GPSr Distance Traveled Accuracy - Odometer Versus Track

[+Kohavis]

I have been using my GPSr for keeping track of how many miles I walk on cache hunts. To me, the exercise is almost as important as the hunting itself, but not quite )

 

For about a month now I've been just relying on the autotracking feature of my handheld. When I get back from my hunt, I download the track into Mapsource (I have an eTrex Venture HC). My GPSr was set up to record a track point every 300 feet. I had a nice list of tracks, dates. places, and miles walked

 

Yesterday I went on what I thought was going to be a 5+ mile hunt in the bosque. I reset the odometer before leaving the parked car. About 4 miles into the hunt (I went after 11 caches) I started watching the miles I was racking up. The hunt was through heavily forested area along the Rio Grande. Eventually it was showing 5.6 miles, then 5.8 and so on. I thought "Great! This is the most walking I've ever done on a cache hunt. It finally totaled 6.39 miles when I had gotten back to the car.

 

When I got home I downloaded the track to Mapsource on the PC and it showed 4.9 miles traveled A respectable distance, but still far short of what the odometer showed. I knew it wasn't the unit's error, because the same handheld was recording the odometer data and the track data. It must be something I was doing wrong. That was a 30% error

 

I thought about it for a while and it occurred to me - DUH!!! The autotrack was set to taking one point every 300 feet. It takes a measurement, then waits until you've traveled 300 feet and records the next point - a perfectly straight line from the previous point. When has anyone been able to travel through the woods in a straight line? Never.

 

I pulled up the Google maps satellite image of one segment of my hike, then did a quick estimate of how I had probably walked from one point to the next, judging by the terrain:

 

 

Then I plotted what the autotracks had probably recorded for that same leg:

 

 

Of course. There's quite a bit of deviation in any walk through the woods - avoiding holes, trees, bushes, mud, etc.

 

So today I changed the interval setting to 50 feet (the lowest distance interval on the Venture HC) and ran a little experiment. I took the dog for a walk. I grabbed the leash, reset the odometer, and away I went. We walked primarily on sidewalks, with a small detour to check on one of my nearby caches, which was hard-pack dirt in an empty lot. I watched the odometer as we went, then checked the total as well as the track distance when I got home.

 

Odometer total:

 

 

 

Track distance:

 

 

There's still an 11% error in the distance. What bothers me is that it was on sidewalks most of the way, with little deviation. It would probably be worse in the woods.

 

The only thing I can figure is that 50 feet is still too long for the interval. I can set it to time intervals down to one second, although 5 seconds would probably be adequate. I think I'll repeat the experiment tomorrow with the new setting. But I'm bothered by the fact that the discrepancy between the odometer and the track was so large - 0.12 miles on a pretty straight path. That's 600 feet or so! I don't imagine the unit does any rounding on the track points, but I can't be certain. I do know that the autotrack feature is more direction-centric than distance-centric. Its purpose is to let you backtrack your steps - like a breadcrumb trail. Perhaps it does round up or down as you go along ...I don't know.

 

Does anyone know more about this? If the error is inescapable, I'll merely record the odometer distance and add it to the track title in Mapsource for archiving. That's at least one workaround for it.

[+trainlove]

There has been much said of GPS odometers.

check http://forums.Groundspeak.com/GC/index.php...=195212&hl= for the most complete.

 

There are also some posts about the spedometer accuracy on GPSr's.

 

Some people think a reading, each second, has +/-10 meter accuracy so slow speeds are bubkis.

Others think that's not the case.

Both mighjt be right, LOL.

But speed, and distance between points will be dependant on how you go from A to B, even 50 feet is never a straight line. Try other settings, like auto detailed, or time based settings if your can do them.

[+Kohavis]

Thanks trainlove. I guess the Vista error in that thread is likely what I'm seeing, since the models are so closely related.

 

I'm not concerned about the speed, just the total distance.

 

I'll give the time-based intervals a try tomorrow and post my results here

[+julianh]

For what it's worth, on my Summit HC, I generally have my track-log set to "Auto - Normal", rather than "Distance" or "Time". The Summit HC has a current track-log limit of 10,000 points, so I always want to get "best value" of my available track-log memory. (This is less of an issue on units with expandable memory card option, where you can have effectively unlimited track-logs.)

 

In "Auto" mode, the unit stores a track-point at a variable time / distance interval, depending on your movement. It will put down quite closely spaced points when you are wandering around, to maintain a fairly faithful record of your perambulations, but when you are driving on a straight highway, you may find you only get a point every few hundred metres or so. The 10,000 point limit will usually give me 3 to 5 days of records before it starts to over-write itself. Works for me.

 

If you want even more detailed track-logs, you could try "Auto - More Often" or "Auto - Most Often". If you are trying to stretch out your track-log (e.g. to record a couple of week's movement without access to a computer for downloading tracks), you might try "Auto - Less Often" or "Auto - Least Often". I haven't played with these options much, so I can't tell you how much impact this has on track-point spacing. However, when I am using the unit for work, and need the best possible detail, I use "Auto - Most Often", and downlao9d the tracklog to my computer at the end of each day.

 

Hope this helps!

[+EraSeek]

Generally the odometer is less accurate than the track log. My odometer over-estimates.

 

However, the photo you have is of a "saved" track. Saving a track edits the track points to I think 500, and therefore the track accuracy as well.

 

You can get passed this by either not saving the track and thus keeping the track in the unit as an active track (you can turn the track log on and off on the "Track Page") or downloading the track record directly from the SD card to mapsource on your computer and viewing it there. The sd card does not edit.

 

Anytime you want to record a certain track section independantly turn the tracking off and back on before and after the section on the "track Page" (not on the card page, which you should just leave as "record to card"), this defines one active track section from another.

 

Any track record can be inaccurate due to conditions and how you have your setup. Generally I leave mine on Auto/most. But sometimes when in the open and wanting the best record I use Time/1sec. If you are in the woods you will get multipathing of the signal so if you have it set too tight- say time/1 sec or distance/30ft or auto/most- your 3 mile hike may record as a 4.5mile hike. Keep that in mind and try and set it for the conditions as best you can.

[+Kohavis]

@julianh - Thanks. That does help. I have the same internal memory you have, so your settings are worth a try. It'll take trial and error, I'm sure. Up until now I've been using the manual settings for track point intervals. I'll give auto a try next time

 

I have daily access to a PC for my daily hikes, so I don't have to worry about the track log wrapping around again. 10,000 points is plenty for one day

 

@EraSeek - The photo I took was of the saved track, but I did that to display it on the unit's map for the grab. You have to save it to show it on my unit. I had downloaded the track log to Mapsource prior to saving, however. I never save the logs in the unit. I prefer to d/l the raw log and then clear the track logs in the handheld for the next trip.

 

My unit has no card capability.

 

I hadn't considered multipath - good point. 1 second intervals seems like overkill, since I don't walk that fast

 

I suspect I'll need to play around with it - maybe try a different setting every time I go out, and see which one is most accurate for my hiking style. It's nice that they give us all these tweakable settings, but sometimes it's daunting getting it set up just right.

 

Again, thank you for your valuable inputs. There's nothing like tapping into someone's experience to make the learning curve easier to get up

 

I tried the time interval - 5 seconds setting today. Took the dog on almost the exact same path, and opur walking time was about 20 minutes. This time the odometer reported 0.99 miles, and the Mapsource track log (245 points) reported "0.9" miles for the total. The track log in Mapsource only has one digit to the right of the decimal point. I sure wish I could set the units to feet instead of miles

 

Not knowing if and how the figure is rounded, or if they just dropped the least significant digit, it could be 0.9, 0.949 or 0.99 being represented as "0.9" in the track log. I don't want to have to take the poor dog on a 350 mile walk to get statistically meaningful data I think I'll manually add up the feet for each of the segments and try to determine the exact feet traveled. Each segment shows feet as the units instead of miles, so there's still a chance I can get a more accurate total

[+Klemmer]

One general tip, applicable to most units, certainly the three I own:

Be sure you have a good solid GPS signal lock (multiple good sats), BEFORE you reset or turn on your odometer or track log. During the initial stages of a unit locking up, it can at first think you are .2 mi or more away from where you really are. As the lock-up process continues, the position improves. This can easily skew your results, usually to the high side. On both an older Garmin & Magellan, as well as my current Delorme PN-20, this can happen easily.

 

In summary: get a good lock first, THEN turn on or reset odometer / track.

 

Of course, as explained well above, there are other issues. But in my experience, good lock before resets minimizes the problems. I only see a few %, maybe 10% max, differences.

[+Kohavis]

Good thing to remember. In fact, I had fired up the unit before the walk and since it was set to auto-track every 5 seconds, there were a number of points showing as the error circle got smaller, with a resulting track I was standing in the front yard with a clear sky. I waited until the error circle had dropped way down before resetting the track log and the odometer again I didn't set out until it had a good lock, and the error was about 6 feet.

 

Strangely, the handheld shows two decimal places in the track length, but Mapsource shows only one.

 

I added up the total feet of all the legs of yesterday's and today's walks:

 

Yesterday

 

Odometer distance reported: 1.13 miles

Mapsource track distance reported (50 ft. intervals, 98 points): 1.0 miles

Sum of each track leg (feet): 5321 ft. = 1.0078 miles

 

Today

 

Odometer distance reported: 0.99 miles

Mapsource track distance reported (5 sec. intervals, 245 points): 0.9 miles

Sum of each track leg (feet): 4968 ft. = 0.9409 miles

 

So either Mapsource is rounding down, or it's simply dropping the extra digits to the right of the decimal. It also looks like the odometer is slightly optimistic. I believe that could be from the same issue EraSeek touched on - multipath. Since the odometer is "continuously" updated, it may see signal jitter as actual movement.

 

It's not that big of a deal. I ran through all these monkey-motions because I'm an engineer and I don't like mysteries or discrepancies in my equipment. But the bottom line is that it has little impact on anything. How sore my feet are at the end of a long hike is probably a more accurate odometer

 

I think I'll try what julianh suggested and test the auto settings (combination dist. & time) with various point recording interval settings. I'll start with the most frequent and see how that turns out. If there's no improvement, I could also just leave the track interval at 50 feet and split the difference between the track log and the odometer when I get home. I'll see how it goes with the auto settings.

 

Thanks again for the advice!

[+Team Chinook]

checkout SportTracks. I use it to track exercise and I upload my current track of my Colorado directly into the software. It shows satellite or terrain or streetlike maps to see where you've been.

[+Kohavis]

checkout SportTracks. I use it to track exercise and I upload my current track of my Colorado directly into the software. It shows satellite or terrain or streetlike maps to see where you've been.

Thanks. I'll do that. I hadn't considered a different app for it, but it makes sense

[+Kohavis]

I think I'll try what julianh suggested and test the auto settings (combination dist. & time) with various point recording interval settings. I'll start with the most frequent and see how that turns out. If there's no improvement, I could also just leave the track interval at 50 feet and split the difference between the track log and the odometer when I get home. I'll see how it goes with the auto settings.

As promised, I tried this setting for the autotrack - "Auto - time & distance" and "Most often" for the frequency of points. I hold my GPSr in my right hand when walking and frequently cycle between the compass/bearing, map and odometer displays. I also have a tendency to swing my arms a little when walking. I made a conscious effort to try and hold the GPSr still at my side when walking.

 

Results:

 

Track as displayed in mapsource:

 

 

Autotrack - 5.1 miles

Odometer - 5.67 miles

 

Error: 10%

 

It appears that I'll never get better than a 10% disagreement between the two metrics. I'll have to use the lower of the two numbers for my record keeping.

 

@Team Chinook - I installed SportTracks. I like it! It provides some nice statistical data and makes more sense than using Mapsource to track my hikes. Thanks for the tip

[+Kohavis]

Good news, I hope. I checked the Garmin site for unit software updates and the version 2.8 software for the Venture HC says that among the "fixes" is "improved odometer distance measurements". Maybe that'll fix the error

[+Kohavis]

One last post before I let this topic sink into oblivion.

 

The firmware update fixed the discrepancy!! I'm using ver. 2.8 of the unit firmware now and all is good.

 

I did a cache hunt hike yesterday. The autotrack for the trip reported 5.7 miles and the odometer reported 5.69

 

 

 

 

So if you're having the same issue, try updating the firmware from Garmin

Edited October 30, 2008 by Kohavis