[Closed] Endomondo and height gain

you weren't going fast enough

I don't know how the endomondo metrics work, but I always found they massively overestimated flattish rides, and underestimated mountainous rides.

Had my first go with Endo, this weekend and it seems ok but not that accurate as the route was a fair way, 1/2 mile ish off, in places and not convinced the heigth gain was right either, the readings on my phone and computer were different too. 😕

I must have too much time on my hands... I uploaded the same track recently onto EM and a few other sports tracker sites (RideWithGPS, Strava, etc).
EM reported altitude gain as 1450m, the others (correctly IMO) report more or less 2000m, also as reported by my Edge 705 at the time.

Not sure why they should behave differently.

Not convinced it's recording the data correctly.

Went out and did a ride up and down a hill a few times, ended up with a reading of 50m up and 140m down.

Next time I'm out I'm going to see if you can run MyTracks and Endo at the same time and see if the numbers look the same.

Anything that records altitude data from GPS will be massively inaccurate. GPS is designed to be accurate in latitude and longitude but not in altitude. If you want accurate altitude data you need a barometric altimeter. The problem is not Endomondo, it's that you don't have an altimeter.

it'll be recording the data fine, it's the reader that will be interpreting it wrongly.

I find if you load a gpx in to different programs you'll get different results. I prefer google earth as that gives you bigger values! 🙂

Not sure why they should behave differently.

They all use different algorithms to work out the height gain/loss.

The unit records data at a set point, ie every 10 metres, which includes location and altitude, the software then fills in the blanks to estimate the actuall height gained, so it always only an estimate.
On top of that, getting altitude just from a gps signal is very inaccurate.

"[i]I find if you load a gpx in to different programs you'll get different results.[/i]"

Probably because they'll take a variety of approaches to deal with the fact that the incoming data is crud. For instance, some will leave it unaltered, some will replace the altitude data with figures interpolated from a known height map, others may use some algorithm to try and massage the data, etc. But fundamentally, if the altitude's come from a GPS signal, it's not accurate.

I don't know how the endomondo metrics work

The following is from a previous post of mine on a previous thread. Exactly the same track loaded to 2 websites gives pretty much identical lowest and highest elevation readings, but the total climb/descent is vastly different. That's for a pretty flat ride, so no idea what that would do to a hillier ride
At least one of them is wrong 😉 but which? or both?
My guess is that Endomondo smooths data a lot more.

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

edit: the above was from an etrex vista hcx with baro alt rather than smartphone

Strava is much better with altitude calculations, try exporting the GPX file from Endo into Strava and see what it makes of it.

[i]Endomondo says:
Min/Max Altitude = 114m / 207m
Vertical Up/Down = 67m / 86m[/i]

Endomondo's clearly doing something stupid when massaging the data then, because (assuming that's a contiguous track) you can't go between 114m and 207m without either doing more than 67m of climbing or 86m of descending.

FWIW I did one ride this year where Endomondo claimed a little over 1000m of climbing and the barometer said a little over 2500. Which gives some idea of how far off the combination of GPS altitudes, low sample rates and dodgy data massaging will be.

Bez - true. Assuming that the 114/207m are also correct (the fact that Everytrail picked out the same numbers suggests it might be, although they might both be picking up on the same duff point). Need a map with accurate contours to compare.

I'll stick the same track in to Strava and some others, but I bet it'll be different to Everytrail.

So whats the answer

"[i]So whats the answer [/i]"

What's the question?

So whats the answer

Simple, go out, enjoy your ride and dont worry about the differences. If youve had a good day out on the bike, who really cares which figure is correct.

"[i]Bez - true. Assuming that the 114/207m are also correct[/i]"

It doesn't matter whether they are or not - point is, Endomondo's doing some data massaging that churns out contradictory figures, which is stupid.

I've always taken Endomondo's altitude figures with a pinch of salt. The more up than down on a circular route phenomenom says it all really.

I'm happy that it seems pretty accurate for distance

I found that my garmin (almost) exactly matches the OS maps for height gain/lost. But if you upload it to the garmin site it defaults the USGS mapping which appears to have really broadly spaced contours and always gives a height arround half to 2/3 of the actual garmin.

There is a tick box to select the gps based altitude or the map based altitude though so if you dont tick that it retains the 'correct' data.

No idea for endomondo though.

In my opinion Endomondo can't do altitude because it is not being fed good data. In my experience of using GPS over the past 10 years GPS devices cannot do altitude (eg it says we are at 100m when we at sea level). I have a Suunto watch which does altitude logging well and I just reference that if/when I want to know height info. FYI the watch has to be calibrated as the air pressure changes with the weather and it uses air pressure to determine altitude. So in my view the phone or your GPS has little chance of being accurate.

more up than down

I take *ALL* the stats with a pinch of salt (other than difference between 2 contours on a hill).

PS MapMyRun came out with near identical figures to endomondo (so another favourite == more BS than others). GMaps suggests 114/207m might be about right (although they're not definitive maps). Will try Strava tonight, since that's everyone's favourite for hills 😉

"[i]FYI the watch has to be calibrated as the air pressure changes with the weather and it uses air pressure to determine altitude.[/i]"

Only if you want absolute figures. If you're only interested in height gain then it doesn't matter if all the samples are out by the same amount.

Bottom line is:

- if you really care about altitude, get a barometric altimeter, use software which doesn't massage or replace the altimeter's data, and ensure your altitude data is sampled at high frequency (ie evey few seconds)
- if you don't have an altimeter but you'd still like half decent altitude data, your next best bet is to use software which can replace the altitude data in your track with data derived from a trustworthy (USGS, OS etc) set of altitude data; again, you will lose accuracy with lower sample rates
- if all you have is GPS data then forget it; regardless of sample rate, altitude readings are never going to be reliable in the first place

OK update of my earlier post. Identical GPX file, barometric, uploaded to 4 different sites.

Endomondo says:
Min/Max Altitude = 114m / 207m
Vertical Up/Down = 67m / 86m

Everytrail says:
Min/Max Altitude = 113m / 207m
Vertical Up/Down = 162m / 175m

MapMyRide says:
Min/Max Altitude = 112m / 195m (by mouseover of plot) or 210m (reported in summary)
Vertical Up/Down = 75m / 85m (assumed that end is 10m below start, which is about right)

Strava says: (with a bit of jiggery pokery to get it to actually report anything)
Min/Max Altitude = 108m / 205m
Vertical Up/Down = 226m / 237m

Now I know why people like Strava 😉
Maybe I'll start playing with that again, too 😉

Clearly Endomondo and MapMyRide/Run (ie 2 of the most popular sites) have filtered (or done something to) the data well beyond the point where the maths make sense. Will file a bug report I think 😉

Before I trust Strava, I want to do a loop in Holland.

edit: and for reference, the min/max elev values in the GPX file are:
113.568m
207.356m

Strava is much better with altitude calculations, try exporting the GPX file from Endo into Strava and see what it makes of it.

Oh, and reverse engineering the data for my file above...

Everytrail just sums the total up/down in the file, which means Strava must be making up some elevation gain/loss from somewhere 😉