Ultratrac calculates wrong distance/speed

I have now used Ultratrac twice while kayaking.

The first time I kayaked over 10km and Garmin tracked the distance as 4.8km.

Second time I kayaked over 30km and Garmin tracked the distance as 14.8km.

Both times the route has been tracked correctly, the route showed correct during the activity and the route shows up correctly in the web app afterwards. Both times the distance and speed have also been shown incorrectly during the activity.

I would hope that a 1000 euro device would be able to correctly calculate the distance between two coordinates.

Garmin chat support was trying to convince me that it is normal for the distance/speed to be completely useless in Ultratrac mode and for distance tracking I need to enable a GPS/etc mode.

Should Ultratrac show a somewhat accurate distance and speed?

EDIT:

Summary for now:

EDIT #2:

When driving a car on fairly straight and open roads:

  • Ultratrac calculated a tracked distance to be 17km (which was shown on the watch and in Connect)
  • After export GPX and import GPX, Garmin Connect calculated the distance to be 31.17km
  • Measuring the distance on Google Maps shows the distance to be 31.2km

So Ultratrac is able collect distance data with 100% accuracy in an "easy to track" activity like driving a car, but Garmin shows the distance completely wrong.

EDIT #3 - 2024-07-11:

I had a year long discussion with Garmin support from 2022 to 2023. They never acknowledged that there would have been any issues with Ultratrac. I then gave up trying to explain to them the issue. BUT it seems now after 2 years, some update for the Fenix 7 has fixed something. My Fenix 7 seems to be now running v17.28.

This week I went kayaking and used Ultratrac (which I would like to remind many of the commenters in this thread, I PAID 1000 EUROS FOR TO GET THIS FEATURE) and it seems like Ultratrac is now able to measure the correct distance while kayaking.

- In Connect, distance according to Ultratrac activity: 16.02km
- In Connect, export GPX, import GPX, distance in the imported activity: 16.09km (so basically identical)
- Same route drawn on a map by hand, got ~16km

I also tried to look at some stats when paddling and the few km lap times that I remember looking at during padding, they seem to be identical in the final lap times in Connect. Which is nice.

During the lap (not the final lap time after the lap is over) the lap pace jumped around, but that is to be expected. This can also be seen from the pace/speed graphs:

Pace graph for Ultratrac activity:

Speed graph for same Ultratrac activity that was exported as GPX and imported as GPX:

  • Do not use Ultratrac.  Frankly, I don't know why Garmin even offers that tracking mode anymore.  Ultratrac is meant to save battery by limiting the number of GNSS samples which results in large inaccuracies in distance calculations.  

    I would hope that a 1000 euro device would be able to correctly calculate the distance between two coordinates.

    It does.  Select any other Satellite option other than Ultractrac.

  • > Do not use Ultratrac

    But I want to use it, it is one of the reasons I bought one of the most expensive watches to get the best battery life. Not in these 1 day activities that I mentioned, but for longer ones.

    > Ultratrac is meant to save battery by limiting the number of GNSS samples which results in large inaccuracies in distance calculations. 

    "Large inaccuracies" are very different from measuring the distance wrong by 100%. Garmin support tried to sell me the "to get more accurate tracking use other modes" idea. This is not about "more accurate results", it is about completely wrong results.

    This is completely a software issue.

    Ultratrac is able to very accurately track you at least when you are moving 5-10km/h in open water. This is proven by the facts that:

    1. The route is tracked "correctly" and shown correctly on maps

    2. The activity can be exported as GPX and imported and THEN Garmin web app itself shows the distance/speed correctly

  • but for longer ones.

    How long would those activities be to make you concerned about battery life?

  • I remember that there was a similar issue on the 735XT which in ultratrac mode produced nigh perfect GPS tracks, but the distance was completely wrong. After about 2 years garmin acknowledged that there was an issue and provided a fix.

  • How long would those activities be to make you concerned about battery life?

    That is irrelevant for this thread, but for example 1 week.

  • Another experiment I did with Ultratrac, further reinforcing the idea that Garmin needs to fix their software.

    When driving a car on fairly straight and open roads:

    • Ultratrac calculated a tracked distance to be 17km (which was shown on the watch and in Connect)
    • After export GPX and import GPX, Garmin Connect calculated the distance to be 31.17km
    • Measuring the distance on Google Maps shows the distance to be 31.2km

    So Ultratrac is able collect distance data with 100% accuracy in an "easy to track" activity like driving a car, but Garmin shows the distance completely wrong.

  • Perhaps related to this are discussions about Fenix 6X measuring the distance short on a difficult terrain, up to 5-10% short. Let me explain that. I think, fundamentally, algorithm is the same between the Ultratrac and the normal every second mode. Even in the every second recording, GPS signal may temporarily drop, for example due to running through a tunnel or for other reasons that the GPS signal may be obscured or have very bad quality. From investigating this issue the conclusion was that the speed and the distance come primarily from the accelerometer when the GPS signal is less than perfect. With Fenix 7 the problem has mostly disappeared because the GPS quality is so much better, but I think the algorithm has remained the same. In Ultratrac mode the watch does short bursts of GPS updates that are followed by longer pauses between updates. However I think the watch relies on the accelerometer during those pauses to estimate changes in speed, and I think it derives the distance from the estimated speed. I think that process of estimating the speed from the accelerometer isn't working correctly - it is way too conservative as was shown with Fenix 6. And I think doing these tests in the car makes it even worse as there is no movement and acceleration pattern that the watch expects. Perhaps you should try repeating them in cycling mode to see if the result would be any different. 

  • I think, fundamentally, algorithm is the same between the Ultratrac and the normal every second mode

    Yup, here is another case of Fenix 6 showing too short distance when hiking https://forums.garmin.com/outdoor-recreation/outdoor-recreation/f/fenix-6-series/286610/fenix-6-pro-calculates-wrong-distance-in-km, while the GPX track is closer to the "real" distance that was travelled, which sounds like the same issue as with Ultratrac, just in smaller scale.

    I think doing these tests in the car makes it even worse as there is no movement and acceleration pattern that the watch expects

    Yup, perhaps as free QA for Garmin I could test some other activity modes while driving.

    But on the other hand driving a car in a straight line at a constant speed should be the easiest possible scenario to interpolate/calculate the distance between known points from some GNSS.

  • I've been a long time user of Garmin Fenix watches, using mainly for multi-day backpacking. I always used Ultratrac for 10 hour hikes on multiple days so the battery would last long enough. Super accurate GPS tracks are not necessary in those conditions. 1 minute interval pinging on GPS is fine.

    it used to work perfectly. The distance calculations were near perfect on my Fenix 3 and 5. Then two years ago it all went wrong, and I got the same half-distance reporting on an otherwise OK track. 

    They messed up something in the software, but as usual don’t acknowledge it. I sold my Fenix 5 after that and got an Apple Watch. Every update Garmin releases breaks some other feature. 

    Only now, with better batteries, am I considering a Fenix 7 or Enduro 2, because it seems I might get a week’s hiking done without having to use Ultratrac. But I’d still use it if it worked, because it would give even longer battery life. 

  • Indeed, the battery of the 7X or Enduro 2 will easily last through a backpacking trip, even without using Ultratrac:

    In GPS-only mode, hiking 10 hours per day, it would last at least 9 days (up to 15 days).

    Even in the super-ultra-mega-battery-killing "All systems + multiband" mode, it would still last at least 5 days (up to 8 days).

    And if you even just plug the watch into a battery pack one time at some point during the trip, then those figures are doubled.

    That's why Ultratrac is basically pointless on the 7x and Enduro 2. There should never be a scenario where anyone would ever need to use it.