Is Garmin/GPS distance measurement inaccurate?

[smutchin]

Sometimes I switch if off at cafe stops and when I switch it back on it thinks the cafe is 70-80 metres higher/lower than it was 30 minutes earlier!

This is because local air pressure has changed while you're in the caff. I always leave my Garmin running during the cafe stop (and outside on the bike) for precisely this reason.

Atmospheric moisture can also affect the accuracy of the barometric sensor.

One thing that could help is saving set locations (eg home) in your Garmin with their altitude.

As others have said, elevation inaccuracies will not cause significant distance inaccuracies. Most likely reason for distance errors is too low a sampling rate - you can change this in the Garmin settings. I find once per second frequent enough to get good results.

 

[andrew_s]

To be most accurate, you should probably use a cycle computer with a rear wheel sensor. The front wheel travels further, on account of wobbles and taking the outside line on bends, as can seen from the wheel tracks (when visible).

GPS is accurate, in that you are generally within a couple of metres of where it says you are, and the distance between point A and point B is correct to within 3 or 4 metres.
GPS takes points one second apart and gets route distances by adding up all the straight line distances. The 1-second points don't normally all make it into the track log, as the default Garmin setting is to economise on saved points (in addition, I would expect that upload sites like Strava would also filter out unnecessary points). This will mean that distances taken from saved tracks won't necessarily give the same distance that was on the screen of the Garmin at the end of the ride, which came from the 1-second points. In general, fewer points means more under-reporting of distance.
In addition, the 1-second points get smoothed. Going along a straight road at cycling speed, the points make a nice straight line, but if you go along the same bit of road at walking pace, the smoothing doesn't work so well, and the unsmoothed random errors in individual point positions give enough of a zig-zag to add significant distance if you use all points. You can also get additional distance at cycling speed, when the initial point of a turn gets smoothed and the track continues straight on for a couple of seconds before making a sharper turn.
Between under recording for straight line segments, over recording due to smoothing overshoots or unsmoothed position errors, and under recording because the GPS is on the handlebars and doesn't travel so far when the bike is banked over on a bend, it's not really possible to be definitive on how accurate a GPS route distance is.

(I loaded a number of tracks on to the OS MasterMap computer mapping at work, in preparation for filling in the large blank areas on OpenStreetMap, so I did get to compare GPS positions with real positions down to about 50 cm).

 

[ColinJ]

This is because local air pressure has changed while you're in the caff. I always leave my Garmin running during the cafe stop (and outside on the bike) for precisely this reason.

I suppose that makes sense, but it seems very odd to me that a GPS manufacturer designs a (then) new device with a more modern chipset and an electronic aid to accuracy (barometric sensor) and it turns out to be less accurate than one of their oldest models.

I have run the devices side by side and checked the readings against the OS map and the Etrex is nearly always more accurate than the 500!

One thing that could help is saving set locations (eg home) in your Garmin with their altitude.

I have done that with my home location. The device automatically calibrates itself when I switch it on at home, but it doesn't correct the error that builds up during a ride. I got back to the same point today and it was happy to report an inaccurate elevation

As others have said, elevation inaccuracies will not cause significant distance inaccuracies.

I was the first one to say it, and I did the calculations to prove it!

 

[Dogtrousers]

It's possible to over think things.

You might even be able to apply the coastline paradox somehow and argue that every ride is of infinite length.

 

[ufkacbln]

Try leaving your GPS on overnight and stationary

There will always be some movement recorded as the signal and other factors vary

 

[huwsparky]

(1) I look up the actual elevations on OS maps.

(2) My house definitely didn't move 30 metres nearer to the Earth's core while I was out on today's 37 km loop but according to the 500 it had! Sometimes I switch if off at cafe stops and when I switch it back on it thinks the cafe is 70-80 metres higher/lower than it was 30 minutes earlier!

My old Garmin Etrex is pretty consistent - if I do a loop then the elevation it showed when I set off is the same (within a couple of metres) that it shows when I get back.

Have you got a case on the 500? Mine was more accurate for elevation on a case especially if it was windy. Also have you set your device to record every second?

On the whole I found my 500 pretty good (but not perfect) much better than the 520 I'm using at the moment but to be honest I wasn't all that bothered.

 

[ColinJ]

Have you got a case on the 500? Mine was more accurate for elevation on a case especially if it was windy.

No, it just sits in the standard mount on one bike, or in out-front mounts on my other bikes.

Today I found a few people people online suggesting that wind blowing over the barometer air holes can cause erratic readings. Somebody suggested sticking a bit of foam over the holes to reduce the wind effect, while not interfering too much with pressure detection.

Also have you set your device to record every second?

Er ... I just had a look and it was set to 'Smart Record'. I've changed it to every second and will see if that makes any difference.

On the whole I found my 500 pretty good (but not perfect) much better than the 520 I'm using at the moment but to be honest I wasn't all that bothered.

My cousin gave me the 500 when he bought a newer Garmin. I found that it was hopeless for navigation so I kept my old Etrex for that and just use the 500 as a GPS-based bike computer. I have it set to auto-scroll through 3 screens showing the main information that I want to look at on rides. If I need help navigating then I take the Etrex as well.

 

[jefmcg]

Yes, auto pause below 2mph.

Wait! What?

You at beginning of ride: "Dear Garmin, please stop recording me when my speed drops below 2mph"
You at end of the ride: "God damn, Garmin, why didn't you record every metre of my trip??"

Try again with autopause off, and let us know how you get on. Autopause is to increase your average speed, at the expense of missing your slower parts. Distance is always going to be less with autopause on.

 

[Pat "5mph"]

We rode together along exactly the same route yet he has recorded 14.00 miles and my GPS is showing 13.63 miles

Maybe Garmin is also influenced by the metric system: we know that miles are not metric, but when you watch the satnav count, it goes "10.98" 10.99" "11" miles.
Maybe try km and see if there's a difference.

 

[Dogtrousers]

GPS records your position in space every x seconds (2 or 3 dimensions, depending on the device).

It calculates distance by "guessing" the route you took between the two points. Mostly simply using Pythagorus and ignoring elevation, assuming your journey is a series of straight lines between the points. More sophisticatedly using elevation and estimating a curve that is closer to your actual line.

So it's always going to be a little off. The more points it records, the more accurate it will be but the shorter your battery life.

Pythagoras won't cut the mustard for the distance between two coordinates. When I've written stuff dealing with GPX data I've used a thing called the Haversine formula. I'm not claiming to have understood it, just used it.
There may be an even more complex variant that uses elevation too, but I've always figured that that would just introduce a load of noise while trying to fix a problem that's too small to warrant fixing.

 

[jefmcg]

Pythagoras won't cut the mustard for the distance between two coordinates. When I've written stuff dealing with GPX data I've used a thing called the Haversine formula. I'm not claiming to have understood it, just used it.
There may be an even more complex variant that uses elevation too, but I've always figured that that would just introduce a load of noise while trying to fix a problem that's too small to warrant fixing.

You got me. I googled a formula when I needed it.

 

[Ajax Bay]

I don't believe the GPS, am I wrong?

Yes. If you have to believe anything, believe a GPS which is working as it should.

I think it is to do with the fact that GPS doesn't taker into account the elevation effect on distance, whereas the wheel sensor does.

This is a good answer.

As others have said, this is not a valid explanation and no 'answer'.
The distance one cycles up a hill and down again is so close to the 'flat' distance that it makes little significant difference, as part of a ride. For example (and simplistic profile) if you have 5km (GPS-flat) at 5% and the same down then the total distance cycled is 0.4% more: 10.04km (an extra 40m). Within an otherwise flat 40km ride the distance measurement difference is less than 0.1%, and you've still climbed and descended (say) 500+m. Flatter and there's even less difference.
I suggest a test for you: map out a long ride on Ride with GPS and follow it, starting the trip on your (attached with a lanyard?) Garmin Edge Touring (once satellites acquired) at the start. How accurate is that? I have confidence that the RwGPS calculations are very accurate. I base this on my 2016 rides, mostly 100km plus where my Garmin 500 distance reading came within 0.5% (often closer) to the accurate route I plotted on RwGPS post hoc.

My house definitely didn't move 30 metres nearer to the Earth's core

But during your ride the barometric pressure (measured by your 510) rose slightly: about a mBar.

 

[ColinJ]

But during your ride the barometric pressure (measured by your 510) rose slightly: about a mBar.

Indeed, but exactly what is the point of adding sensors which actually make the device LESS accurate!

I was expecting the Edge 500 to be more accurate than my old Etrex which usually has an elevation error < 2% and often nearer to 1%, but it isn't.

I have read lots of times that 'unassisted' GPS is not good at measuring elevation but I have found my old GPS to be pretty reasonable at it.

 

[I like Skol]

Yes. All measurements are inaccurate.

No they are not! There is an accepted tolerance and some measurements are within tolerance while others are not. Measuring something as mid range as miles to only 2.5% accuracy is plss poor. We are not talking about light years or Angstroms!

 

[I like Skol]

Wait! What?

You at beginning of ride: "Dear Garmin, please stop recording me when my speed drops below 2mph"
You at end of the ride: "God damn, Garmin, why didn't you record every metre of my trip??"

Try again with autopause off, and let us know how you get on. Autopause is to increase your average speed, at the expense of missing your slower parts. Distance is always going to be less with autopause on.

Stop talking bollocks!!!! The only time I am moving at less than 2mph is when I am stopped. This will only affect average speed. All I want is for the Garmin to record all the positions I have been in and report the distance that connects them. Even on a steep hill that most mortals will get off and walk up I will be doing 5-6mph minimum