This is a great article. I always get a little frustrated when runners claim that a course is "long" or "short" just because their GPS says so. This article explains very well what exactly is involved in course certification, and the limitations of GPS. Well worth the read.
Kilgore Troute wrote:
This is a great article.
Except that most of what he says about GPS is wrong.
dukerdog wrote:
Kilgore Troute wrote:
This is a great article.
Except that most of what he says about GPS is wrong.
Oh, please enlighten us.
I've been through this so many times I'm tired of writing it and I'm sure people are tired of reading it. If you really want to read more you can go to the following.
http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/8301003542
http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/9611072113
http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/5661053823
http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/76210594
dukerdog wrote:
I've been through this so many times I'm tired of writing it and I'm sure people are tired of reading it. If you really want to read more you can go to the following.
http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/8301003542http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/9611072113http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/5661053823http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/76210594
I read the first two of those links and, while interesting, they don't seem to provide much explanation (other than the tangents stuff) beyond the data...
dukerdog wrote:
I've been through this so many times I'm tired of writing it and I'm sure people are tired of reading it. If you really want to read more you can go to the following.
http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/8301003542http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/9611072113http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/5661053823http://measure.infopop.cc/eve/forums/a/tpc/f/89510622/m/76210594
Interesting links, and I think there are benefits to the work you are doing. Thanks for posting.
It seems to me you guys are saying much of the same thing: Basically, the biggest cause of the discrepancy is probably "user error" (not running tangents, etc.)
However, your links don't support your original bold statement at all.
Now, some thoughts:
1.) I read every word in your links. Then I re-read the article. Maybe you should re-read the article in the OP. You said "except that most of what he said about GPS was wrong". He makes statements like: such & such CAN contribute to GPS inaccuracies; and basically GPS is not as accurate as the traditional certified course measurements.
2.) I'd encourage you to use the Forerunners for your tests, and to run with them. To use a different bike-mounted GPS introduces two variables.
There is also a common-sense test: many of us can think of many occasions when we've run side by side with two Garmin-wearers, and at the end they have two different distances measured.
Again, if there is anything specifically incorrect in the OP's article, I'd love to hear it. It seemed to explain both methods quite well. Much of the GPS info listed POSSIBILE causes of error, or quoted accuracy statements directly from Garmin.
It's conclusions agree with yours - beware the user error.
Bottom line, to say "most of what he said about GPS is wrong" is... well, wrong.
The races should paint a bright yellow line along the whole course that shows the exact "tangents" they are using to measure the course. Hiding behind this "tangent" argument seems to be a baseless cop out on the races part.
Of course, if you ask, every runner thinks they are the GREATEST at running all of these "tangents" much better than the next person!!
In the future, gyroscopes might be the most accurate form of measurement (scientifically precise gyroscopes -- not the footpod kind):
http://gizmodo.com/5658408/gyroscopes-the-size-of-sand-will-track-you-anywhere-even-underground
Wouldn't that be cool?
And anyone know how accurate Google Earth is (meaning a link that states its accuracy)? Mainly for a training perspective. I'm always curious if my neighborhood loop is actually the length I think it is.
Where is the Steel Tape dude?
K Theory wrote:
And anyone know how accurate Google Earth is (meaning a link that states its accuracy)? Mainly for a training perspective. I'm always curious if my neighborhood loop is actually the length I think it is.
I'm the USATF/RRTC rep for MI & have been since '86. I felt the article did a nice job of explaining GPS & course certification rules. Very early on ('06) I concluded that GPS users were getting #'s that showed a certified course to be long on a roughly 1 part in 100 basis (for every 100', GPS shows 101'). Everything I've seen since confirms my '06 conclusion. I'm always pleased to hear when GPS wearers share #'s that show a certified course to be long. That means the measurer did a good job.
Regarding the .1% 'added' to each measurement...well, the # IS built into the calculation but, reality is, since a measurer is human, we can't say for sure every mile we're adding 5.28'. We might wobble, take turns a hair wide, swerve to miss potholes/cars/dogs/etc. The .1% is added to ensure each course is long enough - not because our tool is deficient - but because the user is deficient.
If some method of measuring were better out there, USATF would be using it. GPS is not a better way. It's a useful tool, as noted in the article, but measures short (if you stopped where it said the advertised distance of a certified course should be, you'd be short of the finish).
Don't know if this is relevant...
http://www.letsrun.com/forum/flat_read.php?thread=2181687&page=1
I say that most of what he says, and what many other GPS detractors say, is wrong because if even a small part of what they say is right, there would be no way that me and several other people who have carefully tested the devices could get anywhere close to the accuracy we are seeing.
In the past I have gone through their arguments point by point and explained why they are wrong or irrelevant, and then a week later someone posts the same or similar article, so at some point I say why bother. But you sound like someone who is genuinely interested, so here goes.
His first argument is that the manufacturers themselves give the accuracy of their devices to be only within 8 meters, or whatever. The accuracy of the GPS device in determining the absolute position of a single point is irrelevant in determining distance traveled. What is relevant is what portion of that error is systematic and what portion is random. It is not uncommon when you download a track to see all the points shifted a couple meters to the east. This is systematic error and it has no effect on the distance calculation. What does matter is the random error, one point is to the left of you and the next is to the right of you, and all evidence suggests that it is pretty small, certainly no where near 8 meters.
His collie analogy is ridiculous. Anyone who has ever downloaded a track from their GPS knows that you never see one point several meters to to your left and the next point several meters to your right, which is what he suggests with his collie analogy. You don't see points way inside the infield and "up in the bleachers." And again, if this actually happened and that is the only info they used to calculate distance, you'd see errors in the 5-10% range, which no one does.
All of this goes out the window when there are tall buildings around. In that case you will see crazy tracks and bad distance info. But he is suggesting it happens all the time.
His statements that the GPS units calculate distance by simply "connecting the dots" and that the GPS units take measurements every 1 second or 5 seconds, are things that he doesn't know. The units may STORE points only every 1 second, but for all we know they are taking measurements every 10 milliseconds and using those to help calculate distance. And they may be applying real-time filtering to the points as part of the process of calculating the distance, not just connecting the dots. We just don't know. To state these two things as limitations of how accurate the units can be is wrong.
If someone is going to write an article about the accuracy of GPS units, they should start with the empirical evidence and fit their theories to that. Most, if not all, of these articles are written by people who have obviously never actually tested the devices.
The part of his article about the measurement process is very good. The part on why GPS can't be accurate is not.
BTW, while most of my tests were done with my bike-mounted GPS, several of the tests were with a Forerunner, some while riding and some while running. There Forerunner did appear to be slightly less accurate, but on the order of 0.1-0.2% less accurate.
vdawg wrote:
http://www.gmap-pedometer.comK Theory wrote:And anyone know how accurate Google Earth is (meaning a link that states its accuracy)? Mainly for a training perspective. I'm always curious if my neighborhood loop is actually the length I think it is.
I just tried mapping the Lake Wingra 10km course, which has been measured a large number of times and is certified. It so happens that the full loop is 10,040 meters (the Start line is 40 meters past the Finish line). I got 10,160 even though I took a route that should be about 5 meters short (since it follows the bike path coming out of the Arb rather than the road, which is what the race route takes. Presumably 'tangents' might cause part of the problem, but this fits well within the 1% error mentioned for GPS units (and note, the race route is presumably 10,000m + 10m margin so the comparison is 10,050m with 10,160m.
Am I missing something or is the whole "runner threatens law suit" part bs?
Most of my negative experience has been with Garmin devices, and while the specifics of WHY it's wrong may be wrong in the article I can surely say the Garmins as a rule have high variances. Do I have piles of scientific data, no. Do I have the exact reason for the variance, no. Do I know they are inherently inaccurate in general use, yes.
Personal experience, I have a roughly 33.5 mile bike route I ride consistently with a Garmin 405. I don't ride the tangents like I would when measuring, and it's expected some variance, but riding the same route over and over give me a variance of about +/- 1/2 mile. I'll often use the same device when measuring since it's my bike odometer too, and much of the time it's very good but it's regularly off by significant margins.
We have a stretch of road through a neighborhood here in Winston that's been measured by Jones counter for god knows how many races more times than I can count. Not all those measurers could be wrong, but Garmin wearers consistently report discrepancies of up to .1 miles over a mile.
Finally, more than once I've had multiple people come up and point at their Garmin and say this course is long or short. Same race. Same day. Same device. Off by tenths of a mile either direction, sometimes more on longer events.
So, dukerdog, you may have tested the devices nder controlled conditions and found them accurate. Fine. I can say I have more than enough empirical evidence to prove to myself that wristwatch style GPS devices, used in the manner they are typically used by the general public, are highly inaccurate.
And yes, if you look at the way-points you will see readings that are "in the bleachers". I've had runs where the distance between the way-points and time made my top speed show at nearly 100 MPH. Not very often, and the average for the full run would be "normal", but it's clearly you periodically get wacky measurements.
Just out of curiosity, how can I consistently get 5000m + or - a meter when I run repeat 5ks on the local high school track while marathon training? I have a Garmim 305.
I typically run each repeat in the opposite direction - and wear my watch/GPS on my left wrist. Guess which way I get a short reading and which way I get a long reading?
I'm overly impressed with this unit but can see where buildings may cause issues. Older models I had were also poor in the woods while trail running when the leaves were heavy.
In an benign environment the "collie analogy" rarely happens since the biases between successive GPS measurements are highly correlated over many minutes. You can see this by looking at the track for a GPS left stationary for a long time. It will wander around, but generally very slowly, perhaps with abrupt jumps corresponding to satellite changes. In a high multipath environment (buildings, mountains, chainlink fence lines, heavy foliage, . . .) or a high interference environment the collie will wander around much more and the bias can vary rapidly as you move through such an environment.
I don't think that contradicts your experiments such as they are, but the results will differ depending on where the experiment is performed and in some places GPS performance can typically be very poor.
I think we can all agree on the general point of the author's article, that is, racer's GPS measurements aren't reliable, whether or not he has correctly quantified the reason for those errors.
GPS Wearer wrote:
Just out of curiosity, how can I consistently get 5000m + or - a meter when I run repeat 5ks on the local high school track while marathon training? I have a Garmim 305.
I typically run each repeat in the opposite direction - and wear my watch/GPS on my left wrist. Guess which way I get a short reading and which way I get a long reading?
Easy, you're out in an open space with good reception, not many others around, controlled environment. I'll bet if you look at the tracks you'll see that you wander from lane to lane even though you really didn't. I've had courses I've measured wearing one when it was right on also.
I'm overly impressed with this unit but can see where buildings may cause issues. Older models I had were also poor in the woods while trail running when the leaves were heavy.
I'm impressed with the units also, they do an awful lot for something of the size that you can wear on your wrist.
However, that doesn't mean I'll ever take anyone seriously that questions a course measurement based on their Garmin reading. I've seen far too many cases of wide variations over known distances that didn't have "tall buildings". Maybe my unit is defective, maybe the people that report this wild variances have defective units too. Maybe they all took bathroom stops in every house on the course.
Doesn't matter. Since there is no real way to prove any individual unit is operating within specifications, and no real way to prove the operator of the unit used it in a reliable manner, they are useless as a point of contention. For every person that says they got an accurate measurement I can find one that didn't. That in an of itself is reason enough to disqualify them as a arbiter of an accurate distance.
I've read all of the links, but I still for the life of me do not understand why every race I've ever run is long. I thought it was just me, but I've discussed this with all of my friends, and they all say that all of their races are long, too.
Just amazing.