared to a digital elevation model (DEM) to determine the total elevation gain and loss along the line. Two major factors will affect the total: sampling interval and the model itself. The gpx file itself does not contain the elevation data. (While it's possible for a gpx file to carry that information, most companies do not include that information when exporting a track and prefer to use their own DEM to determine it within the program. This is a persistent topic in support forums for digital map/track providers because some do include it in the track and some don't. For the record, we don't.)

    The DEM model itself could be based on different data sources - for example there could be a LIDAR data source built into the model, which would make for a higher degree of real-world accuracy in the model. LIDAR data is expensive and hard to come by for many areas of the country/world though, so it's not always included. Another factor here is that the accuracy of a recorded gpx track may not match the accuracy level of the underlying source of elevation data (it's possible that it is much less accurate.)

    Then you factor in the sampling interval when the elevation for the line is being calculated. This usually depends on the length of the line and how many waypoints are included in the line. The program doing the calculating may sample at every point, or every 50 or 100ft, or some other number that seems random but is based on various factors that I won't get into. The point is, the reported elevation change over the line can and will change depending on the sampling interval. Sampling at every point can introduce a lot of noise and make the total artificially high, while sampling too little could theoretically make it way too low. In the CalTopo terrain stats, you can resample the line at various intervals and see how the total elevation changes. Some info about terrain stats is at https://training.caltopo.com/all_users/objects/existing-lines#terrain

    In your track specifically, the total is higher than the absolute number between low point and high point because you go over some smaller ups-and-downs along the route, each of which contributes to the total reported gain.

    Thanks,
    Julie

  2. 2 Posted by Justin on 18 May, 2021 08:03 PM

    Justin's Avatar

    Thanks Julie. Everything in your response makes sense and it is helpful to hear a little about the internal calculations behind this number. I think the main component of my question that was not answered is if there have been any changes to the way elevation profiles are calculated recently that could account for the changes I am seeing? For example, a change in the default sampling interval or terrain model?

    The track I uploaded previously as an example is one I created in CalTopo by snapping to a trail network and I did not manually modify the sampling interval or anything like that when recalculating the profile. That the reported elevation change in the profile has seemingly increased by 20% in some cases seems suspicious, especially since I have literally not changed anything (same track, same map, clicking "Profile" as before) and getting a substantially different calculation! And I would note for another route I am looking at, CalTopo is now reporting about 2000 ft more elevation gain than other published sources like AllTrails and trip reports, etc. Even if I change the sampling interval to 200 ft instead of the default 100 ft, I am still overestimating by 1000 ft.

  3. 3 Posted by Tom on 19 May, 2021 04:33 PM

    Tom's Avatar

    Justin,
    I'm a fellow CalTopo user.

    Adding to the response above, here are some thoughts from a regular old CalTopo user.

    When using CalTopo to help estimate elevation gain and loss of a recorded GPS track and/or a "drawn" route, I find that the "optimal" sampling interval varies depending on the track or route, and "depends on" things that include:
    - the "precision" of the waypoints on the track or route (ex. how close is the lat long of each waypoint on the track to your actual location when the waypoint was recorded)

    - the "resolution" of the digital elevation model(DEM), and whether the underlying source of the DEM is newer, modern LIDAR data, or older, less resolute data

    ---

    When I am curious about the elevation gain and loss estimate Caltopo provides for a track or route, I find it "interesting" to:
    - Play around with different sampling intervals (ex. 50', 100', 200')

    - From the CalTopop Terrain Stats " page, click on "CSV", to create a file that includes each waypoint being used to estimate the elevation gain and loss, with data for each waypoint that includes:: Lat, Lng, Distance (feet), Elevation (feet).

    I then open this file from a spreadsheet and columns that include:
    - Distance between each waypoint (feet)
    - Elevation Change between each waypoint (feet)

    I then review the "Elevation Changes" between each waypoint, to better understand the total elevation gain and loss estimate CalTopop provides.

    If I see lots of "small" ups and downs in the "Elevation Changes" column I create, I start to "wonder" whether those small "ups and downs" reflect the actual route, or if they are "noise" due to less than 100% precise lat-long estimates, and/or less than 100% precise elevation estimates.

  4. Support Staff 4 Posted by Meghan on 20 May, 2021 01:48 PM

    Meghan's Avatar

    Justin,

    You are correct that how CalTopo calculates elevation profiles has changed recently. Previously we were sampling at 300 points along the line and we recently switched to using 100' intervals by default in order to have more consistent profile numbers, regardless of the length of the line. My guess is that AllTrails is probably using fewer sample points like we used to in order to calculate elevation gain although I don't know for sure how they perform those types of calculations.

    But regardless keep in mind that the raw number isn't as important as comparing to other trips that were computed in the same way.

    Hope that helps shed some light on your question!
    Best,
    Meghan

  5. 5 Posted by Justin on 21 May, 2021 06:09 PM

    Justin's Avatar

    Thank you all for the responses and insights. Very helpful. For the route I was most recently looking at, the distance is about 30 miles, so 300 sample points total works out to one point per 528 ft. Compare that to the new default of 100 ft sample spacing or even the max selectable of 200 ft per sample and that is quite a bit different! I appreciate the suggestion about comparing to other routes computed in the same way as a way to benchmark elevation gain relative to other routes you have completed.

  6. 6 Posted by Tom on 08 Jun, 2021 02:12 PM

    Tom's Avatar

    Meghan, CalTopo: Please, can you explain and elaborate about why CalTopo changed its default method for estimating elevation change using 300 points, to using 1 point every 100'? Thank you!

  7. Support Staff 7 Posted by Julie on 11 Jun, 2021 03:49 PM

    Julie's Avatar

    The change in calculation makes it more consistent across all lines, so that the level of accuracy for each line does not vary as much. The variation in accuracy across different lines lengths has been one source of confusion for users.

  8. Julie closed this discussion on 11 Jun, 2021 03:49 PM.

Comments are currently closed for this discussion. You can start a new one.

Keyboard shortcuts

Generic

? Show this help
ESC Blurs the current field

Comment Form

r Focus the comment reply box
^ + ↩ Submit the comment

You can use Command ⌘ instead of Control ^ on Mac

 

13 Jun, 2021 10:00 AM
12 Jun, 2021 01:42 AM
12 Jun, 2021 12:14 AM
11 Jun, 2021 10:25 PM
11 Jun, 2021 09:54 PM
11 Jun, 2021 09:36 PM