Distance Measurement Accuracy

[Sasha Pachev]

This post is inspired by numerous race reports where the blogger enters the GPS reading for the race distance that happens to differ from the advertised distance even though the course is USATF-certified.

GPS measurement systems such as Garmin Forerunner have a reasonable level of accuracy and reliability, but they provide only an educated guess at the actual distance covered. In fact, it is a miracle they have any degree of accuracy at all. The distance is measured by obtaining the position of the runner at frequent intervals and then performing a computation on how far the runner has moved in between the measurements. A hand-held GPS has the accuracy of position of about 15 meters. Think for a second what that means. Suppose the data is sampled once a second, and your position is off by as far as 15 meters on each measurement. That would create an error that keeps adding up at rates of up to 15 meters per second, that is faster than Usain Bolt can run in a sprint.

Fortunately through a set of very sophisticated error-correction algorithms such devices are able to come up with reasonable distance estimates. But you must always remember that it is an estimate that the device hopes is right. If  Garmin had to pay $10 every time the GPS measurement was more than 5% off the actual distance, they would soon go broke. If we made a requirement of 2% accuracy or pay up, they would go broke much sooner, and a smart runner would be able to get paid $100 a day until Garmin's money ran out. Just do 10x1000 on a windy course in the woods for a guaranteed $100 per workout income.

A GPS measurement is never more accurate than a properly done good old mechanical measurement. This is why USATF does not accept GPS measurements for course certifications, but instead requires a calibrated bike wheel measurement with the help of a rotation counting device known as Jones Counter. While it is possible for a supposedly USATF certified course to have the wrong length (e.g the race director misplaced the start or the finish, the course was measured on tangents while the runners were not allowed to run them, etc), a runner must never question the length of a mechanically measured course based on a GPS reading alone.

[Dallen]

I will mostly support Sasha's claims, but there are exceptions. I have done about 30 separate certified courses in the last 4 years and only 2 were inaccurate. One of these resulted from the race director not appropriately following the certified course and in the other case the course certifier made a mistake in his description of the turnaround  (something like 1st pole north of road instead of intended south). In both cases it was obvious that a mistake was made because everyone had a single bad mile split.

Still this is much better than my experience on uncertified courses which are almost all incorrectly measured to some degree.


Just curious, how accurate/close are your wheel measurements if you measure the same course twice. I would guess at least 10X more accurate than GPS. Probably even better than a track since most of us incorrectly count 4 laps as a mile.

[Matt Konold]

Not speaking for anybody else, but when I entered my garmin distance in the race report for Ogden I did so to record more or less how well I ran the tangents.  I totally agree with you about the accuracy of Garmins can be poor; I've had much experience with that!!  But I kind of like looking back and seeing how well I am running the course (again, as best as mr. garmin can guestimate).  I think we can all tell that when all is said and done, we've all raced the same 26.2, or 13.1, etc.  I think most people understand this vs. what you may be implying that they think that the course is not correctly measured.

While we are on the topic, last year's Ogden course seemed exactly the same (minus the detour on the dirt path for a  portion), however, we weren't allowed to run on the left side of the road down the canyon and use those tangents.  Does anybody know if the certified course was based on running on the right side of the road?  Just curious.

[Jon Allen]

According to Bill Cobler (who certifies courses), Ogden is certified running the best tangents possible, meaning you run on both the left and right side, taking whatever is the best tangent.  He would even run on the shoulder, if it is possible.

Also, some people think that certification mean a course is guaranteed to be a certain length.  Based on my understanding, that is not true.  It simply means that the course is at least that long, and will not be short.  But a certified marathon can be 26.3 miles or more... it just can't be 26.1.

[Paul Petersen]

Very good post, Sasha. I agree 100%.

Yes, a course is certified to guarantee everyone runs a minimum distance. By default, virtually everyone run a little (or sometimes a lot) extra. If you swing out wide to get water at an aid station, you are adding distance to the marathon. And it's very hard to run absolutely perfect tangents. A course with fewer turns will usually have faster times, because there are fewer opportunities to mess up tangents. Add all that to the inherent error of a recreational-grade GPS, and the end-result distance does not represent the actual course. That said, I still find tools like Garmins and GMap to be very useful tools for estimating course distance, particularly for daily running. They are certainly better than what I used before, which was simply a stopwatch and a base assumption for pace. When I started mapping out my old courses, I soon found that most weren't as long as I thought they were!

[AndyBrowning]

How much do you think elevation change contributes to the Garmin error?  It's just one more variable that has to be measured by the GPS and factored in to the distance calculation. 

[adam]

They are certainly better than what I used before, which was simply a stopwatch and a base assumption for pace. When I started mapping out my old courses, I soon found that most weren't as long as I thought they were!

Amen. I quickly learned that a common "4 mile loop" was more like 3.7...

[Paul Petersen]

How much do you think elevation change contributes to the Garmin error?  It's just one more variable that has to be measured by the GPS and factored in to the distance calculation. 

Not much.

[Jon Allen]

Not much.

Paul- what about on something like Logan Peak Run, where you have 7200 ft up and down?  That amounts to almost 3 miles of vertical- does the garmin include that?

[Paul Petersen]

There is very little net effect, even for large vertical gains. I did the math a few years and was surprised to find this out. For Logan Peak, all of the trees, rock faces, and obstructed sky does enough damage on its own. But even a Garmin is still a much better way to measure a trail like that than trying to map over an aerial photo or topo map. The first time I measured Logan Peak, I used topo maps and aerials, and it came up a few miles short, since none of the switchbacks showed up. At least a garmin will attempt to capture that.

[Kevin Frick]

Let me add to the ponderings here.  I am perfectly happy to accept the race distance in a certified course.  How about a local 5K.  I realize the Garmin may not be 100% accurate.  But what if I map out the course on Google Maps or Map Quest and it is different from the advertised couse distance.  I'm just trying to understand how my running is doing and the difference between my "Garmin time" and the true time (or alternatively between the MapQuest/GoogleMaps distance and the advertised distance) is big enough to make me wonder about how I did.

[Benn Griffin]

Online mapping has never been accurate for me, because it is simply impossible to mimik all of the tangents and turns precisely. I have found that frequently the measurements for a 5.5 mile around the block loop at my parents house is off by .1 or .2