Race distance and GPS distance don't match. Which is correct?

Factual Questions
1

Just finished running the Woodrow Wilson Bridge 1/2 Marathon this morning (1:44:53 if you’re wondering, goal was 2 hrs, destroyed it! Back to the question though).

When I crossed the finish line my GPS showed 13.25 mi. On the web site the official time is the same as the time recorded on my GPS, though my GPS thinks I ran 13.25 and not 13.1.

Which is correct? That seems like a fairly large error by either my GPS or the race measurement. I know the GPS is not perfect and has error built in but I don’t expect that much of a delta.

Data point to consider: When I was running past the mile markers they matched my GPS all the way up to marker 11. This has me thinking that the last stretch of the race was incorrectly laid out.

Oh well, I’m happy with my time!

2

I’m assuming it’s a certified course(website didn’t say) but since your GPS and course markings agreed up to a specific point, I’d vote for a screwed up course over the last portion.

I use a Garmin 305 on my bike and it’s dead on compared to Google Earth and each time I ride the same route, it might vary by .01 mile-sometimes.

And your time converts to 1:43:48.

ETA:terrific time!

3

I use a Garmin 305 and see about the same “extra” distance as you do on half-marathons (I don’t complete them nearly as fast as you do though!). I thought that the error came from the path that I chose through the corners. I don’t know how a course’s distance is measured, but in a turn, if you take the inside line it’ll be shorter than the outside line, and if the course is really wide, there can be a significant (I’m just guessing) difference. Maybe your “line” averaged out to match that of the course markings for the first 11 miles, but there were more or different or wider turns in the last two miles where you didn’t run the “average” line or whatever line was used to determine the distance of the course?

4

ETA: simulpost with treadstone

15% of a mile (13.25 - 13.1) is 792 ft. Were the last stages of the course particularly curvy? If they measure the official distance along the shortest possible in-bounds path and you ran those last two miles more on the centerline or longer path, that could account for some of the distance. Over just 2.1 miles (from MP 11.0 to the finish line) the 792 ft is about 7% of the distance. Were you exhausted and wobbly?

Was the location such that GPS signal propagation would have been lousy at the end? i.e. perhaps you had several satellites in good view until those last 2 miles in an urban canyon or running under an elevated freeway?

How far off were you at MP12 or MP13?

5

Thanks for the kudos Pat. I’m proud of that time. I was only trying to break 2hrs and came in well under.

treadstone I thought the same thing and would believe that if it was gradually more and more off during the race but .15 difference in at most 2 miles? I posted a comment on the Facebook page of the race and several others chimed in after me to say they had the same issue. One lady said she ran it last year and the finish line was noticeably further away this year.

I’m not a competitive runner or anything so I’m not too torn up by it. I wouldn’t even have mentioned anything if it hadn’t matched so closely for the first 11+ miles (maybe even 12, I didn’t look at it when it hit that marker).

6

There were more turns in the last 2 miles or so but nothing excessive. I didn’t look at the GPS at MP 12 but I was already showing 13.1 before I hit MP 13. GPS wasn’t conking out or anything.

7

Was that last stretch hillier than the rest of the track? (I guess not, that’d just seem like spite towards the runners, but still…)

The problem of estimating vertical distance is somewhat ill-posed due to the same reason that the problem of estimating coastline length is: the measured distance depends on how fine grained your data points are, because the geometry is essentially fractal. This is known as the coastline paradox, and the difficulty in measuring vertical distance is exactly analogous (here’s an article discussing the problem).

So if the last stretch had a greater portion of vertical distance, then it might be that GPS measurements don’t agree with what you get if you’d just lay out a measuring tape.

8

The course profileshows a climb from 11-12 miles and an equal drop to the finish. There is a similar climb/drop(though less steep) mid race.

9

GPS-measured position seems to have stochastic variation that seems to systematically overestimate the distance, which matters more if you move at running as apposed to e.g. biking speed.

I use a Garmin 305 for running, and this is the track I recorded when I ran a half marathon for training last year, on a 500 m long wood chip track that is about 1 m wide. The GPS device measured the 500 m laps at anything between 505 m and 520 m.

Another datum: On bike tours the GPS accords well with my bike’s odometer, but when I did not pause the GPS device during a beer garden break of about 1.5 hours it said I had moved a total of about 1 km during that time. Apparently this is because, sitting still, random measurement errors make the GPS think you are running around randomly (in this case, actually running into and out of the Neckar river…).

10

I keep GPS data from hiking, and do my own processing of the track data.

I remove the idle time and distance via a simple threshold on the speed for a given leg. It seems to work well. I also compute the map projection, and actual 3D distances.

Example (with abnormally accurate altimeter data - there’s usually some drift):



Generated Sun Oct 02 11:59:49 PDT 2011
Unit System: English
Idle < 0.4 mi/hour
Short Rise/Fall over 0.1 mi
Waypoint distance within 0.5 mi

Waypoint "Blowdown": (37 13.36 N, 122 17.54 W, 1450 ft)
Waypoint "Can Crk": (37 13.16 N, 122 18.71 W, 940 ft)
Waypoint "Doe Ridge": (37 12.79 N, 122 18.07 W, 1070 ft)
Waypoint "Landing Strip": (37 13.52 N, 122 17.83 W, 1669 ft)
Waypoint "Olmo FR": (37 12.93 N, 122 17.88 W, 1150 ft)
Waypoint "Trail Camp": (37 13.48 N, 122 17.45 W, 1553 ft)
Waypoint "Turnoff?": (37 13.15 N, 122 17.75 W, 1240 ft)

Track: ACTIVE LOG 001, Sat Jan 08 10:02:43 PST 2011
	Bounds: (37 11.86 N, 122 20.44 W, 204 ft) / (37 13.62 N, 122 17.33 W, 1681 ft)
		NS Distance: 2.031 mi
		EW Distance: 2.847 mi
	Projection: Cylindrical Equidistant projection, latitude = 37 12.81 N
		North Edge Distortion: 0.0180%
		South Edge Distortion: -0.0210%
	Start: (37 12.13 N, 122 20.37 W, 215 ft)
	End: (37 12.12 N, 122 20.36 W, 216 ft)
	High Point: (37 13.61 N, 122 17.86 W, 1681 ft) at 6.422 mi, 2:19:41
	Low Point: (37 12.09 N, 122 20.34 W, 204 ft) at 12.376 mi, 5:19:35
	North Point: (37 13.62 N, 122 17.86 W, 1680 ft) at 6.409 mi, 2:19:22
	South Point: (37 11.86 N, 122 20.00 W, 939 ft) at 11.696 mi, 5:00:22
	East Point: (37 13.42 N, 122 17.33 W, 1500 ft) at 7.465 mi, 3:21:22
	West Point: (37 12.13 N, 122 20.44 W, 251 ft) at 0.093 mi, 0:01:35
	Length, 3D: 12.36 mi, 0.065 mi idle length subtracted
	Length, map: 12.228 mi, 0.064 mi idle length subtracted
	Time: 5:21:53, 0:50:54 idle
	Speed: 2.3 mi/hour, 2.7 mi/hour moving
	Altitude Difference, start to finish: 1 ft
	Altitude Difference, low to high: 1,477 ft
	Total Rise: 3,312 ft over 5.733 mi = 10.9% (6.2 degrees)
	Total Fall: -3,326 ft over 6.626 mi = -9.5% (-5.4 degrees)
	Steepest Short Rise: 17.8% (10.1 degrees) at 5.017 mi
	Steepest Short Fall: -41.9% (-22.7 degrees) at 4.072 mi
	Waypoints:
		98 ft from "Can Crk": (37 13.15 N, 122 18.69 W, 896 ft) at 2.697 mi, 0:54:05
		41 ft from "Turnoff?": (37 13.15 N, 122 17.76 W, 1133 ft) at 4.933 mi, 1:45:42
		27 ft from "Blowdown": (37 13.35 N, 122 17.53 W, 1380 ft) at 5.39 mi, 1:57:56
		Through "Landing Strip": (37 13.52 N, 122 17.83 W, 1674 ft) at 6.792 mi, 3:07:06
		Through "Trail Camp": (37 13.48 N, 122 17.46 W, 1542 ft) at 7.281 mi, 3:16:32
		49 ft from "Olmo FR": (37 12.92 N, 122 17.88 W, 1193 ft) at 8.495 mi, 3:45:51
		Through "Doe Ridge": (37 12.79 N, 122 18.07 W, 1133 ft) at 8.742 mi, 3:51:29
11

Congrats on the race! 1:44 is well beyond crushing 2:00. Very nice!

As far as the course goes, assuming it’s certified, a few things probably happened. First, you probably didn’t cut the corners. The race designers did. A certified race is measured from the absolute shortest distance that can possibly be taken. It’s hard enough to get those perfect when you’re running by yourself let alone in a pack of other runners.

Secondly, as covered above, GPS can have accuracy issues. It doesn’t account for elevation. It jitters.

Third, the course could be a bit long. As I recall, certified courses are “at least” the distance, not necessarily “exactly” the distance. Obviously they aim for the latter, but hitting the former isn’t unheard of.

The combination of a tad long, a bit of elevation, and jittery readings can easily account for the distances discrepancy.

For what it’s worth I keep track of PRs for certified distance courses, and for a given course itself. It bothers me to no end when a race changes course design every year. I want to measure my performance on the same course!

12

Y’know, a bunch of folks at CERN are also asking the same question in the subject line.

13

What is CERN?

14

CERN is the high-powered particle physics laboratory in Europe.

They recently published some results which (on one interpretation) called into question the whole bit about Einstein, Relativity & the speed of light. One of the common soundbite factoids about GPS is that GPS had to be designed with Relativity taken into account. So in a sense, the fact that GPS works correctly is evidence that Relativity is a correct explanation of the universe.

So an experimental result which buggers up Relativity would be popularly expected to bugger up GPS as well. Hence Chronos’ joking reference to CERN being confused by their GPS readings.