Garmin 62S accuracy?

[yvon]

What about the accuracy between the 62S display, on distance, elevation etc what we see on Basecamp when we download the tracklog? They supposed to correct this on an update, but it still off. Somebody have info on this?
Despite this gap in information, I still love my 62S

 

[DougPaul]

I don't understand the question--do you have a reference describing the problem in more detail?

FWIW: the nominal accuracy of a consumer GPS is 95% probability of being within 10 meters (33ft) horizontally and typically about 20 meters (66ft) vertically (GPS altitude, not barometric altitude), assuming a good skyview and proper GPS orientation. Garmin does not appear to have published a spec for the vertical accuracy for the 62s (with its barometric altimeter), but the spec for the (presumably similar) 60CSx is "accuracy +/- 10 feet, resolution 1 ft, subject to proper calibration".

Of course, bugs in the GPS or computer software can do anything to the accuracy of the data as reported to the user...

Doug

 

[yvon]

Thank you Doug
This is an exemple from last week:
Last Friday we went hiking. When we returned to the parking lot, I looked at my 62S and indicated that we did 10.95km (6.8mi). Arrived at home, I download my tracklog and opens it in Basecamp it indicate that we did 9.5km (5.9mi) The total ascent have also a difference +/- 50ft. I have this problem since I have my 62S and all my friends who own a 62S have the sane problem. Garmin is aware of this problem since the 62S went on the market. Garmin made an upgrade, which was supposed to solve that issue. But this problem has returned or has not been set. (I did not used my GPS for +/- a year 62S)

 

[DougPaul]

Thank you Doug
This is an exemple from last week:
Last Friday we went hiking. When we returned to the parking lot, I looked at my 62S and indicated that we did 10.95km (6.8mi). Arrived at home, I download my tracklog and opens it in Basecamp it indicate that we did 9.5km (5.9mi) The total ascent have also a difference +/- 50ft. I have this problem since I have my 62S and all my friends who own a 62S have the sane problem. Garmin is aware of this problem since the 62S went on the market. Garmin made an upgrade, which was supposed to solve that issue. But this problem has returned or has not been set. (I did not used my GPS for +/- a year 62S)

This is normal.

The GPS internal trip statistics are computed from the 1/sec locations and BaseCamp only sees the tracklog (which is a subset of the 1/sec locations). It is a trivial mathematical proof (using the triangle inequality) to show that the distance computed from the tracklog must be less than or equal to the distance computed from the 1/sec positions. (The 60 series and the 62 series tracklogs are limited to 10K points so the tracklog can only store 2.7 hrs of 1/sec data and thus one usually records points less often.)

Any "fix" to BaseCamp would actually produce an incorrect result given the data that it sees. (I presume that a "fix" would be to multiply the computed distance by a fudge factor slightly greater than 1. You can look at your past trip data and estimate an average fudge factor yourself. Unlike a Garmin-supplied fix, you will at least know what you have done to your data. Note: the fudge factor will depend upon the track recording method and frequency. (main menu>setup>tracks>{record method, recording interval}. FWIW, I generally use Auto, Normal (the default).)

BTW, the difference that you observe in the above example is a bit larger than I usually observe, but it will vary from trip to trip and, as noted above, also depends on the recording method and interval.

Doug

 

[yvon]

Thank you Doug, I am trying to understand your informations. I use the same recording method/interval. Which statistics is wright, the one I see on the 62S or on Basecamp?
For a long trek, ex. 2 weeks, do I use the same recording method/interval (auto-normal)?

 

[DougPaul]

Thank you Doug, I am trying to understand your informations. I use the same recording method/interval. Which statistics is wright, the one I see on the 62S or on Basecamp?
For a long trek, ex. 2 weeks, do I use the same recording method/interval (auto-normal)?

Trail distance is a fractal and has no unique correct value. The distance traveled depends on how you measure it.

So neither statistic is more correct than the other, they are simply different.

Doug

 

[yvon]

Thank you Doug

 

[TomD]

Hmm

Trail distance is a fractal and has no unique correct value. The distance traveled depends on how you measure it.

So neither statistic is more correct than the other, they are simply different.

Doug

Doug, I'm not sure I understand your comment. If I walk a straight mile on flat ground, I presume I have walked a mile, no matter how I measure it. What do you mean by fractal? I can see point to point on a map being different from actual distance walked if there are curves on the trail plus ups and downs that I think wouldn't show on a gps track, but not sure if I am right. Thanks
Ps. I see articles on the net on fractal distance estimation, but they are way beyond my puny math skills.

 

[bikehikeskifish]

If you measured the center of the trail with a 12" ruler, you'd get one distance. If you walked behind yourself measuring with a yardstick, you'd get a different number. Same with the time samples of the GPS. Both are examples of resolution - higher resolution (smaller measuring intervals, but more of them) means better accuracy.

Tim

 

[DougPaul]

Doug, I'm not sure I understand your comment. If I walk a straight mile on flat ground, I presume I have walked a mile, no matter how I measure it. What do you mean by fractal? I can see point to point on a map being different from actual distance walked if there are curves on the trail plus ups and downs that I think wouldn't show on a gps track, but not sure if I am right. Thanks
Ps. I see articles on the net on fractal distance estimation, but they are way beyond my puny math skills.

A classic example is measuring the length of a seacoast by placing various length rulers end-to-end and counting the number of rulers required to go end-to-end of the coast. The total measured length is the number of rulers times the length of a ruler. If you start with 10 mile long rulers, you will skip over many smaller features. As the rulers get shorter, the total length will increase because you will be following smaller and smaller features.

A straight path or road is a special case for which the measured length is independent of the length of the ruler. Any turns of any size break the special case.

Trails, in general, have enough turns and detail (down to stepping around rocks,trees, and mudpits) that they must be considered to be fractals.

From another viewpoint, think of a giraffe, human, and an ant walking the same trail. The giraffe will be able to avoid many of the details, the human will able to avoid some of the details, and the ant have to deal with all of the details.


BTW, this question as been asked a number of times--just search on the word "fractal" to find a number of previous threads.

Doug

 

[TomD]

Will do, thanks.

 

[JCarter]

Apologies for the grave-digging, but I thought folks interested in the accuracy of distance reported by GPS would be interested in this article on the IEEE Spectrum website:

http://spectrum.ieee.org/tech-talk/at-work/test-and-measurement/why_every_gps_overestimates_distance_traveled

 

[iAmKrzys]

Apologies for the grave-digging, but I thought folks interested in the accuracy of distance reported by GPS would be interested in this article on the IEEE Spectrum website:

http://spectrum.ieee.org/tech-talk/at-work/test-and-measurement/why_every_gps_overestimates_distance_traveled

Interesting article, but in a sense it states some of the obvious limitations of GPS:
1) Any device that reports location with non-zero error along a straight-line will over-report the total distance;
2) If you leave your gps turned on in one location for a couple of hours you will discover that it will claim to have moved quite some distance while you know it didn't move an inch.

While gps may be measuring location very frequently, in practice only few of these points might be getting reported in the track file and then the track might show shorter total distance when moving around a curve. Consider a following hypothetical situation: your gps records a track point exactly every 30 seconds and you pace yourself running 90 yards around a circle such that you make full circle exactly every 30 seconds. From your gps track it will appear that you are roughly staying about the same spot while you might run quite some distance over time.

 

[DayTrip]

This is normal.

The GPS internal trip statistics are computed from the 1/sec locations and BaseCamp only sees the tracklog (which is a subset of the 1/sec locations). It is a trivial mathematical proof (using the triangle inequality) to show that the distance computed from the tracklog must be less than or equal to the distance computed from the 1/sec positions. (The 60 series and the 62 series tracklogs are limited to 10K points so the tracklog can only store 2.7 hrs of 1/sec data and thus one usually records points less often.)

Any "fix" to BaseCamp would actually produce an incorrect result given the data that it sees. (I presume that a "fix" would be to multiply the computed distance by a fudge factor slightly greater than 1. You can look at your past trip data and estimate an average fudge factor yourself. Unlike a Garmin-supplied fix, you will at least know what you have done to your data. Note: the fudge factor will depend upon the track recording method and frequency. (main menu>setup>tracks>{record method, recording interval}. FWIW, I generally use Auto, Normal (the default).)

BTW, the difference that you observe in the above example is a bit larger than I usually observe, but it will vary from trip to trip and, as noted above, also depends on the recording method and interval.

Doug

I have this same model and have noticed in the past 4-6 months that my Base Camp distance matches very closely, if not exactly, my Trip Computer now. In the past the Base Camp distance was always less and closer to the map and guide book distances (which I know is "normal" as you detailed in this and other posts). I haven't been updating my Base Camp since the Windows 10 compatibility issue I had a while back so I don't know if there have been recent updates addressing this. I have not changed any settings in years.

Could this be attributed to carrying a SPOT messenger now and maybe the signals interfering with location and adding distance? That is the only thing I can think of that I'm doing different. I keep the units 2-3' apart generally except maybe on some stops.

 

[DougPaul]

2) If you leave your gps turned on in one location for a couple of hours you will discover that it will claim to have moved quite some distance while you know it didn't move an inch.

If you analyze the track recorded by a stationary GPS, it will generally show the measured position to stay within a fairly small area (eg within a radius of 10-20 meters of the average position). (FWIW I have done this with a 12 hr track. The satellite positions repeat every 12 hours.)

While gps may be measuring location very frequently, in practice only few of these points might be getting reported in the track file and then the track might show shorter total distance when moving around a curve. Consider a following hypothetical situation: your gps records a track point exactly every 30 seconds and you pace yourself running 90 yards around a circle such that you make full circle exactly every 30 seconds. From your gps track it will appear that you are roughly staying about the same spot while you might run quite some distance over time.

The loss of information caused by an insufficient number of samples is called aliasing.

The typical consumer GPS computes the location once per second and, as you note, the recorded track is usually a subset of these locations. Garmin hiking GPSes have three modes for selecting the subset:
1) timed with settable times (including 1/sec which saves all the locations)
2) distance with settable distances (every time you move a specified distance from the last point).
3) auto (some combination of time and distance)

The auto mode may be able reduce the aliasing by allocating more samples when one deviates from a straight line.

One can prove mathematically that the length of a subset track is less than or equal to the length of the full (1/sec) track. (Proof by the triangle inequality.) The trip computer is driven by the 1/sec locations, not the saved track.

Doug

 

[DougPaul]

I have this same model and have noticed in the past 4-6 months that my Base Camp distance matches very closely, if not exactly, my Trip Computer now. In the past the Base Camp distance was always less and closer to the map and guide book distances (which I know is "normal" as you detailed in this and other posts). I haven't been updating my Base Camp since the Windows 10 compatibility issue I had a while back so I don't know if there have been recent updates addressing this. I have not changed any settings in years.

Could this be attributed to carrying a SPOT messenger now and maybe the signals interfering with location and adding distance? That is the only thing I can think of that I'm doing different. I keep the units 2-3' apart generally except maybe on some stops.

Probably not.

If the units are interfering with each other, it is more likely to show up as loss of lock or large position errors.

Doug