Garmin GPS and the distance "issue"

[Framerman]

I bought a Montana last year. Don't get me wrong, I like it (minus the gargantuan size). There seems to be one issue that has bugged me to death, and from the searching on the internet and the many emails back and forth from their support, I have no answer to the question of inaccurate distance data on the unit.

Well, I have an answer, just not one that is very acceptable to me for a GPS unit that costs this much. The data it documents runs on one of two options, by time, or by distance. Meaning, it will take a point based on your desired setting. The distance setting is just plain dumb. The best one can set it at is 0.01 miles, which is (obviously) over 50 feet. So, the generally recommended setting is one point per second, which is no issue at all to me for storage.

The issue comes up when I am expecting somewhat of an accurate overall distance reading on the trip odometer. What was explained to me from Garmin is that if you take a break (say a half hour, or even a combined half hour from multiple stops), the unit (obviously) still takes reading. We all know that GPS units aren't really all that accurate, being within 10-20 feet roughly at best. But what Garmin does is accumulate these 10-20 foot distances while you are standing still and (of course) adds them all up. I can sit there in 5 minutes and see this thing ticking off 0.03 miles easily. While that doesn't seem like much, the going error they state is around 15% off, which at this point, I have unfortunately incorporated into my mental notes for the ACTUAL distance I have gone.

Where it is a minor annoyance to me is how far I have left, or how far I have gone (which I know some really don't mind even knowing this in the first place). If I take that file it creates which says 11 miles and put it in to Base Camp, it immediately says something like 10.4 miles. Now why on earth don't they do that right on the unit? My argument to the person I was conversing with through email is that free phone apps have the setting that says "don't record if not moving", or something along that line. Just don't get why they can't integrate that right in.

I guess my question is, do the other GPS units do this as well? Like Delorme? I have heard good things about this GPS (as I should, since I live in Maine), but they don't have any large format screen (old eyes having an issue seeing the screen)

 

[DayTrip]

I believe they do (I also have a Garmin). There are quite a few threads on VFTT regarding this topic (I asked a similar question when I got my GPS and was chastised for not researching). Does seem stupid. Just a casual observation with mine but it seems to have less of an error if I do not carry it in a pocket. Recently I started clipping to shoulder strap of backpack and noticed the error is somewhat lower. I did a 14.3 mile (actual) hike this Sunday with unit clipped to shoulder strap and took several lengthy stops (15-25 minutes). The trip odometer showed me at about 15.1 miles. When I used to carry in my pocket it would easily be 16+ miles.

 

[DougPaul]

I believe they do (I also have a Garmin). There are quite a few threads on VFTT regarding this topic (I asked a similar question when I got my GPS and was chastised for not researching).

Correct--this question has been asked and answered many times...
A few key points:
* There is no unique correct distance. It is actually a fractal and depends upon how you measure it. The more ways you measure it, the more distances you will get...
* Consumer GPSes produce 1 position/second. This drives the trip computer. (The recorded track is a subset of these positions. Any distances computed from the track can be different--most likely smaller.)
* GPSes decide whether you are moving or stationary by your speed and distance moved in the recent past.

For more detail, use the search. Hint: A search on the word "fractal" will bring up many relevant threads.

Does seem stupid. Just a casual observation with mine but it seems to have less of an error if I do not carry it in a pocket. Recently I started clipping to shoulder strap of backpack and noticed the error is somewhat lower. I did a 14.3 mile (actual) hike this Sunday with unit clipped to shoulder strap and took several lengthy stops (15-25 minutes). The trip odometer showed me at about 15.1 miles. When I used to carry in my pocket it would easily be 16+ miles.

A number of naive users carry the GPS in this way...

For maximum accuracy, a GPS needs to be able to receive as many satellites spread as far around the sky as possible. Flesh absorbs the signals so if you carry the GPS in your pocket, you block out much of the sky and some of the satellites which reduces the accuracy.

Some GPSes work best if held vertically, some face up horizontally. The modern high-sensitivity units are good enough that they will usually do pretty well in any orientation. I usually carry mine in the top of my pack to maximize the skyview without the GPS being exposed. (The signals go though dry packcloth without difficulty.) Sometimes I carry it in a small pocket attached a shoulder strap (but the first location is probably better). If I want the highest accuracy (e.g. for mapping a trail), I put an external antenna in the top of my hat. Or I could just carry the GPS (properly oriented, of course) in or on my hat.

Doug

 

[Framerman]

My apologies, I did search but gps can't be included in searches because of the length . I tried garmin and error for keywords.

I carry more in an exterior pocket on my pack, which is located high up.

 

[Mohamed Ellozy]

... I did search but gps can't be included in searches because of the length ...

Old trick, try gps*

 

[DougPaul]

My apologies, I did search but gps can't be included in searches because of the length . I tried garmin and error for keywords.

The internal search function does not index 3 letter words. Use Google advanced search limited to this site.

Searching on the word "GPS" alone is probably too general--the word appears in many threads, most of which have nothing to do with distance measurement. It might be ok in combination with other search terms. (Again, search on the word "fractal". It generally appears only in the context of trail distance measurement on this website.)

I carry more in an exterior pocket on my pack, which is located high up.

Perhaps not quite as good as in the very top of your pack, but it should be good enough. The more of the sky that has a direct line to the GPS, the better. The signals (19cm wavelength) can diffract around objects or reflect off objects (both can cause the dreaded multipath) to get to the GPS, but the accuracy will suffer.

Doug

 

[Remix]

The older Magellan Meridian series had good algorithms to pin--meaning if the unit determined you were stationary, your reported position and track were rock solid. All the Garmins I have owned, with the exception of CAR GPS's, are very noisy when you are stationary. The track ends up looking like a dust ball if you are not moving and the odometer increases with time.

 

[DougPaul]

The older Magellan Meridian series had good algorithms to pin--meaning if the unit determined you were stationary, your reported position and track were rock solid. All the Garmins I have owned, with the exception of CAR GPS's, are very noisy when you are stationary. The track ends up looking like a dust ball if you are not moving and the odometer increases with time.

There is no 100% reliable way for a GPS to differentiate between slow movement in a small area and and being stationary. As such, the manufacturer's have to guess at some heuristics (ad hoc algorithms) and thresholds which result in different characteristics--the Magellans in question appear to have a higher threshold for separating the two states than Garmin. This may result is a prettier, but less accurate track which skips zones of slow movement (ie data is thrown away). In practice, the fuzz balls may not be pretty, but they do tell the knowledgeable user what happened.

There used to be a similar issue with what to do if the GPS lost satellite signals. Some (Magellans, IIRC) would extrapolate (ie guess) the track while the Garmins would declare a loss of signal. However, the extrapolated track was often wildly inaccurate... This also led some users to erroneously think that these GPSes were more sensitive than the Garmins.

Doug

 

[Remix]

There is no 100% reliable way for a GPS to differentiate between slow movement in a small area and and being stationary.

There is a completely reliable way for a GPS to differentiate between no movement and motion...for whatever reason Garmin decided not to implement it on many of their hiking GPS models...leaving a knowledgeable user to realize they claim a superior sensitivity at the expense of presenting more noise to the user....the incorrect odometer readings and dust ball tracks being the evidence...

 

[nartreb]

Yeah, it's not rocket science. Any "movement" that's smaller than the accuracy of the unit (given the constellation visible at that moment) should be ignored for tracking purposes. The user who does a ten-mile hike and is annoyed that the unit displays grossly inaccurate mileage, so that he has to download the raw data and apply a smoothing algorithm in order to see how far he really walked, is the typical user of a hiking GPS.
The user who's disappointed that he did a thousand circuits of a three-foot circle but still sees a mileage total of zero is somebody who has no need for GPS in the first place.

 

[hikerbrian]

Interesting. I recall reading various reviews of smart phone-based GPS's vs. Garmin (on this board and others) where both units tracked a particular route and the mileages were compared at the end. In nearly all cases, the Garmin mileages were longer, leading the reviewers to conclude the Garmins must be more sensitive. Seems like that's the wrong conclusion. Thanks for posting!

 

[DougPaul]

Interesting. I recall reading various reviews of smart phone-based GPS's vs. Garmin (on this board and others) where both units tracked a particular route and the mileages were compared at the end. In nearly all cases, the Garmin mileages were longer, leading the reviewers to conclude the Garmins must be more sensitive. Seems like that's the wrong conclusion. Thanks for posting!

I haven't compared the distances, but I have compared the tracks recorded by a Garmin 60CSx and an iPhone (4, I think). The iPhone track had a number of large excursions from our path, the 60CSx track did not. This suggests that the 60CSx is more accurate than the iPhone, but says little about the sensitivity.

Doug

 

[Remix]

Interesting. I recall reading various reviews of smart phone-based GPS's vs. Garmin (on this board and others) where both units tracked a particular route and the mileages were compared at the end. In nearly all cases, the Garmin mileages were longer, leading the reviewers to conclude the Garmins must be more sensitive. Seems like that's the wrong conclusion. Thanks for posting!

Yes, the Garmin unit I used would "accumulate" the noise and add miles to the odometer, whereas the Magellan would reject noise until it detected a new position that was greater than the maximum possible error for its own receiver and GPS dilution of precision-a theoretical error related to the gps receiver to satellite geometry. Someone published some actual data and I did post it here. The OP's point about being able to choose behavior was a good one.

As I recall the general conclusion was not to put a lot of trust in any GPS odometer--no matter which brand--since all brands accumulate noise while moving. However, accumulating mileage from noise while not moving makes things worse.

 

[Raymond]

Last time I was on the summit of Bondcliff, I changed the batteries in my 60CSx and reset the computer page so that everything but the time of day was zero. As I set the unit down on a rock next to me, I noticed that the trip odometer advanced to 1 foot, and I remember thinking that that was probably about right. Then I ate my lunch.

When I readied to go, 10-15 minutes later, I noticed that the trip odometer still read 1 foot! Can’t complain about accuracy there, although I probably can whenever the trail goes up or down in elevation. The track distance, which usually reads less (a shorter distance) than the trip odometer when walking around my neighborhood, seems to be more accurate in the mountains (a longer distance, more like what the books or signs say the distance should be) than the trip odometer.

 

[madmattd]

In a total reversal of what most everyone else seems to be seeing, my Garmin (Etrex 20) consistently under-reports the mileage compared to the AMC published mileage, usually between 5-7%. I can count on one hand with fingers remaining the number of instances in the last 3 years of hiking (nearing 1300 miles) where the track's recorded mileage was more than the AMC mileage. This is when viewing the track on the computer after downloading. Most of the time I forget to reset the trip computer before starting a hike, so I can't speak to the difference between the trip odometer and computer track distance. I've also on a few occasions gone and removed the extra points from the mothballs where there were longer breaks and it doesn't seem to affect the end distance much. Perhaps the computer software (BaseCamp) accounts for this in displaying the total distance?

 

[DougPaul]

In a total reversal of what most everyone else seems to be seeing, my Garmin (Etrex 20) consistently under-reports the mileage compared to the AMC published mileage, usually between 5-7%.

The Etrex 20 and the AMC are using different methods of estimating the distance and can be expected to get different answers. And as been explained earlier and elsewhere--there is no fundamentally correct distance for trails. That said, one may be able to notice systematic differences between two different methods (but this does not mean that one is necessarily more accurate than the other).

Perhaps the computer software (BaseCamp) accounts for this in displaying the total distance?

The GPS trip computer runs off the 1 per second PVT (position, velocity, time) output of the basic GPS receiver--the computer software runs off the tracklog which is usually sampled far less often than 1 per second. Different data produces different results. (Fewer data points for the same trip generally produces shorter distance estimates.)

Doug

 

[Framerman]

Most of the time I forget to reset the trip computer before starting a hike....

That took me awhile to figure out as well. I went on a few hikes and it said something akin to 60+ miles and I swore up and down that I reset to zero, which I did. I found out later that I HAD to start moving, watch the odometer click into the exorbitant mileage, THEN reset. It's a minor annoyance, same with the trip odometer. I have adjusted to suit, and honestly am relatively pleased with the unit. I'm not making maps for sale or anything, but I do have a little OCD about things like wondering how much longer my knees are going to ache for

I have also tried comparing my SG3 and the Montana, and it's obvious neither are ever going to be super accurate in the first place. I keep my battery well charged on my phone just in case the Montana goes. Not that I NEED a GPS to get out, but it's still nice to know how much further I need to go.

DougPaul, thank you...I had no idea the terminology I was looking for (fractal). Some days I'm just brain farting all day and can't seem to figure things out. I was certain that this topic had indeed been discussed to death, but the search terms I was using obviously was not very well thought out.

 

[DougPaul]

DougPaul, thank you...I had no idea the terminology I was looking for (fractal). Some days I'm just brain farting all day and can't seem to figure things out. I was certain that this topic had indeed been discussed to death, but the search terms I was using obviously was not very well thought out.

Sure--happy to help.

There is no way for you to have known to search for the word "fractal". (Unless you were a very good and lucky guesser...) I only knew to suggest it because I had taken part in many of the threads--the word probably only appears on this website in the context of measuring the length of a trail.

I suggest that you make a reasonable attempt to find what you seek (Google advanced search is more versatile than the internal search) and if that fails, post a question explaining that you have tried and failed and would like some help finding the info or the info itself.

Answering a question takes time and effort and those answering are often more willing to help if it is clear that you made a reasonable try before asking your question. Also if you have a piece of hot news, it is worth checking before posting to see if someone else has already started a thread on the topic--multiple parallel threads are messy.

Doug

 

[DougPaul]

Yeah, it's not rocket science. Any "movement" that's smaller than the accuracy of the unit (given the constellation visible at that moment) should be ignored for tracking purposes.

Maybe not, but it is still far harder than those who have not tried it may think... (Try to write a program to do it. Remember the GPS cannot see the future, so you can only use past and present track points (ie no future points) to decide whether the present track point is moving or not. Postprocessing an entire track (what we do when looking at a track) is much easier. Most of the track points will be fairly easy to classify, but there will be some percentage of troublemakers.)

The errors in the position and velocity are probabilistic and the different sources of error change on time scales ranging from shorter than the measurement interval (1 sec) to hours. (The typical error spec of 10m for a consumer GPS means that the error is less than 10m with 95% probability.) Thus any threshold will make wrong decisions for some range of physical speeds.

The user who does a ten-mile hike and is annoyed that the unit displays grossly inaccurate mileage, so that he has to download the raw data and apply a smoothing algorithm in order to see how far he really walked, is the typical user of a hiking GPS.

I have actually tried smoothing on the altitude track to get a good estimate of the total climb and found that I could get any of a wide range of answers depending upon the degree of smoothing. Try it yourself--you may be surprised...

Solving both of these problems is simple in concept but difficult in practice. (People also used to think that image analysis and speech recognition were easy too, but when they actually tried to write programs to perform these tasks they found them to be very difficult. FWIW I spent a good part of my career developing algorithms for speech recognition.)

Doug