Only recently starting using Endomonodo, more cos I’n trying to learn some local trails well enough to guide on them and it’s useful to look at the ride in a map after than anything, but I’ve been looking at the statistics and there’s a glaring error… I rode 16.5 miles offroad today, Endomondo says I did 1085m of climbing and descending! I can tell you if I did half of that I’d be surprised. Likewise, Wednesday’s road ride, did 30 miles and it says I did 850m of climbing. Now it was far from a flat road ride, but again I’d be surprised if it was even half of that!

Anybody got any ideas why this is happening? Running an up to date version of Endomondo on my iPhone 4, running iOS5 if that makes any difference. And the problem seems consistent. Does it need calibrating in some way perhaps?

Its free?

Endomondo doesn’t “make up” the data – that’s all recorded off your iPhone.

Ok druidh, see where you’re coming from. So why am I getting this spurious data then? Is it a known fault?

I can’t find any data but I guess that the 3D accuracy of the IPhone is probably not fantastic. Dedicated GPS units often come with an altimeter and I think that they require calibration.

Dedicated GPS units often come with an altimeter

yup that’s the big difference. GPS is designed to tell you where you are in the world, it kinda sucks at telling you how high you are though

Have you got it on battery saver? I’ve used Endomondo for a while and tried the battery saver, the GPS trace was just useless, straight lines, 100miles+ distance for 15 mile rides, ridiculous top speeds (I think 500+ mph was the best!)

I used Endo for a while on my HTC, but it became a bit pointless because it was so inaccurate. I now use Runkeeper, which has been nothing but 100% accurate, whether I’m walking, running, cycling or MTBing.

doesn’t “make up” the data

Well they all (sort of) do actually. They’ll all apply some kind of smoothing to the altitude, and how much will affect the end result.

With 0 smoothing, it’ll add up all the +1ft,-1ft inaccuracies as you ride along a totally flat canal towpath, and tell you that you went up X ft and down Y ft, where X and Y are basically nonsense. With a bit of smoothing, it’ll remove that.

Endomondo, raw Garmin readout, Everytrail all give different results, with the same GPS data from the same device. Never tried iPhone. In fact never tried non-barometric alt GPS device, so no idea how they compare.

i’ve found the iphone GPS to be the most accurate when riding.

a lot of guests have altometers or GPS on their bikes and many have iphones.

i’ve only got a map.

the iphones seem to be the most reliable (after the map, of course). usually with 2-3 metres. the GPS are often 50m out and the altometers can be way off depending on air pressure.

i’ve found the iphone GPS to be the most accurate when riding.

Got to be said I’ve just looked back on a couple of routes I did before I updated the iPhone to iOS5 and I’d say the altitude gain/loss was bang on the money on those. Since I’ve updated it seems way out.

Coincidence?

what i have noticed, and this must be a function of the application rather than the phone, is how different applications tell me my altitude is different. Endomondo is fine at recording the route but crap at recording the altitude. Currently trying the Garmin app to see if it is any better.

If i take a route recorded on my garmin, and import it into endomondo, training centre and google maps, they all give widely different figures for ascent descent.

They all use different algorithms for calculating the figures from the raw data.

Going up mont ventoux earlier in the year, from Bédoin, which is a well known and recorded height gain, endomondo underestimated by 300m, google maps was 150 over and training centre 200m over.

ps: I use the training centre figures for bragging rights of course.

I binned runkeeper in favour of Endomondo Pro, the latter is much better at mapping etc. Got nothing to compare it to for altitude, but I reckon it’s alright

I noticed the same running Everytrail on my iPhone – I put it down to the phone’s shortcomings as an altimeter. Oddly it would always over-estimated the height gain rather than under. Otherwise, it’s been spot-on for telling me where I am – very impressed.

Gammon Connect has a function where it uses the 2D trace from the unit overlaid on a map and uses the map to work out altitude data. It roughly halved my climbing too but much more reliable. Does Endo have the same setting?

Endomondo has always been way off for me, I have the paid version.
If I care then I’ll grab the mapping data from Endomondo, import it into tracklogs and get the altitude details from that, update Endomondo. Endomondo is always way lower than actual height gained.

There was a thread on here soon after iOS5 came out saying just that. I did an MTB ride yesterday which came to 3900ft of climbing on my Edge 705. The chap I was with, who did slightly less in real life had 8800ft on his iPhone and 6700 on Garmin Connect, which was still wrong!

They’ve obviously changed something on iOS5.

I never take the altitude reading too seriously, Endomondo’s favourite trick with me is that it almost always records more descending than climbing on the loops I am riding. If only it were true!

2D trace from the unit overlaid on a map and uses the map to work out altitude data

Have to be careful with that too, especially if traversing round a hill or ascending/descending with a steep “cliff” either side. Any small lateral GPS errors can lead to unwanted altitude elevation errors.

No idea how people can say “I think XXXX device is the most accurate”. Only real (user) way to measure them is on a reasonably uniform hill with accurate start+end altitudes. And stick the same data on 2 devices through 3 applications, and you’ll have 6 answers (plus 2 more based on the devices internal elevation integration). Descend the same hill on a bike and you’ll get 6 more answer, many of which won’t match those of the ascent, just because of the substantially fewer data points. Or at least that can be the case after a few hills.

I just use them as an approximation (or bragging rights).

andytherocketeer – Member
…Only real (user) way to measure them is on a reasonably uniform hill with accurate start+end altitudes. And stick the same data on 2 devices through 3 applications, and you’ll have 6 answers (plus 2 more based on the devices internal elevation integration). Descend the same hill on a bike and you’ll get 6 more answer, many of which won’t match those of the ascent, just because of the substantially fewer data points. Or at least that can be the case after a few hills.

Sounds like a good case for a STW mag test. Surely there must be an expert here who can do it.

But how do you get those accurate start and finish altitudes!? Barometric is certainly more accurate, IME any that use GPS are always out. I think consistency is more important than the actual number!

If your distance is in miles perhaps the height is in feet. Never used endomondo though so really I have no idea.

It’s because GPS looks down, more or less.

If you watch a man climbing up a ladder from the side you see how high he’s gone. If you watch it from the top it’s much more difficult.

It’s because GPS looks down, more or less.
If you watch a man climbing up a ladder from the side you see how high he’s gone. If you watch it from the top it’s much more difficult.

Errr, thanks for the physics lesson, would never have got that one otherwise! 😕

Seriously though, just the knowledge I’m not the only one makes me feel better, will stop worrying about how much climbing I’ve done now then. I do think as a GPS tracker it’s bloody amazing though, seems to be dead accurate when you zoom in. And for accuracy sakes, did the same route with my friend yesterday, my iPhone recorded 26.42km, her HTC recorded 26.43. A 10 metre difference could easily be accounted for just taking a few wider lines on corners I’m sure.

Sounds like a good case for a STW mag test. Surely there must be an expert here who can do it.

2nded

There’s got to be a few smartphones at STW towers I’m guessing. Given the prevalence of how popular they are these days, and how many such apps you have got, I for one think a detailed review on which app to buy/use for what purpose would be very well received, and probably more useful to many than another which wheels/brakes/lights/shorts/handlebars/shoes review. We’ve had a GPS unit review in the past, surely now the GPS app for smartphones review is the logical progression?

Chipps take note… And if you want some help, as always happy to help… 😉

But how do you get those accurate start and finish altitudes!?

one of those legacy paper things called a map?

for my regular loops, I just use a value determined from Google Earth. No idea how accurate it is, but I just calibrate the (barometric) altimeter on mine at the start of each ride just for consistency. Most of my rides start from that same landmark in town. When skiing,I just calibrate it to the altitude that’s on all ski maps/lifts at the start of the day.

Sounds like a good case for a STW mag test.

One of these days, I’ll set all of my GPS’s recording for 5-6 hours on my balcony or somewhere. Trim the ends of the GPX files so they’re all identical start/end times, and to remove any warm up/lock errors. Then put all the tracks in a selection of apps to see what the total distance and elevation change is of my house. Just for a laugh. The one closest to zero is more likely to be the one with the harshest smoothing, rather than the most accurate (will need to do some stats of the raw data too).

thanks for the physics lesson,

I know one about cows too, involving the small vs far away concept 🙂

I was trying to say that’s why GPS systems can’t be that accurate unless they use some proper integrated altimeter style data or very accurate correlation with maps…

I find anything off road under tree cover etc gives me errors but I did a road race on the weekend and the mileage and ascents/descents were almost spot on. Not going to be as accurate as a proper GPS system

I’ve got a garmin edge 305.

Thought it was woefully inacurate when compared to the garmin website and mapmyride, but with a bit more effort, counting the contour lines on an OS map, *5 = distance climbed/decended (10m spaceing, divide by 2 for up and down), showed that the 3D gps trace wa spretty much spot on, never more than 20m error in arround 10miles.

I’ve had an issue with cyclemeter since I updated to iOS5. Altitude reading are around double what they were before the update. I wasn’t sure if the issue was with the iOS update or the updated version of cyclemeter. Seems likely the issue is with the phone.

Did the laiest ios update at the weekend – ios 5.0.1 and the altitude reading has gone back to previous stats so all fixed now.

One of the updates on ios 5.0.1 was to add ‘elevation filtering’

OK did a test with today’s short loop of a ride. Identical GPX file uploaded to Endomondo and Everytrail. Barometric altimeter calibrated at start of the ride, and all datapoints before that (ie while locking etc.) discarded.

Endomondo says:
Min/Max Altitude = 114m / 207m
Vertical Up/Down = 67m / 86m
Max/Avg Speed = 35kph / 16.9kmh
Distance = 19.65km

Everytrail says:
Min/Max Altitude = 113m / 207m
Vertical Up/Down = 162m / 175m
Max/Avg Speed = 43.7kph / 16.9kph
Distance = 19.6km

Speedo says 43.65kph max (so almost identical to Everytrail).
Map says start point is 10m higher than end point (so Everytrail is pretty close there too on the total up/down).

Conclusion:
If anyone says “Oh my iPhone/Android/Garmin is accurate for speed/accumulated altitude/etc.” then they are talking out of their backside. GPS only records (x,y,z,t) and it’s the application used to view that data that derives Speed, accumulated distance, accumulated altitude, etc.
My guess is that Endomondo smooths the data a lot more than Everytrail. One of them is wrong. Or both. But imho, Endomondo is more wrong than Everytrail.

Update: and just for a laugh, I tried MapMyRun too.

Chart shows 112m / 195m for min/max altitude. Text says 210m max altitude. And 71m total ascent. Reasonably close to Endomondo stats.

Further proof that it’s not (just) the raw data where inaccuracies originate from, and you have to take everything with a pinch of salt.

Next time I’ll do a hill where I know it’s consistently all up, with not even any small up/downs along the way.