GPS Accuracy and Repeatability

[GPS_Accuracy]

This is my first watch from Fitbit with GPS, but so far I'm finding that it constantly makes me run 0.15-0.2 further on a 2-3 mile run than what my previous Garmin would have me run, and when mapped out on MapMyRun. Not only does it make my routes further, but the consistency between runs of the same routes can very as much as 0.1. I'm wondering if this is normal, or if my watch is defective? I do notice when I review my routes that it often shows me cutting corners by quite a bit. I've also noticed when I run a route that has me run the same sidewalk, there's often a lot of variation between the paths over the same general area.

I'm wondering if this is normal for Fitbit GPS? I run in a suburban area without tall buildings, so I wouldn't expect to be losing GPS connection. And again, I never had these issues with previous GPS tackers.

[SunsetRunner]

Unfortunately, this is an issue with the GPS used by Fitbit.  The trackers use a combination of stride length and GPS to track you and is known to aggressively cut out corners and over straighten bends.  Might be something to do with Fitbit only using GPS whereas Garmin use GPS, GPS + Galileo or GPS + Glonass.  As someone who has owned Fitbits and Garmin, the Garmin is a much more accurate GPS tracker.

[t.parker · Accepted solution]

There may be small differences between two different recordings of the same route due to the nature of the GPS. However, they should be rather marginal. First, you can verify the route by exporting TCX file of your activity from the dashboard and uploading it to https://www.mygpsfiles.com/app/ . The reason of doing it is to make sure the distancd you see is indeed the one coming from GPS data. Before, it wasn't the case and I found out that many times that despite the GPS distance was correct, the one shown by the app (and watch in the real-time) were totally wrong. GPS (in my case it was the Ionic) was quite accurate when it worked but Fitbit algorithms were using stride and steps to artificially correct the distance - it was bringing inaccuracy and always different distance on the same route.

Now, since I don't have working Fitbit watch, you may try to verify another thing by yourself. Start run with GPS, wait until it's connected and start running/skipping in place. Observe the distance on the watch. If distance increases despite you not moving - here's your source of inaccuracy. I kinda hope Fitbit updated the algorithms but if not, the distance computation uses stride and steps rather than GPS data. With Ionic, the data captured by GPS was not worse from any other device I use but Fitbit algorithms were just wrong.

[GPS_Accuracy]

Thanks- that's exactly what's happening. Often when I review my route, the path will show me running through backyards and houses when going around a corner. It's frustrating that it can be that inaccurate, especially for the price point of the Sense.

[GPS_Accuracy]

Thanks- I'll have to give that a try.

Regarding the different recordings of the same route- with previous trackers, I would notice about a max of 0.05 mile difference over a 3 mile run. Not enough to concern me and what I thought was reasonable. Now it seems like I have to run down the road a couple more blocks and around the corner before I even hit the minimum.

[Elfred]

I have been using a Fitbit for a long time. Just traded the Ionic in for the Sense this weekend and saw a deviation in both pace and distance on my usual 5 k route. This morning I started the app on the Android and the exercise on the Sense at the same time and got this much difference in just 5 k!

Is there some calibration the Sense needs? This seems extreme to me.

[t.parker]

@GPS_Accuracy the GPS trackers very often show less than great accuracy at the corners. Very often this is because of the reduced sampling interval in order to keep battery usage low. On Garmin, you have Smart Recording, which forces sampling whenever it detects a significant change in the acceleration, direction, or cadence (I keep it off, it isn't that smart). On Suunto (Spartan family), you have multiple sampling modes which, if you go less than Best, it reduces the recording interval and also cuts corners. However, some trackers will apply additional data smoothing or even aligning the data to known roads and paths. There's not much you can do about it. Also, it could be just GPS drifting (which usually I experience a lot around autumn in the UK when the sky is very clouded).

@Elfred try to test (the way I described in the previous reply to this post) how the distance is calculated on Sense. If it's steps and stride rather than GPS data then there's not much else you can do. If it's purely the GPS data, you may try to start the exercise and wait a short while after you get a GPS lock. It may help.

[Elfred]

Thanks!

As I said, I have been using Fitbit for a very long time so have several years of other fitness data to tend from. I did think about the Garmin and actually tried the Apple Watch but am comfortable with Fitbit.

That said, it is disappointing to see the difference between the two Fitbit devices. I would hope that a newer device would (if changed at all) be an improvement. Of course, maybe it is more accurate and I have been lying to myself all this time…

[t.parker]

@Elfred mind that any device may experience a "worse day". Even if there are perfect conditions outside. Last Saturday, I attended a virtual race and tracked it with two watches and my mobile phone. The result couldn't be better, all the devices came up with pretty much the same distance (a large part of the race was a trail run) and both watches were vibrating on each km at the same time (sometimes they ware like 1-2 sec apart). Yet Today, my 6km run got totally screwed up by one of the watches (registered distance 0.6km less than the real one), both watches experienced some GPS drift, too. I believe weather conditions could affect it. It may happen from time to time. However, my running dynamics are unaffected as I stopped trusting GPS a long time ago and started using a foot pod sensor for measuring my run parameters.

[Mike_Croy]

@SunsetRunner my understanding is that Sense also uses Glonass as well as GPS (but no Galileo).

[SunsetRunner]

@Mike_Croy you are correct the Sense uses Glonass as well.

[Celaon]

 

I've switched from Ionic to Sense and I'm pretty disappointed.  I always rum with my husband that uses Garmin. Before, my Ionic were more accurate that his Garmin, but now my Sense is giving my completely a none sense results. Is possible to do something to fix it?

[Pete-Fl]

I just changed from the ionic to the sense. The same running route I’ve been doing for sometime now measured 5.25 miles on my ionic. Today, using my sense for the first time, it measured 5.49 miles... Something is amiss...

[pdotterw]

I got a Fit Bit Charge 5 for Christmas so am still learning.   I set the Fit Bit to GPS mode that only used the GPS in the Charge 5 (not dynamic mode which would use my phones GPS to save power).  Every other day I hike the same 5.9 mile route through the hills of Marin County in California.   One deep canyon with redwoods, and about half the route is under tree canopy.   Fit Bit dashboard has reported 7.1, 6.9, 6.6,and 5.5 mile for the same route on different days.  For kicks I took my Older Garmin Vista HCX on the same hike.  It registered the route properly at 5.9 miles.   I uploaded the Fit Bit TCX GPS track file that was reporting 6.6 miles to https://www.mygpsfiles.com/app/ .   That website said the data file was a 6.6 mile track (Garmin 5.9).  So it's not the app or dashboard, it is the raw data recorded by the fit bit.   The old Garmin also had WAAS/EGNOS turned on so probably more accurate with those positioning correction aids.   So Fit Bit Charge 5 mapping, distance will be taken with a grain of salt from now on.   

[pdotterw]

More Data:   Per a variety of Fit Bit Charge 5 reviews and comments - I "loosened" the band so that I could fit a finger under the band while on my wrist.   According to Fit Bit this will increase accuracy of the GPS.    It did - sort of.     On suburban streets, through our small towns and in open areas it was OK accurate +/- 15 feet as far as I could tell by the map overlay of my route in the dashboard.   In a Redwood tree grove on my route it was +/- 50 feet.  Then it fell to about +/- 75 feet in switchbacks up a gully on the back of a hill.    The problem is - it dutifully records points with this variability and connects them as if I walked that erratic route.    It adds up.    A 5 mile route becomes a 6.3 mile route as it connects the jagged "off trail"  waypoints.     Perhaps the software could be set to record "less often" (most waypoints are recorded at 1 second intervals) and maybe ONLY when "X" number of GPS satelite signals are recorded (x = 3 or 4 or 5 whatever would yield more accurate triangulation).   Fewer waypoints might "average" out the error rate.    Hopefully the firmware can be improved in a future update.   

[AshLynn043]

My sense is much different that someone with me wearing a versa 2, with their pace showing almost 2 minutes less than my pace. The ionic seemed to be more accurate for me

[AshLynn043]

I’m having the same problem. My Sense is showing I am running a pace nearly 2 minutes slower than the person running behind me wearing a versa 2