Distance accuracy discrepancy between All+ Multiband and SatIQ modes

You cannot compare two different watches in this way, because the distance that each watch records, is the result of combining the GPS data with cadence/accelerometer data. And the cadence-based distance is strongly depending on a proper calibration of the watch. If improperly calibrated, it can make up the difference of up to 10%, and in some cases even more, despite having a perfectly correct GPS track.

What you can do, though, for comparing the GPS accuracy, is opening both activities in Garmin Connect, and saving them as Course. At the saved course, you then see the true distance calculated from the GPS coordinates, without the influence of the accelerometer data. Only that will allow you to see what the difference in GPS accuracy really is, though you'll have to find out which of the watches is closer to the reality.

Adrian s said:

there was an elevation difference of about 5m total ascent and about 8m total descent

That's actually an excellent result, considering that Garmin claims the accuracy of +/-10 feet (~3m) at any given point [source] (and there are likely hundreds to thousands of points in a 20 km run activity)

Thanks for your valuable input as always. 

I'm always analysing the GPS tracks themselves, not the results in Garmin Connect. 

We try to maintain the same Cadence and stride length at all times when doing these tests. Fenix 8 makes it the 4th generation of watches we've ran with and tested throughout the years. I've found Fenix 5 and 6 the most accurate when testing this way (although Fenix 6 was reviewed by media to be notoriously inaccurate with GPS tracks and positioning on the map itself). 

Adrian s said:

I'm always analysing the GPS tracks themselves, not the results in Garmin Connect. 

OK, I see. In that case please post the distances shown on the activity stats, as well as the distances of the GPS tracks. It should help, understanding where the difference in the reported pace comes from. However, as I wrote, I am persuaded that it is more the consequence of the diference of the pace & distance calibration of both watches, and much less the result of the GPS accuracy. The fact that you were running with the same pace and cadence does not exclude it at all.

The best would be posting links to both activities, which would allow for a better analysis, though I understand if you do not wish posting it on a public forum.

Do you mean you wore both watches, or one and your running partner the other?   In any event, a difference of 800mm seems pretty good.  These watches don't use military grade GPS which would be crazy accurate.  Many other things come into play when calculating gps tracks.  Maybe one watch locked onto more satellites than the other getting closer to true data.

One watch for each runner, testing only the GPS distance accuracy by analyzing the GPS tracks only. 

That is indeed the case, yes, one watch was set to all+multiband and one on Auto, as I mentioned, so naturally one would connect to more sattelites. My end goal is to warn people that actively train, that this setting might skew their training data.

Thanks for your response. 

We've subsequently tested both watches in All+multiband mode and the distances were spot on (a few meters difference, but this is negligible in training). 

So I'm led to believe that there are indeed "notable" differences between the two Positioning modes. Notable for people who actively train, as I was mentioning, as resulting paces might place the runner in different training zones if training by pace. 

I base my training heavily on HR and RPE, not pace, but this was my conclusion when testing the newly (2 months) acquired fenix 8's , which I we are still putting through their paces. 

trux said:

If improperly calibrated, it can make up the difference of up to 10%, and in some cases even more, despite having a perfectly correct GPS track.

Do you know why that is? Wouldn't it be logical to calculate all metrics based on the actual (perfectly correct) GPS track distance, instead of basing it on arm swing and accelerometer data which is prone to miscalibration and incorrect data based on different gaits and cadences? From a purely scientific point of view I find it illogical. 

I tried SatIQ, multiband, allsat and GPS only in a local park with exact measured distance on my Epix. SatIQ, multiband and allsat show no measureble difference, while GPS is in most cases short due to some corner cutting, but there is no way I can tell what SatIQ is using at any point. I use allsat for running and SatIQ for hiking due to better battery life.

Differences might be small and only noticeable on large distance, hence I've tested only in a half marathon, not short runs 

As I wrote, the pace and distance reported by the watch do not use the GPS data only. The algorithm is more complex, and using also the accelerometer data, hence any differences in the GPS track in no way mean that the pace and distance shown on the watch are impacted by them. 

Adrian s said:

Do you know why that is? Wouldn't it be logical to calculate all metrics based on the actual (perfectly correct) GPS track distance, instead of basing it on arm swing and accelerometer data which is prone to miscalibration and incorrect data based on different gaits and cadences? From a purely scientific point of view I find it illogical. 

Quite oppositely, it is perfectly logical, and done exactly for the case you mention - varying accuracy of GPS tracks, depending on the satellite options, recording mode, and external conditions (visible skies, weather, underpasses, reflections, interferences,...). Combining the data with the accelerometer allows to eliminate a big part of GPS errors.