GPSs & altitude discrepency

Did a (great) ride last night over winter hill

But interestingly there is a 100m difference in the altitude gain between mine (recorded on an android/strava) & between my mates on a new Garmin Edge.

His (the garmin edge) recorded a 100m more.
Then looking furthur the heights recorded for top of winter hill & another peak,
mine agree with OS maps perfectly his are 81m low for both.

Quite a few factors at play: sample rate; accuracy of location; accuracy of mapping; tree cover; landscape, also elevation data is the hardest part for consumer GPS units to get right. If you go over a small bump, say a metre high, and one of you records a position on that bump and the other records a position just before or after it then you’ll have a 1 metre difference for that section. It can add up over the length of a ride.

In Strava you can click a link that attempts to correct elevation data using the barometric data that some units also record, sometimes it makes significant adjustments, sometimes it doesn’t.

I’ve a Garmin Edge 510 and I find that if I switch it on about five minutes before setting off on a ride then I’ll get more accurate (or at least more consistent with others’ data) elevation data. It’s as if the unit needs to settle down and figure out where it is before you start moving.

I thought Garmin uses barometric altimeter as default and if you “correct” it reverts to (less accurate) map data?

Android/Strava must use map data as default (which usually seems “under” to me).

what happens if atmospheric pressure changes during a ride – our weather can be ‘changeable’ even within the length of one of my rides!

I quite agree with theotherjony’, barometric pressure can vary hugely over a day.

Everything I read says barometric is better especially with the some GPS reset feature it has which might be what whitestone is seeing.

But aatleast comparing both against OS maps my android/strava wins hands down over the garmin using barometric.

Regarding accuracy: Garmin http://www8.garmin.com/aboutGPS/ reckon their units are accurate to within 15 metres. That article mentions WAAS which is a wide area version of Differential GPS which what surveyors use. I worked on a seismic survey crew using a differential GPS system, it was accurate to 3cm in all three axis but took 30 seconds of signals to get that accuracy (the update, at $20,000 a unit!, dropped that time to 3 seconds).

In practice the units seem to be more accurate than that but it does depend on the layout of the constellation of satellites – if the satellites in view are all clustered near the horizon for example then you are going to get a lower accuracy than if they are all high in the sky and nicely spread about.

Next up is the mapping accuracy. What you are looking at is topological mapping rather than features. The earth isn’t an exact sphere, it’s fatter at the equator for example due to its rotation. Mapping agencies get round this by using an ellipsoid There are several of these which are accurate to a greater or lesser extent on different points on the globe so the Ordnance Survey (for example) will use the ellipsoid that is most accurate for the British Isles. GPS systems need to use a generic one. The GPS “map” is actually a grid of points that are the effective height from the surface of this ellipsoid. The grid may be 10, 20, 30 metres between points.

Now when your GPS unit locates where it is, if it isn’t exactly on one of these points then it has to work out its elevation using the points around it. This can be a simple “I’m halfway between A & B so I’ll split the difference” or something more sophisticated like a Bezier Curve which doesn’t assume that there’s an even gradient between each point.

Given the potential error in precisely locating where you are on the surface of the earth and the working out of your elevation then even if you carried two or more GPS units they are likely to give different values for the same ride.

But aatleast comparing both against OS maps my android/strava wins hands down over the garmin using barometric.

garmin beats my android any day. BUT I calibrate the barometric at the start point of the ride, at least to the value noted on google maps (when cycling), or the altitude written on a ski lift station (when in the alps skiing/riding).
By the end of the ride, on a bad day, it’s 10m out when I arrive back at the start point.
Android Strava/Endomondo, is always well out to begin with, and well out at the end, by differing amounts on the 2 phones I’ve used.

Does strava use the vertical data? or does it calculate it using maps?
Noticed this weird section the other day:
https://app.strava.com/segments/5703107

KOM’s VAM is 87200 – pretty impressive!

Happens to me and my brother every time.
His Garmin Edge 300 always give different readings of hight and distance to me using a iPhone 6 with the Strava app.
Anything upto half a mile on a 10 mile ride

don’t know what Strava does (actually I don’t believe anyone can say for certain what any of them do), but Endomondo on Android I’d say is consistently offset 45-55m above what the map says (at least around here).
distance IME across all platforms is pretty consistent. maybe +/- 0.1km most likely due to rounding errors and how many points they use in smoothing.

Standard GPS altitude is not terribly accurate. A Garmin will generally have a barometer, but as said they really need calibrating (I think this is done by marking a location with known altitude? Have to admit I don’t bother usually), and atmospheric conditions affect things. Would movement also affect it as air pressure against the device surely changes? Also is there a difference if you have the thing in a pocket?

Also as said, Strava in particular will apply corrections based on topographical mapping data, though it will prefer the Garmin’s data if you used one. When Strava does corrections it take a little while to update on their web site after upload.

Generally I’m not too fussed on the actual altitude so long as the elevation gain is about right, but even that seems to vary between us in the same group of riders, quite dramatically.

As for distances, there’s a difference if you use a speed sensor as it tracks actual distanced rolled on the ground, not GPS distance. Though needs the correct wheel size calculated, which a Garmin may do automatically. Also I find Strava “corrects” the distance and what my Garmin says doesn’t match what it says.

I don’t really trust the elevation data given by my GPS for reasons explained above, so if I’m downloading files at the end of the ride then I stick them in Tracklogs and see what the elevation model on there reckons.

I think this is more accurate (and usually lower), although I may be kidding myself.

Obviously when it comes to bragging about my ride on the internet, I just use whichever figure is the highest 😉

Obviously when it comes to bragging about my ride on the internet, I just use whichever figure is the highest

IME Strava 😉
Endomondo smooths altitude gain/loss (although what it does in the app, and what it does to Garmin acquired GPX data are different)
Strava makes up some extra altitude gain/loss 😉

edit: and why I always use km (even when I still lived in UK) – bigger numbers are better 😉

This article suggests Strava will use elevation data from devices with a barometer, but not from anything else (i.e. reverts to calculating from map elevation data):

https://strava.zendesk.com/entries/20944466-Elevation-Gain

I use Oruxmaps on android which offers the option of using DEM data (hgt files) which then get written into the resulting gpx log, however Strava will still not use the elevation data from the gpx.

The strava article also mentions smoothing and thresholds, which explains why Oruxmaps generally always states a higher elevation gain than Strava at the end of a ride.

GPS elevation is not the same as height above sea level. The former is calculated using a “reference ellipsoid”. Where I live, the GPS elevation is about 145m but the height above sea level is 100m. The difference varies from place to place. Unless a GPS device has some way of compensating for the difference, it won’t know what the height above sea level is.

As regards height gain and loss over the course of a ride, my Android phone always overestimates massively compared to my altimeter watch. But when I recalculate the height gain and loss in the GPS log by using digital mapping data, it generally comes out pretty similar to what the watch says.

There’s a reason aircraft use barometric altimeters rather than GPS based ones.

No reason to think that should be any different for cyclists 😉

The barometric altimeter on the Garmin is likely to be accurate for short term changes, but will nearly always tend to develop errors over time.

Why? Firstly, the air pressure in the UK tends to change quite rapidly unless a high pressure system is wedged over the UK. As a front moves through pressure can fall by 30-40 hPa over the course of a morning, ie, a false elevation gain of 1000 feet. You can mitigate against this to an extent by calculating the regional pressure setting by GPS.

The other – and I suspect more predominant factor – is that as the wind funnels around hills and valleys the Venturi effect causes the pressure to fall. Only slightly, but if the wind gusts on a continual basis this slight gain, and loss, of altitude adds up over time. Even holding the GPS in a strong wind might influence this, let along topography.

Together you can get more than 2000 feet of false elevation gain, which looks great for bragging purposes but sadly fools no-one.