GPSs always overestimate

[Don't fear the sweeper]

GPS apps always overestimate distances
Have you had a suspicion that your GPS app is overestimating the distance traveled? Here's why.
And discussed by geeks here.
(Apologies for the over-simplified thread title, but it seems the title can only be 25 characters long.)

[Knightrunner]

Since abandoning the phone / app combo for a GPS watch I have been at least reassured by consistency. The phone and app were frequently out over known distances by enough to make the thing near useless. I once did an all day walk with a friend who by chance had the same phone and same app: over about 17 miles we were almost 3/4 mile different at the end! My impression is the watch is a far better set up but I have no science to back that up. The article linked above I suppose applies to apps and watches alike, so I'm not sure what's up. I think it be the devil's work ....

[geronimo]

My garmin 310 consistently records the same distance over repeated courses to within a few yards and the same for measured road races is usually spot on

[sore legs]

My garmin 310 consistently records the same distance over repeated courses to within a few yards and the same for measured road races is usually spot on

Agreed.

[tthef]

Reporting the same distance repeatedly is not the same as reporting the correct distance. I have found that regardless of GPS make, distances carefully measured on a map are invariably 10-15% shorter than done with a GPS; I have always assumed that was just because the on map measurements are not accurate enough, but it looks like there is more to it. Also, aren't race distances often based on GPS readings these days?

This also explains why GPS invariably grossly overestimates ascent and descent; the individual measuring errors there are much bigger relative to the cumulative value than for distances, but I could not get my head around why these errors do not tend to cancel out, now I know. Learn something new every day.

[Witton Park]

Defining the correct distance is the first issue?

If you run a distance on a flat plain of 2 miles from A to B, but it involves a 15% gradient up to a summit at 1 mile and down the other side it is 2.08 mile so already either map measure or GPS measure may be 4% out.

I'm not sure it matters - if I run 15 miles and 2000ft on a training run, 5% either way in ascent or distance is not that important.

[Stolly]

I think my suunto is pretty accurate as far as distance is concerned (less so ascent though). For sure measuring with a piece of string on a map is definitely (if only ever so slightly) inaccurate as it ignores the gradient's added distance.

[Derby Tup]

Suunto? Is that a modern version of a Stolly map?

[scubagary]

My Garmin 910 XT and my old 310XT were very accurate.

We have measured mile markers on local road that police use to calibrate vehicle instruments so I often start my Garmin at start and see if it beeps at the mile point. It is always pretty much bang on to within metres. There are no ascents/descents or corners to affect it though. This gives me confidence that a good Garmin watch is highly accurate.

[helix]

Reporting the same distance repeatedly is not the same as reporting the correct distance. I have found that regardless of GPS make, distances carefully measured on a map are invariably 10-15% shorter than done with a GPS; I have always assumed that was just because the on map measurements are not accurate enough, but it looks like there is more to it. Also, aren't race distances often based on GPS readings these days?

This also explains why GPS invariably grossly overestimates ascent and descent; the individual measuring errors there are much bigger relative to the cumulative value than for distances, but I could not get my head around why these errors do not tend to cancel out, now I know. Learn something new every day.

It should be noted that a GNSS device utilising the GPS system is using the WGS84 datum with an angular geographic coordinate system. OS maps use a different datum (OSGB36) which is then projected to a planar coordinate system; British National Grid (made flat and is a better country wide best fit). If measure between points on a map based on WGS84 and measure bewtween the same points on a map utilising OSGB36/BNG you will get a different measured distance. If your GNSS device can use GLONASS the Russian satellite system it uses the Earth Parameters 1990 – Parametry Zemli 1990 which is very accurate in the northern hemisphere but will still differ from OS maps. Likewise when Galileo comes online (you can buy Galileo ready chips online) it will differ from OS maps again as it uses GTRF. The Chinese system Beidou also uses a different system.

Ascent and desent -wise. GPS is woefully bad at measuring height; people put far too much credence in the accuracy. You can get a 3D fix if your device can see 3 satellites; accuracy will improve the more satelllites it can see but will decrease depending on where those satellites are in the sky, your surroundings, weather and any odd things going on in the atmosphere. You should only really use it as a guide unless you have a device which is barometrically corrected and you calibrate it at the start of your activity.

If you are comparing the climb you GPS device reports to Strava the differences are big. Strava uses a really poor low resolution Digital Elevation Model to adjust your amount of climb. It does not do this to certain devices such as the Garmin Edge 900 and if you're brave enough you can hand edit the gpx markup language to trick Strava into thinking your device is a Garmin Edge and report the correct climb.