Announcement

Collapse
No announcement yet.

Altitude Accuracy

Collapse
X
  • Time
  • Show
Clear All
new posts

  • Altitude Accuracy

    I have a question that extends beyond the EE unit to the entire Edge GPS family. Since mapping contains data in 3D space which includes altitude for a given Lat/Long based on surveyed actual measurements why is this data not used to give an altitude instead of using the inaccurate GPS method or barometric readings (which is not even available on the EEunit)?

  • #2
    Some Edge models have a barometric altimeter. Would you prefer using altitude data from the map in contrary to data from the barometric altimeter? If no barometric altimater is present your question feels more relevant to me.
    I have understood that the mesh with heigts data in a map is rather coarse. The height value for a given point must be calculated then by interpolation from height data of nearby points and I don't know if the results are accurate enough? Ever found a comparative study? Would be interesting.
    Edge 800 HR+CD / Edge Touring Plus HR / Edge Explore HR

    Comment


    • #3
      There is an option for tracks uploaded to Garmin Connect to have altitude correction. By default, this correction is automatically applied for tracks recorded on a device without a barometer. A couple of days ago I went on a 15 mile circuit with both my Edge 520 (which has a barometer) and the EE (which doesn't). In Garmin Connect the Edge 520 reported 270m gained and 270m lost but, if I enable elevation corrections then the values change to 224m gained /223m lost. In contrast, the EE without altitude correction reported 185m gained / 179m lost and 233m gained / 233m lost when altitude correction was enabled. It is usually expected that GPS measurements will over-report the height gained / lost due to innaccuracy in measuring height (see http://regex.info/blog/2015-05-09/2568 for example) but, in my observations, the GPS measurement is giving the lowest values for height gained/lost. Inspection of tthe height profile for the EE suggests that Garmin have applied agressive smoothing for the GPS height data which has over-compensated for the height measurement inaccuracies. I read somewhere that Garmin uses the Shuttle Radar digital elevation model which is an approximately 90m grid of smoothed data measured on a 30m grid so the correction in Garmin Connect will give lower values than the barometer. There's also the minor matter of the 10m difference in the corrected values for the two devices. This might be caused by the Edge 520 having a less accurace GPS (as shown by how well the track follows the roads I used).

      Comment


      • #4
        Originally posted by BrandonS View Post
        ... why is this data not used to give an altitude instead of using the inaccurate GPS method or barometric readings (which is not even available on the EEunit)?
        Cyclists are generally more interested in elevation gain rather than knowing the absolute altitude. The elevation data in the maps might be too far apart to get an accurate measurement of gain.

        Comment


        • #5
          Maps do not always have elevation data. For example, the map on my Edge 800 does not appear to have elevaton data. I went to the maps page and touched the screen at a location where I know the elevation. The latitude and longitude data are displayed, but no elevation data.

          The Edge Explore does have a DEM file. When I touched a screen at a location where I know the elevation, the box displays latitude, longitude, and elevation. This works with the GPS turned off. So I assume the elevation data is not from the GPS.

          A few days ago, I mounted bothe the Explore and the 800 on my bike and took a 40 minute ride. I compared the elevation gains reported from both units and the corrections applied by Garmin Connect and Strava. Here are the results in feet of elevation gain.
          Device Garmin Connect Correction Strava Correction
          Edge Explore 190 448 480
          Edge 800 623 454 477

          I'm not sure what to make of this. I'll try this again.


          Comment


          • #6
            Originally posted by asmcdaniel View Post
            Maps do not always have elevation data. For example, the map on my Edge 800 does not appear to have elevaton data. I went to the maps page and touched the screen at a location where I know the elevation. The latitude and longitude data are displayed, but no elevation data.
            I don't think the 800 wouldn't be able to use elevation data even if the map included it.

            The newer units have elevation data in a separate DEM map. The elevation data doesn't really change. So, keeping it in a separate map means the maps that do get updated frequently are smaller.

            Originally posted by asmcdaniel View Post
            I'm not sure what to make of this. I'll try this again.
            Garmin provides the correction for units that don't have barometers. That's essentially an admission that GPS-only elevation data is less accurate than DEM data and that DEM data is less accurate than barometric data. (DEM = digital elevation model.)

            The corrected numbers for the two devices (Garmin or Strava) should be very close. There's pretty much one database of DEM data. So, the corrected numbers from Garmin and Strava should be close. The numbers are fairly close (as expected).

            The 190 number is from GPS-only (which Garmin treats as the least accurate).

            The DEM numbers (the corrected numbers) might miss short gains that the barometer sees.

            In any case, 600 feet is probably close to the noise level (so it's probably hard to get an accurate number anyway).

            People seem to expect that the gain numbers will be as accurate as distance numbers. That's probably not correct.

            It might make more sense to treat the gain as an approximate number.

            Last edited by dpawlyk; 01-02-2019, 05:34 PM.

            Comment


            • #7
              Originally posted by dpawlyk View Post

              The 190 number is from GPS-only (which Garmin treats as the least accurate).
              The Edge Explore is not GPS-only does have a DEM file. I think that the Explore gets the elevation from that file because it can display the elevation of a point with the GPS reciever turned off.

              I do have statistics from the .FIT files from the Edge 800 and Edge Explore for this ride. I was trying to compare temperature data collected from the two units, so i failed to correct the elevation displayed by the Edge 800 at a known location. That said, the standard deviation of the elevation data in the .FIT file recorded by the Edge 800 was 16.3 ft. and the standard deviation of the elevation data from the Edge Explore was 12.7 ft.

              I plan to make sure both units are wel prepared and try this again.


              Comment


              • #8
                Originally posted by asmcdaniel View Post
                The Edge Explore is not GPS-only does have a DEM file. I think that the Explore gets the elevation from that file because it can display the elevation of a point with the GPS reciever turned off.
                It may not. The unit might not read the DEM file as part of recording the activity. It might only read the DEM file when asked to.

                Since the unit already has GPS data and because it has to deal with the case of not having the DEM data, it's simpler/faster to just use the GPS data (just like all the other units). Keep in mind that the devices have to minimize what they do when recording an activity to conserve power.

                There's a reasonable chance that the unit has the same DEM data that is used for the correction. If that is the case, the number would match the corrected numbers.

                Originally posted by asmcdaniel View Post
                I do have statistics from the .FIT files from the Edge 800 and Edge Explore for this ride. I was trying to compare temperature data collected from the two units, so i failed to correct the elevation displayed by the Edge 800 at a known location. That said, the standard deviation of the elevation data in the .FIT file recorded by the Edge 800 was 16.3 ft. and the standard deviation of the elevation data from the Edge Exp inlore was 12.7 ft.
                Is standard deviation meaningful for elevation?

                I suspect the Explore elevation profile is smoother than that for the 800.
                Last edited by dpawlyk; 01-02-2019, 10:05 PM.

                Comment


                • #9
                  Originally posted by dpawlyk View Post
                  ... It might only read the DEM file when asked to.
                  You may be right. When I diabled the GPS receiver, the elevation data field shows three dashes. I would guess that the ddata displayed in the data field is the data recorded to the activity file.

                  Originally posted by dpawlyk View Post
                  ... it's simpler/faster to just use the GPS data (just like all the other units)
                  Do all the other Edge units use GPS for elevation?
                  .
                  Originally posted by dpawlyk View Post
                  Is standard deviation meaningful for elevation?
                  I think so. Observations from two devices measuring the same thing. Can you suggest a better statistic to measure the variability?

                  Comment


                  • #10
                    Originally posted by asmcdaniel View Post
                    Do all the other Edge units use GPS for elevation?
                    I believe they all do. The elevation might be a simple by-product of also computing the surface location.

                    Some will also use the barometer (if there is one).


                    Originally posted by asmcdaniel View Post
                    I think so. Observations from two devices measuring the same thing. Can you suggest a better statistic to measure the variability?
                    What is the standard deviation of? Standard deviations only make sense if an average makes sense.
                    Last edited by dpawlyk; 01-03-2019, 09:36 PM.

                    Comment


                    • #11
                      Originally posted by dpawlyk View Post

                      I believe they all do. The elevation might be a simple by-product of also computing the surface location.

                      Some will also use the barometer (if there is one).
                      I put my Edge Explore and Edge 800 side by side on a railing and collected data for about an hour. The altimeter was set to a known elevation and i gave both units a few minutes to get a good satelite fix.

                      After looking at the data, it's obvious that the Edge 800 is using barometric pressure to calculate altitude and the Edge Explore is using the GPS data.


                      Originally posted by dpawlyk View Post
                      What is the standard deviation of? Standard deviations only make sense if an average makes sense.
                      I think the second set of data I collected is a better test because the units were stationary. Any variation recorded was drift. For this test the standard deviation of the Edge 800 was 3.80 ft and the standard deviation of the Edge Explore was 7.23 ft.

                      So does anyone know why there is a DEM file on the Edge Explore?

                      Comment


                      • #12
                        Originally posted by asmcdaniel View Post
                        I think the second set of data I collected is a better test because the units were stationary. Any variation recorded was drift. For this test the standard deviation of the Edge 800 was 3.80 ft and the standard deviation of the Edge Explore was 7.23 ft.
                        That is a reasonable use of standard deviation. As long as the barometric pressure wasn't changing, I'd expect the barometer variation to be less. It's less but not by a lot.

                        Originally posted by asmcdaniel View Post
                        So does anyone know why there is a DEM file on the Edge Explore?
                        So the unit can provide an estimate of elevation gain for routes. Some of the units will calculate alternative routes. You can use the estimated gain to pick a hillier or flatter route.

                        Comment


                        • #13
                          My best guess about the elevation measurement is that it's highly smoothed GPS data. I went for a ride today with the DEM turned off in the mapping but elevation was still reported and measured. It missed the top of the hill on my circuit and later I also noticed, while on a gentle climb, that the reported elevation was slowly dropping. The result of the smoothing is that the elevation gain / loss is significantly under-reported. Garmin need to refine the algorithm used by the EE so the net result is about right.

                          Comment


                          • #14
                            I asked Garmin support about the altitude measurment on the EE and was told that it uses the GPS data. The DEM file is only used for displaying the elevation profile for a course.

                            Comment


                            • #15
                              Thanks for the confirmation. So the accuracy of the elevation displayed is dependent on the Edge GPS.accuracy.

                              Comment

                              Working...
                              X