How accurate is Strava in measuring height gain?

[Globalti]

Last night's hooligan ride produced widely varying measures of height gain:

Rider A using Strava: 3337 feet

Rider B using Strava: 3048 feet

Self using Bikehike after the ride: 2914 feet.

I find Bikehike accurate on distance measuring; my Cateye measured 37.82 miles and Bikehike gave me 37.35 miles so I've every reason to think Bikehike is pretty accurate on height as well.

Anybody got any thoughts on this? I bet @ColinJ has something to say..

 

[HLaB]

Not very. Discrepancies are due to the recording units barometer or not. Reseting the Strava elevation should give the same but I've never tried it.

 

[welsh dragon]

I don't think any of the aps are accurate for anything in my opinion. Strava endomondo, mapmyride are all innacurate regarding distance travelled let alone anything else. They can be out by as much as 4 miles, and the further you go, the less accurate they are.

I only use endomondo as a rough estimate and nothing else. I'm not bothered about pinpoint accuracy anyway, but take them all with a pinch of salt.

 

[GrumpyGregry]

Very not very accurate.

 

[fossyant]

Don't worry about it. Take it with a pinch of salt

 

[Colin_P]

I cycled to my GP's the other day and was in there about 45 minutes, paused the Garmin.

In that time I had somehow climbed about 100ft. A weather front and change in pressure obviously wafted in.

I take the figures with a pinch of salt, including distance.

 

[ColinJ]

Last night's hooligan ride produced widely varying measures of height gain:

Rider A using Strava: 3337 feet

Rider B using Strava: 3048 feet

Self using Bikehike after the ride: 2914 feet.

I find Bikehike accurate on distance measuring; my Cateye measured 37.82 miles and Bikehike gave me 37.35 miles so I've every reason to think Bikehike is pretty accurate on height as well.

Anybody got any thoughts on this? I bet @ColinJ has something to say..

I think that most of the possibilities have already been covered in other posts but I'll add my opinions ...

You would expect 2 riders doing the same route to get the same numbers from Strava. Usually, GPS is good at tracking position but not quite so good at tracking elevation so I wonder if Strava is looking at the tracklogs for elevation data or using the positions to look up elevations elsewhere? Ah - I just found THIS - sometimes one way, sometimes the other! (Strava may work differently now - that was 5 years ago.)

I have been plotting my routes using Memory Map for 11 years now and the elevation gain figure that it comes up with is always about 20% higher than I expect looking at the elevation profiles. It takes into account every little undulation in the road so every contour line that gets crossed gets counted. When I think of climbing I am thinking of actual hills/mountains, not small elevation variations that hardly register when I am riding them. I am more interested in the bigger climbs so I knock the 20% off to come up with a figure for a route, but MM is showing a more accurate figure.

I now ride with a Garmin Edge 500 on my bike which shows accumulated elevation gain for each ride. When it is behaving itself it comes up with numbers similar to MM but sometimes it loses track of what elevation I am actually at. I have programmed the device to recognise where I start my home rides from so it calibrates its altimeter before I set off, but I have returned home from rides and found that it thinks my house is then up to 60 metres higher or lower than when I left! Obviously those inaccuracies can add up if they occur a few times during a ride. DESPITE the GPS having a barometric altimeter, it is actually less accurate then the non-barometric altimeter on my ancient Garmin Etrex which I run alongside the newer device! (I check them at local summits which I know the elevations of.)

TBH, I am not too bothered about the exact amount of climbing on a route. If I am planning a ride for other people then I will give an indication of how hilly the route is but often just say that it is extremely hilly, hilly, undulating or flattish. If pushed, I would come up with a number which would be about 90% of what Memory Map tells me that it is.

 

[jefmcg]

[QUOTE 4937997, member: 9609"]the hillier the ride the more accurate - flat roads suffer badly from noise (esp with gps/barmeteric data) , up a metre down a metre up a metre down a metre (it all ads up) when in truth it was just flat. If you are on a hill that noise is eliminated as you are going either up or down.[/QUOTE]
That does not tally with my experience. It's very flat where I live, and I can do a 20km ride with 0 elevation. No noise at all.

 

[Sea of vapours]

Her'es another entertaining and often very significant error which can creep in when allowing Strava (et al) to calculate elevation based purely on track (that's a recorded track, as distinct from puttnig a route into some software).

On proper hairpins, the variety found in the Alps and Pyrenees, there is often a drop-off at the edge of the road; a steep drop-off. So, when you go round a hairpin then, depending on the recording rate you've set in the gps device, the actual track may go across the apex of the hairpin, over the drop-off. When Stava or whatever then calculates the acent/descent the algorithm does not seem to disregard this data, so you appear to suddenly drop 20-30m and then gain the same a couple of seconds later. This, needless to say, can produce vastly exaggerated elevation figures on roads with lots of hairpins and steep drop-offs by the side of the road. I did a day up a climb in the Pyrenees last year which my barometric device recorded as a little under 3,000m ascent. As an experiement, I uploaded it to Strava minus the barometric data and the result was, iirc, around 4,000m. Not an insignificant error (and technically correct based purely on the track, but not meaningful in any way)!

 

[Racing roadkill]

GLONASS is a sack of poo. The wheel sensor based systems are better for distance recording, IMO. The altimeters are universally flakey for relatively small elevation changes ( relative to aircraft elevations between ground and flight altitudes for example).

 

[Globalti]

Jefmcg on a bike ride, yesterday:

 

[vickster]

Somewhat unlikely that it's her

 

[jefmcg]

Somewhat unlikely that it's her

I'm in the tag-a-long

[QUOTE 4938214, member: 9609"]they must have improved enormously since I used them, is that GPS or barometric pressure ?[/QUOTE]

When did you stop using it? 2011? I've been getting consistent elevation on flatland since I first started using Strava. GPS only (I did have a phone with a barometer, but strava did not use that data)

[QUOTE 4938214, member: 9609"]and where are you that is so flat. ? even a competitions standards snooker table could have up 5metres of rise and fall in 20km ![/QUOTE]

@Globalti is nearly right

View: https://www.strava.com/activities/19751519/embed/965568877defd50a917fba5fe0c053242b8192bd

 

[Mr Celine]

Here's a screengrab of the strava segment on the hill that @User9609 is shaking his bike at in his avatar photo. I've done the hill a couple of times but not on strava and I don't recall the vertical pitch at all, though I do have a triple and had to use the granny ring.

 

[Zimbob]

Capture Strava variance... by zimbob.co.uk, on Flickr

I've found it's maybe not so much Strava, as more the device.... A fair discrepancy on the run above as an example, mine recorded on a Garmin Edge 200, my lady friend's on an 820. Oddly, her's shows as slightly longer too, when, if anything it ought to be the other way round owing to her Privacy Zone setting.

It's not a big deal really, but I feel the hills more than she does so it's a trifle unfair