GPS measures long?? Why?

[kanangra]

I have long suspected that my Garmin GPS measures long and have now confirmed it. On Day 1 of a walk in the Bogong wilderness a while back the GPS measured 56km. According to the "Rooftop" map of the area the true distance is 48km. (By the way these maps of the Snowies are excellent. They have contours at 1:50,000 scale and show much more detail than the Govt toppo maps.)

Has anyone else noticed something similar? I was wondering why this happens?

K.

[GBW]

I've noticed the same.

Just did a section of the AAWT from Baw Baw to Mt Howitt and the distances in John Chapmans book dont match my Garmin log. I realise his distances are approximate but all my Garmin tracks are longer by 15-20%

example:

Whitelaw Hut to Stronachs camp = 9km Garmin = 11.9km
Stronachs Camp To Thomson River Camp = 11.8km Garmin = 14.2km
etc.

Ascents and Descents are greater also.

[Mark F]

It is to do with error, high sampling frequency, people and tracks not travelling in dead straight lines etc. The distances you use as a reference measured from a map are not accurate either. They suffer from a very low sampling frequency, measurement errors etc. There have been a few threads about this on bpl and I think here as well.

The best analogy I can think of (from a Geography exam at UTAS) is measuring the length of a coastline. Choose your sampling distance - it is like a stick of a fixed length. A handheld Garmin gps uses about 3 to 4 metres but gives a reading to 1 metre but also has an error of 5 to 20 metres which often cancels out over a long distance. If the sampling distance is long then many little ins and outs are missed and curves are short cut making the measured distance shorter. As the sampling distance decreases you are measuring more often and finding smaller indents etc that get included in the distance measured. If the sampling distance becomes small enough you are measuring around small pebbles or bits of driftwood and extending the length measured.

[madmacca]

GPS data points can contain an error of up to 25 metres (but usually considerably less than this). So if you plot the reported points together, it will show a slight (and longer) zig-zag path, rather than a straight line.

At walking pace, in some cases, the error will be behind you, possibly fooling the GPS unit into thinking you backtracked (and thus adding to the distance).

[bumpingbill]

distances in John Chapmans book dont match my Garmin log.

In your case, it could well be Chapman's book.

While walking the Tassie South Coast a few years back, expletives over Chapman's distances were regularly heard.

[icefest]

In fact there is even a Wikipedia article on this exact thing: http://en.wikipedia.org/wiki/Coastline_paradox



There's been several discussions about Chapmans distances in the past. Search the forums. My theory is that he has calculated the distances via a dividing caliper.

[Gadgetgeek]

another thing that can affect the end result is whatever dead-reckoning algorithm the system is using. depending on what info its using for figuring out where it is, that can cause some of that zig-zag, sometimes its corrected for, sometimes not.

[GBW]

Going on the Coastline paradox you would expect gps distances to be less than actual distances, not more. Looking at my tracks up close, there doesn't appear to be a lot of zig zag between the points, and it's hard to believe Chapman could be that far out with distances...could he? I have an athletics track near me and I'll check it out when I get home.

[icefest]

Going on the Coastline paradox you would expect gps distances to be less than actual distances, not more. Looking at my tracks up close, there doesn't appear to be a lot of zig zag between the points, and it's hard to believe Chapman could be that far out with distances...could he? I have an athletics track near me and I'll check it out when I get home.

It's actually the other way round. The GPS has smaller base units than map based reckoning.

I'm assuming Chapman used the 25k topo and then measured the distance (probabaly with a reasonable distance like 5mm which is 125m on the ground). The GPS, with decent smoothing algorithms applied will be reasonably accurate with a 10m sampling distance.



This is all ignoring the fact that some GPS units will calculate diagonal distance in 3 dimensions, which can be significant when you are ascending/descending a lot, which excludes the SCT.

[South_Aussie_Hiker]

GPS data points can contain an error of up to 25 metres (but usually considerably less than this). So if you plot the reported points together, it will show a slight (and longer) zig-zag path, rather than a straight line.

At walking pace, in some cases, the error will be behind you, possibly fooling the GPS unit into thinking you backtracked (and thus adding to the distance).

Um, yes and no.

Selective availability has been turned off for years. The only reason a modern GPS unit would have an error of 25m would be because of poor coverage/not enough satellites. With full coverage, the error even on civilian GPS is normally significantly less then 15m. Baro aided GPS will always be significantly better too.

In terms of the errors causing zig zagging or backtracking, this would only occur in places with degraded coverage. You seem to imply that the "up to 25m" error is random in nature - and that is incorrect. A unit with full satellite coverage which is 10m in error does not mean the position solution moves around randomly within 10m of the actual position. For all intents and purposes on a bushwalking GPS, the error will remain essentially constant in direction and distance. It will not change quickly enough to cause back tracking or zig zagging and subsequent over estimation of distance.

My money for your distance discrepancy is on one of two things:
1. Satellite coverage at times was poor (will show on a map of your track)
2. Chapman distance is wrong.

My money is on number two. The vertical profile which the GPS takes into account may explain it a little, but would not account for such a large error.

[neilmny]

This is an interesting topic.
Do you have any ideas on this one Sth Aussie.
On the same walk at the same time , which would contain the odd random slight track deviation, in other words we don't walk in each others footprints, my son inlaws Garmin GPSMAP?? always records a longer track than my Vista Hcx by around 8%.
We have endevoured to use the same recording settings to solve this but the difference is still there, although they are fairly generic descriptions such as "auto" and "more often" etc.

[michael_p]

This discussion reminds me of the ancient proverb - "Give a man a GPSr and he will know where he is. Give him two and he will never be quite sure."

On a serious note. There are umpteen factors that can screw up a GPSr measurement. I've been using gps for 10+ years for mapping of walks and mtb riding tracks and have encountered anomalies a plenty.

Several times I have re-mapped tracks usind the same GPSr under virtually the same conditions and found the new log substantially different to the old one. One time my GPSr was out by 200m most of the day then amazingly jumped to within 3m near the end of the day.

I don't have a definitive answer to Kanangra's question. Possible errors that could affect the measurement:
- Satellite tracking problems and the unit wandered a lot.
- Recording too many waypoints. This can make a big difference as distance is measured from waypoint to waypoint.
- Variation caused by the mapping software. Oziexplorer has given me some interesting results when compared to the GPSr odometer.
- GPSr odometer is not reliable in my opinion and should be taken with a pinch of salt. I've gotten some interesting variations between the track length in mapping software when compared to the GPSr odometer.

Just some thoughts,
Michael.

[Mark F]

For all intents and purposes on a bushwalking GPS, the error will remain essentially constant in direction and distance.

I will disagree with this. It may be true on a plain but slopes and steep country reflection of gps signals from ridges and the limiting of sky view cause the size and direction of errors to move quite markedly as you move around relative to the slopes.

The main cause of error is ionospheric delay which is why surveyors use base stations and post processing systems to help remove the problem.

My own expectation is that the GPS will provide a slightly longer distance than we may expect due to the frequency of readings and associated errors, while Chapman et al working from a map will underestimate the distance due to the much larger measuring unit. It really comes down to the coastline paradox discussed above.

An interesting topic but does it really matter? When I walk I watch how I am progressing on the map and extrapolate that and terrain to come to give an eta. Distance only becomes a problem when I am racing against the big stop watch in the sky.

[photohiker]

Openstreetmap has multiple GPS tracks around major routes. If you have a close look, you will see that there is significant variation between tracks. Reasons could be the location and orientation of the GPS on the wearer, the accuracy of the GPS hardware and the location of the GPS satellites on the day and hour the track was recorded.

[JohnStrider]

So are Garmin GPS the best way to track the distance you are covering?

I use an app on my phone called Map My Hike.

[South_Aussie_Hiker]

I will disagree with this. It may be true on a plain but slopes and steep country reflection of gps signals from ridges and the limiting of sky view cause the size and direction of errors to move quite markedly as you move around relative to the slopes.

Um, nope. Youve answered your own question though - on slopes, the loss of satellite coverage is causing your problems as you move around.

If your error moves around randomly, you have poor coverage - which could perhaps be improved by
A) orienting your GPS so it gets the best sky view (on your head or on top of your pack)
B) buying a barometrically aided GPS
C) buying a unit with a bigger, higher antenna.

[South_Aussie_Hiker]

Are Garmin the best?

Well not necessarily. Garmin make high end aviation GNSS equipment, so they know what they are doing. Generally for bushwalking, the best one is any one with barometric aiding and the best antenna.

All GPS use the same calculations - it's essentially spherical trigonometry corrected for the fact that the earth is not a perfect sphere. WGS84 is the reference used for GPS.

A phone is perfectly fine given it has good coverage. In areas of poor coverage, a dedicated GPS with a better antenna will peform better.

Some GPS units are also capable of receiving other constellations, most commonly the Russian GLONASS (which incidentally, uses a different geodetic system to WGS84).

Michael_p, if you were out by 200m for a day you had poor coverage - perhaps due to terrain, your receiver location, or satellite unserviceability or a satellite "hole". When it corrected to 3m, that's because your receiver was receiving more satellites.

[tastrax]

That track is also at the bottom of gully so I suspect the signals are being bounced around a fair bit. Traces up on the spur track I reckon would be much better aligned as they have better "view to the sky"

[Maaxxx]

200 m out all day sounds a bit like a datum error but that wouldn't explain the sudden return to sanity at the end of the day.

[GPSGuided]

Take a look at your GPS walking log and there'll be lots of random movements at points where you rested. Those add up to quite a bit of additional distance. Clean up those spurious points and you'll be much closer.

[Hallu]

Nothing to do with a bad GPS, actually more to do with a precise one. To me it's clearly coastline paradox. Phone GPSs aren't that precise and should give you shorter distances than hand-held GPSs like the Garmin you have.

[icefest]

Phone GPSs aren't that precise and should give you shorter distances than hand-held GPSs like the Garmin you have.

Only if they have a longer sampling distance. If the sampling distance is the same then the more precise receiver will have the shorter track.

[slparker]

Phone GPSs aren't that precise and should give you shorter distances than hand-held GPSs like the Garmin you have.

Only if they have a longer sampling distance. If the sampling distance is the same then the more precise receiver will have the shorter track.

Is that true? Error will be greater in the GPS that uses less frequent interrogation of sattelites (presumably the phone) and that error will be to effectively shorten distance.

precision will not necessarily result in a greater distance error (or at least not by much - it is within the parameter of precision error) but will result in an error in location only(i.e. imprecision of co-ordinates). I don't have enough knowledge of how GPSs work out total distance to know for certain.

[neilmny]

............ Phone GPSs aren't that precise and should give you shorter distances than hand-held GPSs like the Garmin you have.

Assuming the precision relates to frequency of the sample this seems logical to me.
The greater spacing would "smooth" out the track and therefore shorten the overall distance.

[icefest]

I think we are discussing semantics.
Clarification (IMHO):
Precision I assumed is the likelihood of you being in the location that the GPS says you are. Rereading the wiki says this is not right.
Precision seems to be the amount of significant figures that your GPS device spews out. This is overly high with both phones and handhelds.

Accuracy
Is how far you are from your stated location. This improves with a better antenna, more satellites, better GPS device & relative complexity of internal mathemagical algorithms. This is generally less wrong in a dedicated GPS. The more accurate the GPS is the shorter the track will be for the same sampling distance.2



The other issue is sampling frequency, which is entirely explained by the coastline paradox (more samples, more distance).

[slparker]

yep, that graphic shows it nicely - but there's a few permutations here:

1. If a GPS is very accurate but has low sampling frequency the track will be shorter than actually walked (because location is precise but sampling is low - coastline effect)
2. If a GPS is not very accurate and has low sampling frequency the track will be shorter than actually walked (unless the inaccuracy has a very wide parameter, that is: the variance in each sample is many metres from actual position).
3. If the GPS is not very accurate and has a high sampling frequency the track will be longer than actually walked( because many samples will be taken of an imprecise location).

[TheGhostWhoWalks]

I had a similar issue. When analysing the tracks I noticed that I walked a km while sitting for lunch. You need to reduce the sampling rate, so that you have to move ten or more metres for the GPS to record. A stationary GPS will appear to move as it recalibrates position. This vastly improved my accuracy.

[icefest]

I had a similar issue. When analysing the tracks I noticed that I walked a km while sitting for lunch. You need to reduce the sampling rate, so that you have to move ten or more metres for the GPS to record. A stationary GPS will appear to move as it recalibrates position. This vastly improved my accuracy.

I will usually clean up my traces when I get home. Usually by deleting all the rest breaks and then applying a 5m average. I prefer maximum density for recording as I can someone's use it in the worse reception areas to see where the track is afterwards.

[Zone-5]

I simply set my Etrex 20 to log data points after a time period and not after a certain distance is traveled.

[Happy Pirate]

Just to jump in, remember also that Garmin GPSes work exclusively in the WGS84 Datum which is a generalised global spherical projection. To measure lengths and areas it has to take spherical measurements in Decimal Degrees and convert them to linear measures using a generalised global average trigonometric function.
Localised paper mapping is usually in UTMs which are optimised for linear measurements on a much more local scale.
Steve