Can anyone decipher this Strava reading?

Fursty Ferret

I don't use Strava but there's a hill local to me that I climb quite regularly and I always wondered what the gradient was, so I had a quick peek on Strava and there appears to be some peculiar readings. The hill in question, Blowingstone Hill in S Oxfordshire, seems to have an ave gradient of around 10% over the course of its relatively short duration, but with occasional spikes of seemingly ludicrous proportions, according to Strava. Or is that entirely normal? I have no idea. See below

http://app.strava.com/segments/616102

simona75

You cant trust the micro level gradient measurements on strava (you will see on some hills they have measurements of over 100% which would be an overhang )

I've been up Blowingstone Hill a few times and i would say 10% is about right (with a short sections of 14-15%) I prefer Dragon Hill road myself as you can stop at the top and admire the view

xcmad

I can beat that, check this out. A simple down and up dip, but according to Strava its covered it rocks!

http://app.strava.com/segments/1891951

No idea why this happens.

AchillesLeftKnee

GPS only altitude readings are a very dubious prospect indeed, and if the segment was created from GPS only data, that's the way it ends up looking. Segments created with altitude data from devices with barometric altimeters tend to be a lot more accurate, although will often contain what appear to be steps instead of smooth gradients because of the relatively low sampling rate of the devices (using a Garmin in averaging mode, for example).

Strava doesn't exactly cover itself in glory when it comes to handling the data either. Given that the workings of Garmin's averaging method are very transparent, smoothing out gradient data isn't exactly a tough coding task.

Fursty Ferret

Interesting. Makes one wonder to what extent you can actually trust these sorts of tools.

AchillesLeftKnee

Fursty Ferret wrote:

Interesting. Makes one wonder to what extent you can actually trust these sorts of tools.

I think it depends on what you're looking to get out of it.

Personally, I'm only really fussed about my solo performance vs. my previous solo performances on a selection of segments, so in that sense it works pretty well for me. The timings aren't perfect because segment start and end points are interpolated between samples from the Garmin, but for longer segments it's close enough.

TakeTurns

Check this. It's actually completely flat, but strava says some parts of it are 116% grade :shock:

http://app.strava.com/segments/1010469

racingcondor

The basic problem is that Strava calculated power data, altitude data and gradient data are all rubbish. It's fun and a reasonably useful tool for tracking your progress over time but I wouldn't expect it to be all that accurate.

Seems like Strava are working on data quality though, I was going to post the Embankment (London) section with a fastest average of 67mph but I can't find it anymore (which, it has to be said is a good thing...).

barrowmatt

I have the Strava running app and I can do a loop from my house and have an overall elevation difference +35ft when I get back!

ShutUpLegs

http://app.strava.com/segments/725282

Gradient 472%

pinarellokid

would definately be a job to even walk up that one.

gerroffandmilkit

pinarellokid wrote:

would definitely be a job to even walk up that one.

Don't know about clipless, more like a requirement for crampons on that last section!!!

ricky1980

lol GPS elevation is funny. A guy at work swears that GPS elevation can be accurate to 1mm and top speed shown is deadly accurate.

so basically some of you must have been running up vertical hill cliffs while carrying your bikes on your back...spiderman like...or KUNG FU masters...correction KUNG FU GODS

Guanajuato

Have a look at the OS map - that'll give rather more accurate elevations.
http://www.streetmap.co.uk/map.srf?X=43 ... &A=Y&Z=120
So it climbs from 123m to 205m in about 3/4km - I'd make that pretty much bang on 10%. The contours are fairly evenly spaced, so it looks like it should be a pretty steady gradient.

GPS is accurate to about 10-20ft horizontally (it can be better) but the vertical accuracy is in the order of +-100ft - it was never designed as an altitude system. The signal is effected so much by things like humidity and cloud cover that, by itself, the signal couldn't possibly be relied upon to give anything other than a vague idea of elevation.
Whats worse is that Strava will use satellite-based elevation data to 'correct' elevations. However, that struggles with tree cover, 'urban canyons' and sharp elevation changes, such as you'd get going up/down a steep, wooded hill.
As an example, the GPS on my phone says I'm at 100m elevation. yet a Castle on a hill that's a good climb as 93m, so my real elevation is about 45m.

Fursty Ferret

Guanajuato wrote:

Have a look at the OS map - that'll give rather more accurate elevations.
http://www.streetmap.co.uk/map.srf?X=43 ... &A=Y&Z=120
So it climbs from 123m to 205m in about 3/4km - I'd make that pretty much bang on 10%. The contours are fairly evenly spaced, so it looks like it should be a pretty steady gradient.
.

It's a weird little hill. It's not quite uniform in terms of gradient. First 30 yds are gentle, next 100 or so really do cause some pain and then it 'levels' out for a steady climb at around 10% but the damage has been done by that point. Still, 'tis a mere pimple compared to many.

mikeeye

My take on this is that Strava is just showing the data that has been uploaded by each rider. So if you look at that segment as ridden by different people, you get a different trace.

E.g. here are two traces I have for the same segment ridden on different days and recorded with different devices:



My guess is that the trace you see on the main segment page is the original trace as uploaded by whoever created the segment - but that is just a guess. Arguably, the main segment page's trace ought to be smoothed, but it's probably not unreasonable to leave the traces from each individual ride untouched.

Pumpkin Positive

ShutUpLegs wrote:

http://app.strava.com/segments/725282

Gradient 472%

Is that the hill by the glider club?

If so, 472% sounds about right to me...