Very interesting results. I'm 6'7", maybe the extra height means it takes longer to calibrate.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
3 Runs, 3 incorrect distances
- Thread starter jaconrad15
- Start date
- Sort by reaction score
Very interesting results. I'm 6'7", maybe the extra height means it takes longer to calibrate.
I have no idea what the accuracy of modern GPS watches is for running/walking distance (Google didn't turn up any definitive answers), but in most of the criticisms being made of the watch, the implicit assumption seems to be that the GPS watch is 100% correct.
If you have two GPS-based measures (say a Garmin and a Strava on the iPhone) and they give the same distance and the Apple Watch gives something different, this doesn't mean the Garmin/Strava are right and the Watch is wrong, it means the Garmin and the Strava used similar data and methods. You could wear two Apple Watches and get essentially the same distance, for example.
If you really want to say something about accuracy, you need to run a route where you truly know the distance, e.g. hop on a treadmill and measure the speed with a hand tachometer or run in the inside lane on a 400-m track.
I've done four runs with the Apple Watch (without a phone) ranging from 2-11 miles and the average error in distance compared to what I map out on runningmap.com when I get home is under 3% so far. Good enough for me!
If you have two GPS-based measures (say a Garmin and a Strava on the iPhone) and they give the same distance and the Apple Watch gives something different, this doesn't mean the Garmin/Strava are right and the Watch is wrong, it means the Garmin and the Strava used similar data and methods. You could wear two Apple Watches and get essentially the same distance, for example.
If you really want to say something about accuracy, you need to run a route where you truly know the distance, e.g. hop on a treadmill and measure the speed with a hand tachometer or run in the inside lane on a 400-m track.
I've done four runs with the Apple Watch (without a phone) ranging from 2-11 miles and the average error in distance compared to what I map out on runningmap.com when I get home is under 3% so far. Good enough for me!
Last edited:
odd though
I just thought it off though that ONE of the apple watches was exactly the same as two GPS devices and the other Apple Watch was off by quite a bit.
I just thought it off though that ONE of the apple watches was exactly the same as two GPS devices and the other Apple Watch was off by quite a bit.
Well it sounds like you had one with a phone and one without (i.e. one using GPS and one not) and with different calibration histories, in that case I would expect them to give different measures.
indeed, although the one WITH the iphone connected was the one that was most off, and the GPS from THAT phone gave a different reading to an app on that phone and it lined up with another phone with GPS but which was not connected to a watch..
we'll see how it rolls out.
THIS.
Having researched this the last few days, it appears that the AW doesn't use the GPS on the phone unless you have this turned on. This is not the case with other apps using GPS. I think this is further supported by the old Nike+ app which basically did the same thing the AW does in using GPS and the other sensors to "calibrate" an external sensor to be more accurate.
FWIW:
http://support-en-us.nikeplus.com/app/answers/detail/a_id/35280/p/3169,3179
I ran a 4m and it was off by .2M and the pace seemed odd. After I turned on the above, I ran the same track today and it was dead accurate.
The instantaneous pace the AWS gives during my runs is hilariously off, like it's not even close, we're talking off by a 1:00/mile or more, in either direction. This is without ever running with the iPhone+watch, just the watch all by itself.
The average pace at the end seems fine but I assume that's just total distance / total time.
Did you have that setting turned off as well? As it essentially turns off the motion coprocessor in the a8. Leading to bad results.
I can't remember ever changing it but on my phone Location Services is on.
I don't see any settings specific to motion calibration and distance, it's just "Location Services on/off" but maybe I misunderstood your previous post.
I've done an 8km run with my TomTom watch as well as the AW + iPhone. It was a known route, and the AW was off by 200m. Was expecting it to be a bit more accurate considering it's using GPS from the phone.
This happened to me too. Many times the pace was way up or way down. I wonder what data points during the run it's using to calculate that instantaneous pace.
This happened to me too. Many times the pace was way up or way down. I wonder what data points during the run it's using to calculate that instantaneous pace.
Are devices with on-board GPS substantially more accurate than this (2.5% total, or 0.025-km drift per km)? I've never used one but that sounds pretty good to me.
Than I think the problem is you need to bring the phone alone for a few runs, so that it gets a chance to learn your strides.
So. The phone should already know the strides? I have ran with my phone for every single run for as long as motion tracking was available on iPhone... Yet Apple watch still showed 0.18 less than nike gps app. I plan to take my next run with a gps watch to compare.
I am still wondering where the GPS data is? If the run is tracking GPS as long as I have my phone why can't I see my route?
I don't think it works that way. The phone only provide the location related information to the artificial intelligence software on the watch.
I've heard that the "pairing" process takes the long history of GPS and steps/acceleration data on the phone and uses this to calibrate the stuff on the watch that's running when it has no GPS.
I doubt Apple is in any rush to release the details so we'll probably have to live in ignorance though
This was pretty close to my experience on my first run. The Workout app was about 0.03 lower than Nike, but I started the Nike app shortly before the workout app. I ran on a flat track so perhaps that helped with the accuracy.
I don't have mine yet tomorrow
but is there a way to make sure the Workout is using the iPhone's GPS and not some calculation based on steps (as it would do if the iPhone was not available)?I've heard that the "pairing" process takes the long history of GPS and steps/acceleration data on the phone and uses this to calibrate the stuff on the watch that's running when it has no GPS.
I doubt Apple is in any rush to release the details so we'll probably have to live in ignorance though
That's unlikely. From what I understand the iPhone by default does not record a list of your gps data because of privacy concerns. It's a feature that need to be opted into by the user and even than it's only use for specific purposes.
Interesting although it's worth noting that I did not run on a flat track but a rather constant up and down route. I actually asked a friend today if he knew of anywhere flat to run! He did not.
I've got location calibration on etc and have done a bike ride and a run with the phone running no other apps. Both activities are over courses of known distance (based on literally dozens of runs/rides with a variety of devices). I was running a garmin edge on the bike and a vivoactive on the run. The bike ride was spot on over 15m, the run was 0.13m short.
I'm not convinced outdoor run is using the GPS at all, whereas I think the bike must be.
I'm not convinced outdoor run is using the GPS at all, whereas I think the bike must be.
I just did a quick run this morning, it measured 3.27km compared to the TomTom at 3.45km. ~5%. Multiple times when I looked at both the pace was varying quite wildly, unfortunately.
It's not a big deal right now, as I wasn't expecting it to be a dedicated running watch, and needing to carry a phone with me isn't something I enjoy doing.
Just checked the exact distances, it showed 7.76km compared to the TomTom watch at 8.01. Little over 3%. 2-3% shouldn't be a big deal, but it throws me off when I've been running these routes for so many years.
And even worse, the pace is always all over the place.
----------
Actually similar happened for me. Did a long bike ride as well on the weekend, the computer on the bike measured identical distance to the Workout app on the Watch/iPhone.
It's not a big deal right now, as I wasn't expecting it to be a dedicated running watch, and needing to carry a phone with me isn't something I enjoy doing.
Just checked the exact distances, it showed 7.76km compared to the TomTom watch at 8.01. Little over 3%. 2-3% shouldn't be a big deal, but it throws me off when I've been running these routes for so many years.
And even worse, the pace is always all over the place.
----------
Actually similar happened for me. Did a long bike ride as well on the weekend, the computer on the bike measured identical distance to the Workout app on the Watch/iPhone.
I wonder if I should try and run somewhere flat then. my runs I have done so far have not been very flat. I'll try this on my next run. though I would think that part of the calibration would be the hill component... Since your stride changes slightly going up or down hills. And the GPS data could easily be used to help adjust the calibration here...