Apple Watch 4 GPS Accuracy (44mm Nike Edition)

Test Route

I’ve done my GPS ‘test’ on the Apple Watch 4 (44mm Nike Edition). The Apple Watch 4 (AW4) uses GLONASS as well as GPS but let’s call it GPS as no-one knows what GNSS means, except you clever lot reading this πŸ˜‰ . The even cleverer ones amongst you will point out that the GNSS chip is GALILEO compatible, so the Apple Watch 4 is probably using that as well.

The results OF THIS ONE TEST are very interesting and unusual, if I may say so. I’ll hold fire until a later review/post with my overall verdict based on lots more runs/rides but here is my commentary on the formal test result from today.

It scored 77% using the standard workout app

That’s pretty much the same as a good Garmin on a good day. ‘Normal Good’ but not in the same league as the V800, Ambit, Coros Apex 46mm or a Spartan Trainer – also on a good day. There is some subjectivity in this test and I was perhaps very slightly erring on the harsh side of fair – being kinder it would have scored 79%, so no material difference.

The overall distance was under-reported by the AW4 by just over 1% and I typically find that almost all devices are accurate to within +/- 1% over this 10-mile route. So when something strays out of that +/-1% range then something unusual may be going on.

Looking at some of the specific segments of the test, the Apple Watch did very well in easy GPS reception conditions and fared less well when the GPS-going got tough. Nothing unusual there.

>> DETAILED TEST RESULTS SPREADSHEET AND SOURCE FILES HERE <<

But there is something strange going on with the GPS track. Dcrainmaker picked this up to a degree in his nice sports-focussed review of the AW4 (link to: dcrainmaker.com) when he noted that the Apple Watch cuts corners. I thought, when I read the review, that there may have been some special use-cases that caused this. But no. I agree with what he says, it does cut corners. Sometimes not by very much, sometimes by ‘quite a bit’, but it does cut them consistently. That could explain the under-reporting of distance. Perhaps. Let’s park that thought for a minute.

Take a look at this segment of the test.

click if you really want to.

The Apple Watch 4 is in green (Ambit 3 in blue, Coros Apex in red). Apart from the detour into the river, the Ambit is probably spot on for the circle. Except it’s not a circle it’s a square. Yet the Apple watch has turned this into a circle more so than any GPS device I’ve seen.

Off to the left of the above image, there is an out and back with a fairly tight turn around a tree as shown next…

The Apple Watch doesn’t spot this at all and just assumes it is a simple out-and-back (green).

The following image shows the Apple Watch cutting some corners. Fair enough, it happens. But look at the rest of the track, it is very, very smooth, almost perfect-looking. And that same apparently perfect smoothness is typical of most of the rest of the track.

But it can’t be perfect, can it? Because it clearly cuts those two corners (and more besides)

So I’m pretty sure that the Apple Watch is doing some manipulation of the GPS track here. It’s either smoothing the raw data it gets and/or adjusting the GPS track based on its inbuilt motion sensor (a bit like Suunto’s FusedTrack but on a finer scale).

Random selection of 15 seconds of data, shows different GPS values per second. Clickable

The upshot of all of this is that the Apple Watch 4 may be a little bit ‘smoke and mirrors’. The GPS track looks really, really good. It’s deceptively smooth and deceptively close to where you’ve been – but not quite close enough. It almost fools you into thinking it’s awesome. But it’s not. It’s on the good side of average just like Garmin.

A final point on the smoothness of the track. I have seen a similar smoothness in Garmin devices that use GALILEO (and GPS and GLONASS).

Possible consequences

  • Those of you wanting accurate distances could be disappointed
  • Those of you wanting good running pace on the AW4 might actually have some luck because of the excellent smoothing. I’ll dig into that later.
  • Those of you wanting a sweet-looking, post-run GPS track will be super-happy.

That’s my 2 cents worth. Thoughts?

Corner cutting when cycling too (yellow). And some of that corner-cutting makes the track out by MORE THAN 5 METRES which we would normally call FAIRLY BAD. Just sayin’ :

Clickable, Apple (Yellow), Wahoo (Green), Garmin 935 (Red)

PS: The test was conducted WELL AWAY from my smartphone…about 2 miles away.

PPS: Apple watch can be calibrated for distance (Source: apple.com). This is PROBABLY not relevant to what I am doing here, although if they are massaging the GPS points based on accelerometer readings then it might be relevant. I am totalling the distance from the recorded GPS points in 3rd party software (consistent to all my other previous tests). My other tests typically record GPS-only and GPS+GLONASS if GPS-only is unavailable (as in this case) so it is consistent for me to exclude as much ‘fancy stuff’ as possible.

PPPS: My first run after this was with the STRYD app using effectively 1Hz recording (1 GPS point per sec). That track is not as good. The track is more jagged. It *DOES* make a better attempt at sharp corners and does NOT cut the corners BUT much of the track with STRYD‘s app is away from where it should be when compared to that produced by the Apple workout app.

Edited: Version 1.02

Support this site with purchases at these partners - should click to a local choice in your country

12 thoughts on “Apple Watch 4 GPS Accuracy (44mm Nike Edition)

  1. Hi! Two silly questions: did you leave the phone at home? The AW uses its GPS only if it is unable to use the phone’s (to save battery).
    The other question is regarding how you record the run. I mean, if the AW is somehow changing the raw data, it must be the Workout app. Maybe using the Strava or Endomondo app the results are more β€œreal” and you can then separate the gps signal from the accelerometer correction.

    • yes i leave the phone at home and i start the test well away from home.
      the workout app may be changing the data, the gps chip may be changing the data. I am planning to use the STRYD app at some point I wil lfind out from the developers their sampling rate.

      • Please, do not take my comment as criticism.

        I’ve seen the comment of Lee Davis and my point is the same. I’ve been testing an AW the last few days but I’ve been unable (IT tendinitis) to test the GPS. But I’ve read several posts in different blogs talking about the differences between using Workout and another app (Strava mainly), and I wanted to know your expert analysis.

        saludos! πŸ™‚

  2. There’s been lots of comparisons done in the past and results show that the default workout app cuts corners and ends up being comparable to Garmin (as you say), but using 3rd party apps like Strava or Workoutdoors which appear to have more frequent sampling provides very accurate results and no cutting of corners. I’m a 10+ Garmin user, but decided to try an Apple Watch from the S4 and must say I’m very pleasantly surprised. Just fails me on all day hikes though πŸ™

    • maybe. but zooming into the details of the lines drawn on my maps (above) show that each line is the same distance from one point to the next which to me implies the same ‘sampling rate’

      • I’ll dig out the link to the YouTube video I watched just last week. The guy did the same track, running, walking, cycling with a garmin and an Apple Watch. The Garmin was simply there each time to provide a comparison because there’s no guarantees you can follow your last route exactly. But the one key finding was a difference between using Apple’s workouts app and the Strava app and the GPS plots were very different with no smoothing and greater accuracy.

        • that would be interesting. STRYD say they use an effective 1Hz sampling rate which they thought was what most apps used. HOWEVER there is a higher sampling rate which seemed to be intended for navigation BT which was very battery hungry. they didn’t know whihc sport recording apps, if any, used that mode. I’ll have a look at the FIT file as well and see how the gps points vary in there. (I don’t think any of this wlll affect my results…but could be wrong)

  3. Hi there, just found this post and I’m facing a similar issue when comparing my Apple Workouts distance and the exported GPX file.
    According to Apple Workouts I’ve run 8.190 meters but when I check the GPX file it says 7.970 meters.
    Where this distance difference is coming from? if the AW is supposed to use the GPS, why the resultant GPX file shows less distance?
    I’ve tried to get a response from Apple but they just say: The AW uses the GPS.
    This difference between the Apple Workouts and the GPX file is driving me crazy.
    Any thoughts?

    Thanks
    Dani

    • there could be an issue of the AW switching over to its own gps from that of the smartphone. as a test, start and stop the WELL away from the phone.
      there could also be an issue with the start of a workout waiting a few seconds to capture sesor info (hr and gps). maybe start a dummy workout for 2 minues walking around. discard it and then start your real workout.
      i’m not saying do it each time, just to eliminate these possibilities.

      • I go running without the smartphone, so the AW should only get its own GPS signal. Do we know how long it takes to get the GPS signal when I start the running workout? it could be the reason of the difference between the distance in the workouts app and in the GPX file.

Leave a Reply

Your email address will not be published. Required fields are marked *