Training: Running with a GPS on a track; how accurate are they?

Joe English – 3:51 pm
Training: Running with a GPS on a track; how accurate are they?
Coach Joe English

A reader named Mike G. from New York sent in the following question that has had me doing some testing and research the last few days. Here’s what Mike asks:
I run with a Garmin Forerunner 305 GPS and Heart Rate Monitor unit and I love it. I’ve noticed something kind of weird though. It seems like it says I’m running slower than I am when I wear it on my track. Is there some reason why a GPS would be less accurate when running on a track?

Let’s start with a little background on what GPS monitors are and how they work.

A GPS monitor, such as the Garmin Forerunner 305 that you’re using, tracks your speed and distance when you’re running or riding. It does this by locating your position on the surface of the earth by reference to a series of satellites in orbit around the planet. These satellites, part of the Global Position System (GPS) network send out very precise timing information that can be received by a device on the ground. By measuring the difference between signals coming from the various satellites, the device can detect where you are, where you were, and how fast you moved from place-to-place.

The accuracy of the GPS unit depends on a few things, such as the accuracy of the device itself; whether there are obstructions between you and the satellites (such as buildings or trees); and how often the GPS takes its readings. The Garmin Forerunner 305, according to Garmin documentation, is accurate to “less than 10 meters”, which is pretty good. This unit takes readings of your position once every second, which again, is pretty good. But the documentation also notes that “accuracy depends on view of the sky” and that with a “clear view of the sky” is about 99% accurate or 95% accurate in more “typical” situations.

So in summary, the GPS were talking about measures your position to within less than 10 meters every second with a typical accuracy of about 95%. This should result in a very accurate reading of your speed, especially when you are moving in a straight line. When you’re moving straight ahead, it is very simple for the device to calculate the rate of speed, even if some of the measurements are lost due to obstructions.

However, when you’re running on a track you aren’t running in a straight line all of the time. In fact, when you’re running on a track you are running in a fairly tight circle for half of the time. This complicates things a little. When you’re cornering, the device is measuring your position each second and drawing essentially a straight line between its points of measurement. If there are obstructions that cause it to miss even a few of those measurements, the shape of that curve can get off and my guess was, prior to testing this, that the device might read your speed as slower than it really is.

Before I tell you about my experiment, I found a research article that talks about this in way more technical detail that I can go into here. So if you’re into this sort of stuff, you may want to check out an article in the Journal of Biomechanics called “Accuracy of non-differential GPS for the determination of speed over ground“. At the end of the summary of the article is the following line that seems to square with my thinking on the subject:

GPS data loggers are therefore accurate for the determination of speed over-ground in biomechanical and energetic studies performed on relatively straight courses. Errors increase on circular paths, especially those with small radii of curvature, due to a tendency to underestimate speed.

I also found another article on the web where a runner tried a similar experiment to mine and found that on two different tracks, with differing views of the sky, he had different results in terms of his distance covered and speed. He found that the accuracy of his GPS was off by about 16%.

Timing yourself on a track
Before moving on to the experiment, let’s talk first about how to time yourself on a track and how useful it is to run on one. I’m a big fan of the track for learning to pace yourself, because you can get feedback on your pace every time you pass a particular point on the track, whether that be 100 meters, 1/4 mile, 1 mile, 5 miles — you get to decide. All you have to do is count the laps and you know how far and how fast you’ve run. The finer the increments you break down your workout into, the more precisely you can practice pacing technique.

I prefer to visually check my splits every 400 meters (each lap) and I record a lap time on my watch every 1,600 meters (every fourth lap). This way I can compare mile to mile and I know how I’m doing within each mile if I’m falling behind or getting ahead of my pace.

When you time yourself on the track, be sure that you know the distance of the track to start with. Most running tracks at high-schools and colleges are 400 meters around in the first (inside) lane of the track. That means that you 1,600 meters every four laps, which is very close to a mile.

To be precise, however, 1,600 meters (4 laps) is not exactly 1 mile. There are 1,609.3 meters in a mile. This is very close (less than 10 meters difference), but if we’re trying to be very accurate then we can adjust this distance by doing the following math. Take your per lap speed in seconds and divide that by 400. This gives you your speed per 1 meter. Then multiply that number by 9 to see how many seconds off your mile splits would be. I’ll show you in an example using 6 minutes per mile:
6 minutes per mile = 1:30 or 90 seconds per 100 meters
90 seconds divided by 400 meters = 0.225 seconds per meter
0.225 seconds per meter multiplied by 9.3 meters = 2.09 seconds per mile

So if you were hand timing yourself running miles on the track, the pace would be off by roughly 2 seconds per mile. That’s really very close, but for sake of argument, let’s be accurate.

The experiment
I went to a 400 meter track and wore my own Garmin Forerunner 305 on one wrist and my regular watch on the other. The track that I run on has trees around it on one side and trees in the middle, so the view of the sky is somewhat obstructed, but not badly. I ran 7 miles aiming for 6 minute pace for the sake of easy math (meaning I had 1:30 splits on the 400, 3:00 on the 800, 4:30 on the 1,200 and 6:00 on the 1,600). I recorded the following splits on both devices:

gps hand timing table 2 Training: Running with a GPS on a track; how accurate are they?
Table: GPS vs. Hand-timing

The Garmin surprised me on the first lap by not capturing the first lap split until I was nearly 75 meters into the first turn of lap 5. This showed me that the calculation of distance, and therefore speed, was going to be slightly off. Each mile that went by, the new lap started further into the first lap of my next mile. When I finished the run, which I know to be precisely 6.96 miles the GPS had recorded 6.66 miles and the pace was slower than both my hand-timed and mile adjusted numbers.

Bottom Line
The bottom line is that when running on a circular track, your GPS device is not going to be as accurate as the measurements that you can take from the distance that you know that you’re running on the track itself. Your results from your GPS will vary depending on the view of the sky at your track, the speed that your traveling on the corners, and the accuracy of your GPS unit. While GPS devices are very helpful out on the road, the track is a distinct place in that we know its length to a certainty. Take advantage of that aspect of the track to capture your own hand-timed data on the track and adjust it if necessary for the 9 meter difference between the 1,600 length of the track and an actual statute mile.

Good luck out there runners!

Coach Joe English, Portland Oregon, USA
Managing Editor, Running Advice and News

share save 120 16 Training: Running with a GPS on a track; how accurate are they?
Filed under: Running,Training Tags: Accuracy, fitness gear, Fitness GPS, GPS, Running, Speed and Distance measurement
2 responses so far, want to say something?
1. Dean HebertDecember 8th, 2008 at 6:15 am

This is an excellent article. Though I have not done a test like this, it does support what I believe happens on the track. The bottom line is that GPS systems really should not be used on the track. Save them for the open roads, canals and trails.

2. JoeSeptember 1st, 2011 at 7:23 pm


Thank you so much for this article. This confirms my experiences with 2 different GPS devices – one an old Garmin handheld from the early 2000′s and my new Garmin Forerunner 405. The only differentiation I have between my results and yours is that I believe it is inaccurate on every course except straight line running. As an experienced runner, and you know what I am talking about giving your experiment above, I know when I am running a 7 minute mile pace, a 6:50 mile pace, 6:40, etc. I can guess what pace I am running to within 10 seconds a mile every time, and within 5 seconds most of the time (all of this, on flat surfaces). The only time I don’t accurately estimate my pace is when I am running slower than 7 minute miles.

My experience has been that the old GPS device seemed around 10% off, measured against trail markers or car measurements, most of the time, even in mostly open spaces. The only time my old GPS seemed accurate was when I went running at the beach one time and found it was registering my probable pace correctly. As for my new 405, I am finding the mile splits to be off as well, sometimes upwards of 20-30 seconds. This is on greenways and trails (though I’ll concede that my perceived trail speed could be inflated:-)), and also on roads. What I have found to date is that it seems around 4% off, which is still a lot when talking about 400 seconds/mile.

I could understand the difficulty of the GPS providing accurate measurements on a track (though, with a 13 ft inaccuracy, it should provide readings of being 13 ft outside the loop (making the lap longer, and speed faster), as well as, 13 ft inside the loop (making the lap shorter and speed slower). My mind’s eye can only see a jagged-lined loop track line, but hopefully it would average out (?) to new a smooth line assuming the software on the device does some sort of leveling.

My issue is with everyday road or greenway running – my 405 still seems to get it wrong, giving me splits that I know are 10 seconds or more off. It seems that the accuracy should increase as the GPS takes the starting point, +/- 13 ft for the range of inaccuracy, to any point down the line. As the distance increases, the 13 ft (26 if you add inaccuracy on the far end in a worst case scenario) becomes less and less significant for the whole of the run, and in the end, taking the total time divided by the total distance (possibly 26 ft off and assuming the jagged lines smooth out). So, might I find that, if I ever download the runs to my PC at home, they will say my average pace will be significantly faster than the splits they give? I surely hope so.

This discussion was precisely what I was searching for, and I do appreciate your article, even 3 years later.