GPS Accuracy

is there only smart and seconds settings? I thought there were more choices with previous models

That could mean that metrics like Vo2Max are wrong.

Just for curiosity, how is the pace with the correct calculations from coordinates.

Oh ok, i think I didn't get what your variation means. You calculate the difference between the GPS distance between two points and the "Garmin distance" recorded on the device, right ?

Just did a first quick test on the bike with my 6 Pro Sapphire, tried GPS + Galileo and GPS + Glonass, Smart Tracking enabled.

Fenix on the wrist below my winter jacket.

https://connect.garmin.com/modern/activity/4460937382

https://connect.garmin.com/modern/activity/4462441987

It doesn't seem to be so far off, not as in your example... 

Nike ride accuracy is often very good. Inside Paris, mine are often really really good. I'm in the middle of the street, so it's easier to see the satellite. Maybe the fact that the watch is moving less also.

I did what beta-L-user did, calculating the distance between GPS points, and comparing it with the "Garmin" distance. I looked at big differences, checked the time, and looked at my run to see when those happens. For this first test, it was a run in Paris, a difficult one for the GPS. I'm very often next to tall buildings, so it's not a very clean track. What I see in my sheet is making sense : Garmin is using the accelerometer of the watch, since I had no GPS at the start, yet the watch was recording some distances. When I was slowing down or stopping to cross a street, often the GPS was not slowing down much, while the "Garmin speed" seems to make more sense. When some GPS distance and speed was clearly wrong (like >30km/h), the Garmin speed was more normal.

In the end, the GPS distance was longer than the Garmin distance (no surprise). Retracing the run the best I could remember, it made the Garmin distance closer to the real one than the GPS distance. So for THAT run, their calculation is a good one, it compensates for the GPS errors. 

Next test, on a run where the Garmin distance seems really too short (600m almost, on a 17km run). I didn't retrace that one (too hard, often in the forest, so I'm not sure what is the real path on the map). But looking at how the Garmin distance and GPS distance differs, it's obvious that there are significant difference when the speed is really low (steep parts). Otherwise, the Garmin distance is always a little bit lower, and it adds slowly. 

I did what beta-L-user did, calculating the distance between GPS points, and comparing it with the "Garmin" distance. I looked at big differences, checked the time, and looked at my run to see when those happens. For this first test, it was a run in Paris, a difficult one for the GPS. I'm very often next to tall buildings, so it's not a very clean track. What I see in my sheet is making sense : Garmin is using the accelerometer of the watch, since I had no GPS at the start, yet the watch was recording some distances. When I was slowing down or stopping to cross a street, often the GPS was not slowing down much, while the "Garmin speed" seems to make more sense. When some GPS distance and speed was clearly wrong (like >30km/h), the Garmin speed was more normal.

In the end, the GPS distance was longer than the Garmin distance (no surprise). Retracing the run the best I could remember, it made the Garmin distance closer to the real one than the GPS distance. So for THAT run, their calculation is a good one, it compensates for the GPS errors. 

Next test, on a run where the Garmin distance seems really too short (600m almost, on a 17km run). I didn't retrace that one (too hard, often in the forest, so I'm not sure what is the real path on the map). But looking at how the Garmin distance and GPS distance differs, it's obvious that there are significant difference when the speed is really low (steep parts). Otherwise, the Garmin distance is always a little bit lower, and it adds slowly. 

Example of where the Garmin distance & the GPS distance differs the most : I have created a map of my run, with the color of the dots changing according to the % of difference between the distances. If it turns red, it means that Garmin is counting less distance than the GPS, if it counts more, it turns blue. 

It's easy to see that when I'm crossing a street and pausing a bit, the watch is trying to adjust and count less distance. What surprises me a bit more is that it's not really "catching up" after I start crossing the street, when the Garmin speed should be faster than the GPS speed. But on that run, I can't really complain, since the GPS distance was too high due to the poor GPS conditions. I will play a bit with my other run to see if this kind of visual data can tell me more.