Does arm swing affect the accuracy of a GPS sports watch while running?

December 17, 2009 10:27 PM

I'm training for a marathon and I use a Garmin GPS watch. I'm wondering: of course it transmits to the satellites as I move forward; but does the fact that my arm (with the watch attached) moves back and forth - as opposed to my entire body moving forward as a single unit - have anything to do with the accuracy? I apologize in advance if this sounds vague or else is even irrelevant.
posted by holdenjordahl to Health & Fitness (11 answers total)

Best answer: No, GPS is not really that sensitive, especially in watch form. A GPS only updates about once per second, and the best accuracy it can get is still down to within a few feet, not the 12-18" arch that your arm moves in. Your arm motion will be averaged out of the picture.
posted by autojack at 10:32 PM on December 17, 2009

Response by poster: great, thanks, autojack...
posted by holdenjordahl at 10:34 PM on December 17, 2009

(Also, your device doesn't transmit to the satellites -- your watch only receives a signal from them.)
posted by xil at 10:35 PM on December 17, 2009

Just repeating for posterity- GPS never transmits anything to the satelites. It is completely passive.
posted by gjc at 10:55 PM on December 17, 2009

Newer automotive and cell phone GPSes will augment the readings from the satellite signal with an accelerometer. It's conceivable that if you were using a standard GPS unit (maybe like the one in an iPhone), your flailing arms would trash the accelerometer reading, and you might get a slightly less accurate trail.

Your GPS watch almost certainly isn't fancy enough to have accelerometers at all, and even if it did this problem would probably be too small to care about.

I am not an expert on GPS but once wrote some software to integrate GPS and accelerometer readings
posted by miyabo at 11:10 PM on December 17, 2009

Here is an article about the accuracy of the watches - it gives you an idea of accuracy. As has been stated above, the accuracy you require to measure arm swings just isn't there.
posted by defcom1 at 2:48 AM on December 18, 2009

Also, not to worry you unduly, but I recently ran a marathon with a Garmin Forerunner and found that it regularly underestimated each mile by about 6% (as in, it told me I'd finished a mile when I'd only gone 0.94 on the trail). This is entirely speculative, but I suspect that--since I live in the Mid-Atlantic U.S. and ran in Ireland--changes in latitude may throw it off somewhat.
posted by kittyprecious at 8:15 AM on December 18, 2009

Response by poster: actually, kittyprecious, i've run races from 5Ks to ultras and each time i've wondered (speculated) whether the mileage markers were correct...i wonder if they use the same kind of GPS watches for those.
posted by holdenjordahl at 9:52 AM on December 18, 2009

I've heard plenty of stories of misplaces mile-markers, but all the same I'm more troubled by the fact that my watch told me I was "done" a full mile before the end of a certified-length course.
posted by kittyprecious at 10:29 AM on December 18, 2009

Kittyprecious - Are you sure you ran the certified route - i.e. exactly on the "blue line"? If not, you did run more than the official 42.195km. Add to that the inherent inaccuracies of the GPS, and your deviation over the course isn't that unexpected.

If a race has a "certified" distance is has been measured by a trained operator, usually using a trundle wheel type of measurement tool.

Most GPS watch manufactures only claim 5% accuracy, but I think normally they are a lot better than that. I think even 5% is pretty great for a tiny little device sitting on your wrist picking up signals from a satellite 20,000km above the earth transmitting with the same amount of power as a 50watt light bulb.
posted by trialex at 2:43 PM on December 18, 2009

Best answer: Well, one thing I think causes problems with GPS distance measurements is that because it's taking a limited set of data points (1 per second or less, usually) and interpolating your route from it, instead of the absolute measured distance the polygon it generates will (IIRC) always be inside the actual route, and therefore slightly shorter. And that's assuming perfect accuracy, which isn't a given.

(Sort of like this: Archimedes approximation of Pi.)
posted by Kyol at 9:56 PM on December 18, 2009