GPS Discrepancies

[Mick F]

Yes, the plot thickens, and will continue to thicken.

Back in the old days, ramblers/walkers/hikers would count the contours to work out the total ascent. This idea still works.

We now rely(?) on technology to give us "facts" on a screen and different technologies give us different answers.

Man with two clocks .....................

[Psamathe]

If you climb 1m over 3 miles do you call that an "ascent" or do you call it flat and level?. What about 1m over 1 mile? or 1m over ½ mile? etc.

And given the already discussed interpolation algorithms I don't actually see any plot at-all let alone any thickening. Different devices are using different sensors and different algorithms.

Ian

[Mick F]

Most, if not all, of my rides produce 100ft of ascent per mile.
Always have done using the Edge305, the Edge705 and the Montana. These had/have a barometric altimeter.
I also have an Edge20 but that doesn't have the BP altimeter. Its total ascent info is different whereas the others agreed.

I have GPS records going back to 2007 using my devices, but I upload them into Ascent - AppleMac program no longer supported sadly - and I have all the rides to look back on to compare and contrast.

The program smooths out the perturbations of elevation, speed, cadence, temperature, and heart-rate.
I no longer have a HR strap or a cadence device, but I had all that with the 305 and the 705.

Here's a trace from 2010 riding the Raleigh Chopper. I had the 705 and all the kit back then except the Garmin Tempe to measure temperature.

My favourite ride locally, and done clockwise this time.

[mikeymo]

As for ascent as a separate subject, if you have a device with a barometric altimeter to assist the GPS elevation, you will find that the headwind will affect the altitude. Heading into a stiff breeze the hole in the body to sample the air pressure will see a higher pressure and consequently the device will read low. Put your hand in front of the device to shade off the wind, and you will see the altitude increase. The figure when I've experimented, was in tens of feet difference.

That's an interesting point. My iPhone case, a Topeak thing, is reliably waterproof, and is even supplied with a little sachet of grease for the O-ring. Perhaps when it's shut tight it might prevent the altimeter functioning accurately?

[simonhill]

Interesting reading all this, but it's not only about the recording, there is the more serious side if using the data.

On tour in Vietnam a few years ago, had a long 140km ride to do. No accommodation and wanted to finish before dark. I knew thatt the last 100kms were all downhill - steep descent off plateau (wheeeee), then long run down a valley to the coast. What I didn't know was what the first 40 KMS was like. I knew it was 'hilly', but no details. I did a search online and found the ride logged a number of times on (I think) mapmyride. When I checked the ascent, which would all be on the first 40, the amounts varied from 500ish m to 1200ish. Obviously a big difference in time and effort when riding a loaded touring bike.

Given the 3 very different figures, I planned for the worst. If there had only been one figure logged - it could have caused problems in estimating and planning the ride. Lesson learnt - ignore people's claimed ascents.

[KTHSullivan]

Different devices are using different sensors and different algorithms.

I get that but why are they so adrift. One would expect to have different operating systems within two different GPS devices but one would also assume that they would be an approximation of each other if used for navigation. Presumably the devices are produced to a prescribed standard reference somehow.

[Mick F]

I have a plan .............

I have found a 30odd mile ride recorded on a Garmin Edge705 from September 2013 riding Mercian.
I will do it tomorrow riding Moulton using my Garmin Montana mounted as usual on the stem, and my Garmin Edge20 on a mount on the 'bars next to it.

I can therefore report after my ride tomorrow, of distance and total ascent from three separate devices as uploaded into Ascent - my AppleMac program. I can report the raw stats from the Edge20 and the Montana of course, and compare to the Ascent program's smoothed info of all three ..................... but the raw stats from the Edge705 in 2013 have long gone.

[Audax67]

Concerning heights recorded by devices or estimated in mapping programs:

1. The total climb displayed could well be greater that the total of elevation records in the GPX file simply because the unit will only add a record to its GPX file every so often (e.g. according to a fixed time interval or maybe when you change direction) whereas the displayed value will include every climb in between.

2. Some units incorporate both GPS and barometric altimeters. They may display the barometric total but plug the GPS elevations into the GPX files they output.

3. Elevation estimates in GPX mapping programs are mostly based on files from the old Shuttle Radar Topography Mission. These divide a country up into squares, every one of which is assigned the average elevation measured within the square. The squares used by the mission were originally 3 seconds of arc (~90 metres) on a side, so if your path along a hillside was close to the edge of a square the elevation given would put you either much higher or much lower, but rarely at the right height. There's a dead flat section of cycle track near me which, according to the SRTM files, suddenly does a vertical climb of 20 metres.

I believe the files were later issued with a resolution of 1 second of arc, but I've never used them.

Anyway, profiles derived from such files are always as jagged as hell (OK, so are the roads) and need to be ironed out a bit (OK, so do the roads) using a smoothing algorithm (doesn't work on roads ). The result of this can vary immensely depending on the initial parameters the programmer sets. I used to pass the responsibility on to the user.

The sad upshot is that the more programs you submit your GPX file to, the more different results you will obtain. At this point, most people give up trying to understand anything and go for a pint.

[Mick F]

Concerning heights recorded by devices or estimated in mapping programs:
.................... most people give up trying to understand anything and go for a pint.

From your lips, to God's ears!

[Jdsk]

Different devices are using different sensors and different algorithms.

I get that but why are they so adrift. One would expect to have different operating systems within two different GPS devices but one would also assume that they would be an approximation of each other if used for navigation. Presumably the devices are produced to a prescribed standard reference somehow.

There's the calculation of height gained and lost as above, the inherent problem of measuring elevation by GPS, and whether a barometer is built-in and how good it is.

There are reference methods for barometers, but I doubt if they're mandated for these consumer devices. There might be something somewhere in the small print for devices sold in the USA. I'm not aware of reference methods for the other two sources of error.

Jonathan

[Psamathe]

Different devices are using different sensors and different algorithms.

I get that but why are they so adrift. One would expect to have different operating systems within two different GPS devices but one would also assume that they would be an approximation of each other if used for navigation. Presumably the devices are produced to a prescribed standard reference somehow.

If they are using GPS then are they using the same GPS or is one making use of GLONASS and or Galileo or just the old US system?

If using barometric pressure then different sensors will have different sensitivity e.g. raise both 1 mm and they probably wont register a change but raise both 10m and they probably will both register 10 m but there will be some point of "raise" where one sensor will record no change and the other record the change. So hypothetically to illustrate raise both 20 cm and one might record 20 cm the other record nothing - so ride 20 Km going up and down 19 cm and at the end one will record ascent of 0 the other something significant.

And then there is calibration, I'd expect the Large Hadron Collider uses better quality/calibrated pressure sensors that a bargin basement cycling GPS and some will have more effort (and expense) put into calibrating them. Some might have vent holes sensibly placed (to avoid the effect of pressure changes through movement as MickF described). Some might try and be clever by automatically setting elevation when at a "known" altitude point, others might not.Then there is averaging - I'm sure the devices will have some averaging algorithm to avoid sudden false pressure changed (e.g. shock wave from oncoming lorry not becoming a 20 m ascent and immediate 20 m descent) but different units will have different averaging algorithms some will filter out small humps in the road, others might not.

etc. etc.

Ian

[Mick F]

Don't forget that GPS devices with a BP altimeter is only for fine tuning the elevation. The two things combine to give the figure, not solely the BP altimeter.

Devices with only GPS altitude measuring - like my Garmin Edge20 and the Garmin Edge605 - are at a disadvantage, hence they are/were at the cheaper end of the range.

The Edge25 and the Edge705 have/had BP altimeters.

[VinceLedge]

My family has a mix of apple watches and garmin that we use. Distances are always fairly close, but with small differences. Altitude gained is again always similar with differences in the order of a few 10s of feet - hilly round here so usually negligible.
Biggest differences are between the apple and garmin devices, apparantly they use different map systems as a base.

[Jdsk]

As for ascent as a separate subject, if you have a device with a barometric altimeter to assist the GPS elevation, you will find that the headwind will affect the altitude. Heading into a stiff breeze the hole in the body to sample the air pressure will see a higher pressure and consequently the device will read low. Put your hand in front of the device to shade off the wind, and you will see the altitude increase. The figure when I've experimented, was in tens of feet difference.

That's an interesting point. My iPhone case, a Topeak thing, is reliably waterproof, and is even supplied with a little sachet of grease for the O-ring. Perhaps when it's shut tight it might prevent the altimeter functioning accurately?

Could do.

(iPhones have had barometers since the 6.)

Jonathan

[Jdsk]

My family has a mix of apple watches and garmin that we use. Distances are always fairly close, but with small differences. Altitude gained is again always similar with differences in the order of a few 10s of feet - hilly round here so usually negligible.

Sounds good.

Thanks

Jonathan