UKC

Hi something I've noticed on returning from a cycling holiday last week is that my Garmin (Edge touring) seems to be reporting height gains over a day's ride way in excess of the Strava reports of anyone else on the ride with me (and, in all honesty, way in excess of what I think I've done !). Any ideas why that might be ?

There appears to be a height correction button on Garmin Express which is set (without it set the height gains are even more !)

In reply to GrahamD:
Are you clicking height gain correction on Strava? That's supposed to be more accurate.

I think the Edge Touring doesn't have a barometric altimeter, so the height is all calculated from GPS, which isn't as accurate. Strava can take it from map data.

EDIT: Unless it's a Touring Plus: https://support.strava.com/hc/en-us/community/posts/208848007-Recognize-bar...

In reply to andy:

Its just an Edge touring, not touring plus. When I look at the recorded ride profile it looks right (altitudes pretty much as I'd expect), so it doesn't look that innacurate in terms of absolut accuracy.

Not being 100% familiar with my way around Starva, where is the Strava height correction ?

In reply to GrahamD:

GPS is very inaccurate on height gain. Don't know why. Perhaps the triangulation works better horizontally compared to vertically. I take all height gains as approximate.

In reply to GrahamD:

If you click underneath the climbing number it gives you the option to "correct elevation".

In reply to andy:

Thanks Andy. On my Dashboard Activity Feed I've got an "Elevation" reading, but clicking the Elevation Icon isn't doing anything not to worry.

In reply to GrahamD:

Why the GPS gets it way out, is due to jitter in the height calculated. GPS is very good at 2D fixes but height can bounce about a bit - and the why of this is a bit umm, complicated. But GPS device takes this as gospel and adds up all the off by 1 or 2metres to the total ascent/descent. Decent mapping applications understand this and correct for it. Usually the best way is to use the NASA shuttle radar height dataset, or the OS Opendata topo set, fix the datum differences and interpolate between the dataset grid points to get the height at the track point. Adding those up gives you a more accurate height gain. It's a lot better, but given you know about the height data resolution you can look at nearby points to see trends and further reduce anomalies. Which presumably the Strava thing will do for you.

In reply to beth:

Thanks Beth, Strava claims to average and cross reference to data bases but clearly still not getting the same results as other users (presumably using barometric) for the same route. Unfortunately I can't find any real controls to play with in Garmin or Strava to play with.

In reply to GrahamD:


Because the GPS height figures are quantised, noisy and the height gain accumulator can only ratchet up.

Take for example a walk/ride along a tow path. The height shows as 65 or 66m, it flicks about because of inaccuracy in the measurement but each time it jumps from 65 to 66 the accumulator used to store height gain adds your 1m 'climb'. Over the course of the walk, despite having gained no height at all you may see a few hundred of these 1m 'gains' summed together into a pretty impressive height gain. Another GPS may take a recorded trace, low pass filter the records to reduce the noise then calculate the gain which would normally give a more realistic result but it does rely on some assumptions about the terrain and or your maximum rate of climb, it'd likely miss a lot of detail were you to fly with it.
jk

In reply to GrahamD:

It might be a Premium feature.

In reply to GrahamD:

My Edge 810 seems to under - report altitude gain, and deducts move when I use Garmin Connect to correct.

For example, recently did the Fred Whitton route in the lakes.
According to sportive website the height gain is 3950m, according to the GPX I plotted it was about 3600m, but I recorded 3176m and this then got corrected to 3102m.

My mate doing the ride with me, and using an Edge 500, recorded 3546m, which seems about right.

I can accept that measurements during the ride will be inaccurate, be they based on atmospheric pressure or GPS data - but I'm not convinced the Elevation Correction feature is using accurate data.

In reply to jkarran:

I can understand how the noisy gps data can lead to a cummulative error if unfiltered. I thought that both Garmin and Strava used the raw data to cross reference to standard terrain data though, so I'm surprised that with that supposed level of sophistication I'm seeing up to 30% difference between mine and 4 other riders (who's readings are clustered within about a 10% range)

In reply to andy:

I'm too tight for 'premium' Strava features

A quick google of Strava didn't throw up any info on whether the parameters were settable or not.

In reply to GrahamD:

Terrain data apparently isn't that good. It's ok where slopes are shallow but in hilly areas (even if you're actually following a flat valley floor or a plateau top) it rapidly gets out of step with reality. I've been putting together a glide computer that has a height over terrain feature using barometric, GPS data plus a terrain map to calculate and. Out of curiosity, to better understand its limitations I've been driving with it. In the vale of York it is never more than +-10ft out but by the time you're out to Harrogate in the hills it's frequently +-100ft out. The horizontal position is spot on with the trace following roads in the mapping (while obviously not being tied/corrected to those roads like a car sat-nav would be) so it must be the resolution of the terrain model or the barometer yet the latter works fine in gently rolling terrain.
jk

In reply to beth:


The problem is to design a correction algorithm that accounts for the coarseness of the SRTM data.

This gives some idea of the complexity of height datasets... http://www.cgiar-csi.org/data/srtm-90m-digital-elevation-database-v4-1