GPS Accuracy

Essex Man

Anyone using My Tracks and an Android phone to record data for their rides?

I am mapping routes with Bikeroutetoaster, and as an example a route I took at the weekend was mapped with 1500 ft of ascent, but my phone's GPS recorded 2000 ft - which is right? What about proper bike GPS units e.g. Garmins, does anyone have experience of how they tie-in?

Presumably bikeroutetoaster is getting elevation data from OS.

Of course I want to believe my phone....

antfly

They are both wrong, basically.

cougie

Distance will be pretty much bang on - altitude off a gps isn't very good.

danowat

cougie wrote:

Distance will be pretty much bang on - altitude off a gps isn't very good.

This ^

However, you can run your data through packages / pluigins (Sport tracks with elevation correction, and Garmin Connect to name two) that will attempted to correct the elevation and altitude using terrain models.

dmch2

bikeroutetoaster uses altitude from a satellite that has relatively poor resolution. So if a road cuts between too hills it won't be able to tell.

Similarly GPS is rubbish at height, it's heavily optimised for position. My phone frequently reports an accuracy of a few metres position but more like 150m for height.

SBezza

GPS accuracy is dependent on the number of satellites it can see. If you have a 12 channel GPS and get 8 good signals, you will get a pretty good position and altitude determination, if you have the barest minimum of 3 , you will get a poor position and altitude determination in comparison.

GPS's devices are pretty good at altitude, as the signal from the satellite takes a known time to travel a fixed distance, so by calculating how long the signal has taken to reach the device it can work out at what altitude, again this needs a good number of satellites to be seen by the device, not always easy with trees, and building obscuring the view of the sky.

If in doubt take a figure in the middle, and be happy with that, not that the amount of ascent actually means much IMO.

antfly

They might be ok at recording altitude if you are actually in big hills but when you are on the flat they will say you have done hundreds of feet when you haven't done any. You really need a barometric altimeter if you want to know how much you have climbed.

notsoblue

The new Garmin Edge models come with barometric altimeters, I think this is probably an indictment of just how inaccurate altitude measured by GPS is.

Essex Man

OK, thanks for all the views. They are only numbers, I'm not really bothered, I was just wondering if one was "correct".

Of course, when I'm posting on internet forums I'll make sure I'm quoting the over-estimated figure via GPS for any given route

Bar Shaker

dmch2 wrote:

bikeroutetoaster uses altitude from a satellite that has relatively poor resolution. So if a road cuts between too hills it won't be able to tell.

Similarly GPS is rubbish at height, it's heavily optimised for position. My phone frequently reports an accuracy of a few metres position but more like 150m for height.

There is no simple way of putting it but that is absolute rubbish.

As a pilot, I can assure you that any GPS that has a 4+ satellites fix (most, including phones, will have 8-12 at any one time) will know exactly where you are in 3 dimensions. In 3 dimensions, altitude is measured just as accurately as horizontal position.

Before posting again on the subject (especially if not starting with "my guess is...") I suggest you have a read up on exactly how GPS works.

ShutUpLegs

djheaton321

Bar Shaker wrote:

dmch2 wrote:

bikeroutetoaster uses altitude from a satellite that has relatively poor resolution. So if a road cuts between too hills it won't be able to tell.

Similarly GPS is rubbish at height, it's heavily optimised for position. My phone frequently reports an accuracy of a few metres position but more like 150m for height.

There is no simple way of putting it but that is absolute rubbish.

As a pilot, I can assure you that any GPS that has a 4+ satellites fix (most, including phones, will have 8-12 at any one time) will know exactly where you are in 3 dimensions. In 3 dimensions, altitude is measured just as accurately as horizontal position.

Before posting again on the subject (especially if not starting with "my guess is...") I suggest you have a read up on exactly how GPS works.

Being a pilot does not make you an expert on GPS. When it first appeared in the ATPL syllabus, the altitude error of GPS was stressed, so you obviously haven't remembered your studies very well have you? Just to remind you, the errors in altitude are caused by GPS models assuming the earth is a sphere when it is actually an oblate spheroid. My source for this is the Civil Aviation Authority, though the link below also backs up this theory, albeit in a slightly more technical language:

gpsinformation.net/main/altitude.htm

Note the last line, I would guess that you are a light aeroplane pilot?

BTW Shut up Legs, good post!

Bar Shaker

djheaton321

Altitude is still the weakest point in the information output but using articles written about units manufactured in the last century (Garmin GPS 12XL) is not very helpful.

In the 90s, computing power meant that whilst the unit could receive the time stamps from up to 12 satellites, any more than 4 were useless as it couldn't process the information from all but the 4 strongest signals (1 for a reference and 3 for a 3D position, based on a mathematical map of a theoretical surface position). Today, units can process the signals from more than 12 satellites, simultaneously. DOP is a fraction of what it used to be and the use of earth sphere models is no longer needed once multiple fixes are being processed. An example is the Garmin x96 range of aviation units or the glass cockpit models that can be used to run autopilots.

Computing power has moved on massively and GPS units now offer CAT I/II capability so the last line to which you referred me (owners of light aircraft should make sure their insurance is paid up before using GPS to assist landings) is also rather consigned to the last century.

The US have a term for GPS IR approaches, but I forget what it is called. Either way, pilots using it are not dying on a regular basis but this is for approaches - no one with the brain to get a licence would try to land on a GPS only, mostly because obstacles less than 300ft high will not be charted. Then again, you wouldn't use an altimeter alone either unless you could get a QFE/QNH and even then your altimeter will likely be less accurate than your GPS, for altitude. GPS will never be a substitute for an ILS system, but the advances in computing power mean GPS is no longer the toy it was in 1998.

I fly for fun so have no interest in the ATPL syllabus.

That said, perhaps phone GPS systems are no better than the Garmin units of the 1990s, which would explain the OP's problem.

danlikesbikes

Garmin GPS units are all pretty good BUT even the garmin website has its own option to adjust the calculation for height. Thats only because a paper map provided by OS is going to be super spot on and the GPS will be out by circa only a few meters but is enough to make a difference on height climbed.

rich164h

Danlikesbikes wrote:

Thats only because a paper map provided by OS is going to be super spot on and ........

ummmm....no. No it's not. It will be accurate for the trig points but not the areas in between to the level that people believe, especially in rural areas. That said, it's likely to be more accurate than a cheap non-differential GPS receiver measurement.

There's a reason why the OS doesn't supply height data to government departments as part of the existing pan-goverment agreement for geo-spatial data. For cartographic mapping there's nothing bettter in the UK (although even then, the further away from the South East the worst the absolute accuracy of features becomes - although the relative positioning is bloody good), but not for height.