Unbelievable GPS accuracy - bike mode

Simple- far higher speed compared to running, which makes life far easier for any GPS device.

Higher speed would mean less recorded points per same distance travled(in both cases measurement is every second). Not sure how this could contribute to a better track- every wrong measurement would be even more obviouse. My guess, would be that either better result is due to difference on how bike and run apps are working or simply becase when biking wrist (and watch) is a bit further away from the body and respectively less impacting GPS signal. I`ll try to run and record via bike app and share results.

Whilst running or even walking, the watch is constantly moving back and forth with your arm swing and the watch is unlikely to have the bezel, that the GPS antenna sits under, facing towards the sky; so it's actually a pretty rubbish situation for GPS reception.  When cycling, as well as the greater speed (which makes the filtering position noise between the 1 second points easier) your wrist is more likely to be at a more optimal position for the antenna as you hold the handlebars and you're not constantly swinging it with and against the direction of travel.  You'll get even more awesome GPS tracks if you mount the watch to the handlebars.

[To help explain the speed thing, say you moved 2 metres between each 1 second point running and 5 metres cycling, but the GPS offset 'noise' is 1 metre; then that's going to make the running track more 'wobbly' than the cycling track (if you applied no smoothing algorithm at all).

I don't think is so much the speed that contributes to better results.

I get better results with walking than running.

I think is the rapid swinging of the arm the issue with recording runs

What you say makes a lot of sense.

As mentioned earlier, I recorded running activity using Bike app. Some observations based on that experiment:

1) Accuracy is good (comparable with ordinary case when I`m using Run app), but far worse comparing with the accuracy achive while biking+Bike app

2) Intentially, I went trough some of the same streets/alleys to visualize the deviations. In the 1km stretch below you can clearly see how bike (blue line) is exactly on the alley and run (going and back orange line) is following the alley on the left side pretty consistently. This behaviour I observe on a regular basis with my runs - recorded track always have offset towards left side.

Conclusion: As Crispin_Ellisdon suggested, while running (especially in challenging conditions), watch is having hard time to produce spot on track. Probably, reason is either hand movement or watch`s close position to the body(vote on this one since accuracy is much better on open field, while hand movement still persent). Based on that I`d suggest that problem with accuracy is probably harware related and won`t improve over the time with updated firmware(unfortunatelly). 

Bike Activity:https://connect.garmin.com/modern/activity/4759107369

Run Activity: https://connect.garmin.com/modern/activity/4760620376

Thanks for your test.

What I do to compensate this issue is to run with the watch uncovered by clothes and to keep the watch facing the sky and my palms down. I get used to it and feels natural.

Hmm, I might do a test run with the watch mounted on my shoulder (should be easy with a hydration vest on). Cadence and HR come from the chest belt, so it might be fun to see if the eternal shifted 20 meters to the left/right/wherever is solved that way.

It's the higher speed.

The reason is a little bit subtle; a lot of the issues with GPS position, especially around buildings and trees, are caused by multipath signals. The faster you are moving, the more easily the watch can identify & reject the reflected signals and calculate your location only from the direct signals; a large part of that is Doppler frequency shifts which are different for the direct and multipath signals, and which the GPS chip derives very precisely.

The same was observed here: https://forums.garmin.com/outdoor-recreation/outdoor-recreation/f/fenix-6-series/219040/shift-in-gps-track-depending-on-wrist, seems like the track always switches to the side of the wrist your watch is on when running.

Crispin_Ellisdon said:

Whilst running or even walking, the watch is constantly moving back and forth with your arm swing and the watch is unlikely to have the bezel, that the GPS antenna sits under, facing towards the sky; so it's actually a pretty rubbish situation for GPS reception.  When cycling, as well as the greater speed (which makes the filtering position noise between the 1 second points easier) your wrist is more likely to be at a more optimal position for the antenna as you hold the handlebars and you're not constantly swinging it with and against the direction of travel.  You'll get even more awesome GPS tracks if you mount the watch to the handlebars.

I completely agree with this. When running, for most of the time the face of the watch is facing sideways, not up at the sky like on a bike. With the Fenix series, because most of the case is made from stainless steel, this degrades the already weak GPS signals from the satellites as it's harder for signals to pass through steel than plastic, and when the watch is constantly being swung around from your arms this makes it even harder for the watch to lock on to a solid satellite fix. I don't think speed is necessarily the reason for the better tracks, although that helps as it results in fewer location anomalies and hence a smoother track.