GPSrChive Discussion Forum

In another thread the topic of Automatic> Most Often track recording came up:

gpsrchive wrote: Mon Dec 28, 2020 2:54 pm I suspect you had the Garmin recording controls set to Automatic > Most Often. If you configure the Garmin recording for Time > Once every Second, You will likely get a tracklog that is just as detailed (or more) than the watch.

gpsrchive wrote: Mon Dec 28, 2020 4:48 pm Be aware, when set to short time intervals such as this, if you spend a lot of time stationary it can result in 'spider-web' tracks that require some cleaning up after the fact.

And on other sites this has been discussed a bit with contradictory observations. Many say that they see more accurate tracks using Time> Once Every Second rather than the Automatic> Most Often setting. But some say just the opposite:

If you do slow activity (f.e. hiking), I found out that smart recording is more precise because it will reduce GPS deflections.

And the reply:

I have no idea why this should be the case, but I have found it to be so as well. I wonder if there is some element of error-correction going on in smart recording that is absent from 1s recording. If the reception is not great (trees/buildings etc) a gps walk with 1s recording is guaranteed to give you a track with the sort of zig-zags I don't usually see on smart. For that reason alone, I generally stick with the latter. Perhaps it is the default setting for a good reason.

Myself I have switched back and forth several times because I can’t really decide which gives me a more accurate track. When I have it set to Time> One Second, sometimes I think I get a less accurate track which, especially in tree cover, seems to zig zag back and forth from side to side. The “true” track seems to usually be somewhat in the center of these zigs and zags. Set to Automatic> Most Often, I get fewer Zig Zags, but the track can sometimes wander off many feet in one direction and stay there for a bit and also sometimes cut off corners or switchbacks a bit.

I was wondering if GPSrChive has determined if there is really a best setting to get the most accurate track? Also what others have their 66 set at, and why.

edit: I do realize the real answer probably has "sr" in it, but for now I have the st.

For GPSMAP 66st. software 6.00
In the beginning i always used 2 second logging as i know from earlier units that this is an optimal setting for my usage.
2 second logging gives a very accurate tracking, but does not exhaust the unit or memory, especially older units struggled with 1 second logging.
(tried 3 second logging as well, but this is too sparse)

On the GPSMAP 66st after some time i tried Auto and Most often tracking setting. For all my use, such as driving, cycling, running (seldom with gps), hiking the auto setting is as accurate as 2 second logging, but uses about 30%-50% as many trackpoints, often less if the travel pattern is straight or there is a pause. It does detect sideway moves (turns) very accurately and places the points very correct compared to roads or other paths (i have lock to road off).

Since the total logging time is greatly extended using auto, compared to time, auto logging with most often setting has become the natural choice for me.
Fast cycling or running tricky sections is the only time it may be a little off, but i can't say i have ever seen the time based logging being more accurate, at best the same.


Side note: i have also detected that if travelling fast (say in a car) the expedition mode wakes up much more frequently and places track points rather accurately as well. When the GPS is not moving it may be many minutes between the track points saved.

Prior to the release of the Garmin Oregon 6x0 series, all devices would record track log data at the user specified rate while always updating Trip Computer statistics once per second. This frequently resulted in users complaining that their Track Log and Trip Computer statistics did not match each other after a hike or other activity. This page explains why that happens.

The common work-around was to set the tracklog recording settings to manually record a track point once every second to effectively match the Trip Computer recording interval and provide statistics that closely mirrored each other. For many users, this was the only workable solution.

With the Oregon 6x0 (and most new models since), Garmin included additional controls that allow the user to choose when the Trip Computer accumulates data: 'Always' or 'During Activity'. Selecting 'During Activity' alleviates the 'mismatched statistics' issue so often previously reported, while selecting 'Always' retains the legacy behavior for those users who prefer it.

This meant that users could now tie the Trip Computer data accumulation directly to the Tracklog Recording status, freeing them up to use any tracklog recording interval or method they desired instead of being stuck with one track point per second as with the older devices.

Most of those older devices were only capable of using GPS and perhaps WAAS/EGNOS to determine their position, which, when coupled with a once-every-second tracklog recording setting, could lead to some very messy 'spider-web' tracklog recordings each time they encountered a weak signal due to environmental factors beyond their control. These always had to be cleaned up after the fact on a PC or Mac using tracklog editing software.

Newer devices with the additional tracklog recording controls also have access to more GNSS systems and can therefore mitigate some of the 'spider-web' recording issues that were previously unavoidable.

This leaves many users now wondering just what tracklog recording settings they should be using to achieve the best results for their respective activities. And the simple answer is, there is no one setting that will always provide the best results for every user under all circumstances. (but we may get close!)

Using shorter time intervals (such as one track point per second) can result in a recording with the maximum amount of detail, but may require some post processing to thin some of the excessive and unnecessary data after the fact.

Using longer time intervals can reduce the amount of post processing necessary, but only at the expense of losing some detail.

My rule of thumb has always been that activities with slower movement and/or frequent changes in direction require time intervals with more frequently recorded track points. Faster activities, or those with few changes in direction do not require as many recorded track points to retain reasonable accuracy.

Now, you may be slowly hiking some switchbacks one day, and then sprinting down some forest service roads in your ATV the next, and neither activity would benefit from using an identical tracklog recording configuration.

This is where the 'Automatic' settings come into play.

When configured for 'Automatic', the device is always comparing your current position with the last recorded position to calculate when to record the next track point in the tracklog. This setting results in less recorded track points when travelling at higher speeds with little directional change (driving down the interstate) while also allowing detailed path recordings while navigating twisty trails at lower speeds. This means the user can use one setting for multiple activities while achieving acceptable results.

The user can also choose between 'Least Often', 'Less Often', 'Normal', 'More Often', or 'Most Often' to 'fine-tune' the 'Automatic' setting to their liking.

For most of my activities, I find that 'Automatic' and 'More Often' seem to get the job done with satisfactory results, but that may not be true for everyone.

I have yet to be able to find a good use for the 'Distance' based recording intervals.

Thanks for all that information! So if you assume 100% use is hiking, what sorts of track differences would you expect to see when comparing the same trail, recorded twice, first with Automatic> Most Often and then a second with Time> Once Every Second?

In my experience close to none in terms of track accuracy, just less data with the auto setting.
I have tried it a few times but since i don't keep logs for long i cannot show the comparison.
If you feel for it just try some 1 km loop twice or walk the same way one way and back with different setting, save after each section.
Just let the gps fix for 15-30 minutes before the first round to avoid a gradually better position.

Auto logging should lead to fewer auto saved tracks and less need to manually save the track before the log is full.

From what the gpsrchive user described, it seems that the calculation of the track position is enhanced on the newer devices.
On old devices the auto setting always gave less accurate tracks than time based (1-2 sec) in my opinion and experience.

987 wrote: Wed Dec 30, 2020 10:09 pmIf you feel for it just try some 1 km loop twice or walk the same way one way and back with different setting, save after each section.

As I said in the original post what I seem to see, although I am not 100% sure, is:

When I have it set to Time> One Second, sometimes I think I get a less accurate track which, especially in tree cover, seems to zig zag back and forth from side to side. The “true” track seems to usually be somewhat in the center of these zigs and zags. Set to Automatic> Most Often, I get fewer Zig Zags, but the track can sometimes wander off many feet in one direction and stay there for a bit and also sometimes cut off corners or switchbacks a bit.

I am wondering if GPSrChive believes that is what is expected, or would the expectation be that Time> Once every second should give a more accurate track recording for hiking? I know it should give the maximum amount of detail, but is that detail expected to be more accurate than Automatic> Most Often? I don't care about the memory space or post recording cleanup. I don't store the recordings on the GPSr and I always do cleanup anyway and have become efficient and fast with that.

I did some tracklog testing today using a GPSMAP 66s (tracking set to one point every second) and GPSMAP 66sr (tracking set to automatic, more often).


TRIP A

During a 13 mile drive, on city streets and highways, with speeds ranging from 5mph to 75mph, neither unit faltered, providing essentially identically accurate tracklogs. The 66s recorded the trip with 1199 track points, where the 66sr only used 264 track points.



TRIP B

Once I parked and reset both devices, I walked across a parking lot and into a building, where I stayed for nearly an hour before exiting and returning to my vehicle. The 66s recorded a trip distance of 0.68 miles using 3114 track points, and the 66sr recorded a distance of 0.20 miles using only 159 points. Both exhibited 'spider web' tracking, but this is expected when indoors or under heavy cover. In this scenario, the 'automatic' recording interval was far more accurate than the 'once every second' recording interval.



TRIP C

Now I reset the devices again and proceeded to walk around a large building. Most of the walk was in the open with some partial 'shade covering' obstructions in some areas. This time, both units recorded a very similar trip distance (0.44 vs 0.47 miles), the 66s using 1293 track points and the 66sr using only 93 track points to do so. There was about a half mile an hour difference for the reported moving average. Interesting to me was while the 66s track looked smoother due to all those additional track points, it also showed me walking through several walls, and even cutting through a building, none of which actually happened. This has nothing to do with the track log settings, of course, but rather the difference between Single Band/Two System vs Multi-Band/Multi-GNSS capabilities.

Interesting. Thanks for that. Trip C would seem to indicate that outside the sr does indeed provide a more accurate track.

I also did some testing today with two st units and my Foreruner watch. I hiked two trails that followed either side of a gravel road and a stream, (switching over and back a few times). I had the watch and one of the st units set to one second, the other st set to auto>most often. I was really just trying to see if I got a better result with the one second recording vs auto>most. I don't think I really did but it was quite hard to be sure. Both settings had periods of good accuracy and other periods of being off by 30 or 40 feet, interestingly not always at the same time. Both the st units and the watch recorded a generally similar looking but the watch was about 20 or 20 feet to the north of the two handhelds the entire time. Even when the watch track looked like a carbon copy of an st track, it was always to the north. After I analyzed the tracks I only kept two of the st recorded ones, so don't have them to further look at. But I think I came away with the decision that using the Automatic>Most Often setting is pretty much the same as the one second setting for getting a good track especially if you stop moving for more than a few seconds during the hike. And I definitely will take the watch off of one second as it make no real difference with that. But the more I play with this issue, the more I want a 66sr!