Wheel measurment question

When this GPS questioning of courses first reared its head, RRTC was asked to formulate a statement on it, basically to assist race directors who were getting accused of having long courses by GPS-wearing runners.

As part of this, I called the tech guys at Garmin who noted that their consumer models were accurate to within plus or minus something in the range of 3-5 meters (commercial and military units are more accurate, but either expensive or unobtainable). If that variation is carried over the course of even a 5k, it would account for the discrepancies.

If you're interested, there's a lot of discussion on the Measurement Message Board

http://measure.infopop.cc/eve

and a nice experiment was done by a surveyor in Australia

http://www.cage.curtin.edu.au/%7Egordonsj/course/

I've used both the 60csx and the Forerunner on trail routes through the woods on winding trails, and the accuracy relative to the Jones is the same. Previous experience has shown me that the Forerunner has problems with tall buildings like you said, don't know about the 60csx.

Your examples of people giving you bad results from GPS units and online mapping sites only proves that any tool is inaccurate if used incorrectly. If someone makes an error calculating the calibration constant for their Jones, people don't suddenly declare that Jones counters are all inaccurate.

I believe we, as course measurers, need to stop addressing the issue of people telling us our courses are inaccurate based on their Forerunner, by simply declaring that GPS devices are not as accurate as the Jones. Everybody knows that and it's pretty much irrelevent. They aren't questioning the accuracy of the Jones, they are questioning the accuracy of the course. If the course is actually .2 miles off it's not because the Jones is inaccurate. It's because the measurer made a mistake, and today's GPS units are capable of detecting such a large mistake.

If someone told me after a race that their Forerunner said the certified 5k was 3.3 miles, I'd tell them to ask 3 other Forerunner wearers if they got the same thing. If they find 3 who did, I'd be worried.

[quote]JimG wrote:

As part of this, I called the tech guys at Garmin who noted that their consumer models were accurate to within plus or minus something in the range of 3-5 meters (commercial and military units are more accurate, but either expensive or unobtainable). If that variation is carried over the course of even a 5k, it would account for the discrepancies.

Here's a direct link to the measurement forum thread Jim mentioned.

http://measure.infopop.cc/eve/forums/a/tpc/f/508102977/m/7421068171

If you download a track from a GPS unit you don't see the points bouncing around 3-5 meters. As I mentioned before, that's probably because most of the 3-5 meter error is bias error, and has no effect on a distance calculation since it's the same for all points. My downloaded tracks often show me riding a few meters to the left or right of the trail I was actually on. But it shows that same shift for the entire ride.

Here's another thread from the measurement board that includes data from experiments. Only one other person posted data from distance calculating experiments (On the thread Jim mentioned) and his error was similar.

http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/8301003542

If you're really interested in how courses get certified, and the methods used, check out:

http://measure.infopop.cc/eve/ubb.x

It's the Road Course Measurement Bulletin Board, and it's a prime place to get the real dope.

I cannot understand why some , a minority I think, of GPS wearers do not realize that it is almost impossible to run the absolute shortest distance in any race.

I have raced with a Garmin since 2005 and I normally expect

to get a reading of 6.25 miles in a 6.21 ( 10K ) and around

26.41 in a marathon which is pretty close to 26.21.

I expect the above and realize that a 6:43 pace on my Garmin means I am really running at a 6:46 or 6:47 pace.

survival wrote:

Common sense would dictate that GPS and esp. on-line course measurements would be short.

But, after reading this board, most of the course measurers' experiences say otherwise. In your common sense assessment, you're still assuming that the update points are perfectly accurate. If they were, then I imagine that you'd be correct, however, if there is a random error of between 3 and 5 meters, then the straightaways would measure long with the GPS. One poster mentioned this earlier. You end up with a slight zigzag, which will also add up, and there is usually much more straight road in a race than there are turns.

dukerdog wrote:

If someone told me after a race that their Forerunner said the certified 5k was 3.3 miles, I'd tell them to ask 3 other Forerunner wearers if they got the same thing. If they find 3 who did, I'd be worried.

I wouldn't. In fact, I'd expect the other 3 to agree. From what I've read, some GPS errors tend to be a function of local position. They can be caused by, among many other things, regional atmospheric effects as well as reflections of the GPS signal off of surroundings (including, potentially, the ground).

At Mt.SAC, they don't need the wheel. The cross country course is now measured by how much money they bring in every year. The course gets shorter so everyone gets a PR. Then when long standing records fall, you even get free publicity. And then more people will come. Now they even charge for parking to sqeeze more money out of you.

How long until the Mt. SAC race is just a START/FINISH line with no distance in between?

I have measured about 100 races. As USATF measurers know, I normally must ride a course in both directions. I ride with a Garmin Legend HCx, which has a high-sensitivity antenna. However, when riding on a paved bike trail, riding the shortest possible route, the downloaded track will periodically show my route separating by up to 8 feet, even though the trail is only 6' wide, and my ride is within 1' of the other direction. GPS devices are fine for "pretty close" work, but certainly not for precise work. I use mine as a support tool, but it is not what is used to certify the course.

When runners have questioned my courses, I will re-visit the course and verify. I have yet to find a mis-measured course, but have found the course to be laid-out incorrectly. As most courses are measured at least 3 times, it is pretty hard to repeatedly make a mistake of the magnitude that GPS devices report. It is simply GPS inaccuracy - and I don't make that comment as a denigration to GPS devices.

As stated by someone earlier in this thread, if a course is not certified, it is a fun run. With that in mind, for 2008 and beyond, I am building my Website to list as many CERTIFIED races around the country as I can. If anyone knows of certified races in '08, please email details to

info@RaceMeasure.com

. This way, runners can find a certified race in their area, so they can be confident in the accuracy of the course.

Also, if measurers would like to participate in RaceMeasure, please go to RaceMeasure.com and read the Measurer's Info, and let me know if you would like to join us.

I don't know of any measurer that does not take this job seriously, so I hope OldXCGuy has received sufficient input to his posting to answer his original question. Measurers can make mistakes, but our entire procedure (including having our work reviewed by the area certifier) is designed to discover and correct errors before we submit our paperwork. Not impossible, but I believe it is unlikely that a certified course is incorrect AS MAPPED. Can't say the same for the layout on race-day.

I borrowed a Jones Counter once and was totally disappointed in the measurements. I probably had a bad device/unit. It seemed that I had to ride the bike very slowly - like maybe 8 mph to get somewhat consistent readings. Is there a prescibed speed that is needed with these units to ensure accuracy?

dukerdog,

I understand your argument about bias error, but from some of the plots that I have seen, I still think that the positional error I mentioned on the prior page (zig-zag effect) plays a role. Take a look at the points that were plotting using GPS from the Australian surveyors link. There are clear errors there. The real question of interest is how the GPS units deal with those errors. I imagine that if everything seems to be falling in a straight line (i.e., that zig-zag is within the expected range of uncertainty), you could have the software just assume it's a straight line. This is problematic though if someone is running along a trail and going up and down switchbacks. Over-smoothing in this scenario will give a short reading rather than long. As Garmin has not published their methodology, none of us really know what's going on inside the receiver unit, and what (if any) data massaging has been applied to the points it is reporting to you.

One of the guys I run with regularly wears a Garmin (the boxy looking one) and it almost always indicates that our courses and races are long. This includes certified courses that I have double-checked (with very good agreement) on Google Earth.

As I am new to this game, I will defer to the judgment of experienced measurers here.

Marty Moose wrote:

I borrowed a Jones Counter once and was totally disappointed in the measurements. I probably had a bad device/unit. It seemed that I had to ride the bike very slowly - like maybe 8 mph to get somewhat consistent readings. Is there a prescibed speed that is needed with these units to ensure accuracy?

Hmmm. The Jones counter if installed correctly will measure correctly at any speed. I routinely ride quite fast on my second ride because I feel it aids in maintaining a sraight line. No problems with accuracy.

Sounds like you're was either defective or skipping spokes. I zip-tie mine to the spokes to make sure that never happens.

RaceMeasure wrote:

Not impossible, but I believe it is unlikely that a certified course is incorrect AS MAPPED.

I agree. But occasionally it does happen. One such case was reported here in a different thread. A quick measurement with Google Earth or other mapping tool left little doubt that the course as mapped was not the advertised distance. For that one, it was probably a case of an incorrectly drawn map, not an incorrectly measured course.

Jim Fiore wrote:

As Garmin has not published their methodology, none of us really know what's going on inside the receiver unit, and what (if any) data massaging has been applied to the points it is reporting to you.

Or, for that matter, that they are reporting to you all the information the unit is receiving. They output a track to you, but there's no reason to believe that's all they are using to calculate distance. I've downloaded tracks that show a bit of zig-zagging and quite a bit of corner cutting, yet the distance the Garmin reports is within 0.1% of the Jones measurement.

These devices have to use a lot of very complex mathematics to perform their basic functionality of finding you on the earth's surface. It's a little hard for me to believe they calculate distance simply by connecting the dots.

Has anyone stepped forward to claim they measured the 2km

XC loop at Indiana State University?

Remember the Pre-Nationals race back on Saturday the 13th?

Does anyone know who certified it?

I am not the only one doubting or skeptical when I see

dozens of women in the 6km run incredible PR's and

then lots more guys in the 8km run record times.

Does NCAA follow USATF and IAAF rules and regs. for things like SCPF - Short course prevention factor?

I have seen way too many stressed out coaches at the last minute walk a course with a wheel in hand and see the wheel spinning while they are standing still and yet they declare the course close enough.

Can someone shine some light on that 2,000 meters in question? Or shouldn't it be a minimum of 2,002 meters for each loop?

Just seems a tad suspect when these women ran faster 6km times than Vicki Huber, Sonia O'Sullivan, Carole Zajac,

Jennifer Rhines, Kathy Butler, Amy Skieresz, Carrie Tollefson, Katie McGregor, Erica Palmer, Kara Grgas-Wheeler, Tara Chaplin, Shalane Flanagan, Kim Smith,

Johanna Nilsson or Sally Kipyego ever did during their collegiate careers.

Cross-country courses are not certifiable.

history professor wrote:

I am not the only one doubting or skeptical when I see

dozens of women in the 6km run incredible PR's and

then lots more guys in the 8km run record times.

Does NCAA follow USATF and IAAF rules and regs. for things like SCPF - Short course prevention factor?

I have seen way too many stressed out coaches at the last minute walk a course with a wheel in hand and see the wheel spinning while they are standing still and yet they declare the course close enough.

Can someone shine some light on that 2,000 meters in question? Or shouldn't it be a minimum of 2,002 meters for each loop?

For the most part, cross-country course accuracy is, and always will be, an inexact science. An SCPF for cross courses...haha. You waste energy worrying about how long the courses are.

Jim --

I own a 100 meter steel tape but I can't find it. If I ever do, it's yours. It's an "add-meter" style -- the body of the tape is marked only every 5 meters (brass soldered to the steel). Then on one end is a fully-graduated (5 meters by mm) segment. To use it the person at one end holds one of the 5m marks over the point, and the other person uses the graduated end. Then somebody gets to add the two, thus the name. In fact the tape is 105 meters long. It's very narrow, but durable. I used it while land surveying in the Solomon Islands rain forest and it held up nicely for years.

On the matter of GPS units. I have several. From the point of view of a surveyor, every one I have (and every one mentioned here) is a consumer/hobby unit. But that's appropriate! If you want to get into GPS as a land surveyor would, then you're into "differential GPS" at a minimum, and your unit is going to cost you a lot more than my Garmin 60cs cost me.

One problem I haven't seen mentioned in this thread has to do with hills (or even undulations). A bicycle with a Jones counter is always following the road surface. A distance measured via EDM (for example) does not. If the road is perfectly even, then the EDM measurement should be the same as the calibrated bicycle measurement. But if the road surface rises and falls, it won't be -- the calibrated bicycle measurement will be longer. I used to use a calibration course that was measured by EDM, but the road undulated. I used it anyway, having figured out that the different was not great.

What this has to do with GPS-measured courses is that a GPS unit which does not or cannot compensate for hills will give a short course, other things being equal. Imagine starting on the flat, running up a hill, down the other side. The GPS unit is really dealing with latitude/longitude pairs, and it calculates distance by calculating where it is in relation to where it's been.

A Garmin rep told me recently that their units attempt to compensate for this by factoring elevation changes as best they can, to estimate travel along the actual route. But -- as anybody with a GPS should know -- ordinary units do elevation poorly. If you doubt this, take your unit to the seashore or to a point of known elevation. My 60cs has a barometer and thus is better than most, but it's still not very good. Why, you ask? For a GPS unit to do altitude as well as it does lat/long, it would need a sat on the other side of the earth, and that's impossible.

I like my units (including a Garmin 205) and use them all the time. But they don't come close to a calibrated bicycle. I stopped measuring courses for certification ten or fifteen years ago. If I started again I'd always have a GPS unit with me because it protects from blunders.

And as for starting races at the correct points -- when I was in the timing business I routinely saw race directors not able to remember (or find) the points, or shifting them for one reason or another. It was rare for the director to have the certification map on hand. All this is bad behavior but it happens.

dukerdog wrote:

These devices have to use a lot of very complex mathematics to perform their basic functionality of finding you on the earth's surface. It's a little hard for me to believe they calculate distance simply by connecting the dots.

I always assumed that they used some basic curve fitting/smoothing like a spline or simple data averaging (in this case I think averaging might make more sense than a spline). Straight up point-to-point does sound rather goofy considering.

The hill thing has been covered in other threads. Even for the hilliest course someone might consider running, the effect of assuming zero elevation change on a total distance calculation is negligible.

Runtime,

I disagree with the barometric altimeter being more accurate. I had two Garmins with GPS altimeters, then got a barometric model. Measuring courses where the Start and Finish are the same line, the barometric gave me up to 15'-different elevations for the same point, regularly. These readings were only 15 minutes to half an hour apart, so it's not that the weather changed that much. I have since returned to GPS-altimeter models.

The problem now is, these high-sensitivity antenna models correct their elevation so often, it seems to double the gross gain in a course. This is another example of why GPS units should be viewed as a fairly-accurate but imprecise tool (unless you spend the big bucks for a professional model), to be used for reference and support, but not the absolute authority.