As Dukerdog correctly mentioned inaccuracies with 2-6 is mainly due to user error. In my very early days I used the measuring wheel- always walking at a steady 20 minutes per mile, never switching hands except to get closer to the curb (tangents). Most users wouldn't take this much care and therefore I believe would yield a less accurate measurement then google earth.
row row rowbury your boat wrote:
For the sake of the GPS watch then, (I know we obviously shouldn't use it for accuracy) are we on correct assumption that it will always measure short? Is there an anomaly that could cause it to be long?
Depends on what you mean.
Will the GPS device itself always measure a course short or long? No. The device itself will sometimes measure a bit short and sometimes measure a bit long.
Will the average person measuring a course with a GPS come up with a distance that is longer than the actual course distance? Usually.
Will the average person measuring a course with a GPS while running a race come up with a distance that is longer than the actual course distance? Almost always.
I'll assist David here.
The GPS device is a "fairy" accurate device, considering the purposes and limitations that are defined by how it collects and computes position.
There are 24 satellites circling the globe at various altitudes and speeds and traveling at different directions. Your device needs to find three satellites to determine your location to within 15 meters 95% of the time (two standard deviations). 95% of the time it is accurate within 15m, 5% of the time the error is greater than 15m. With a fourth satellite your elevation can be determined.
I haven't found any information as to the frequency of calculations your GPS runs per minute, but if you take a look at some of the maps on the link I am supplying, you'll get the whole idea how this is all done.
If you go on a 100m run, your gps is not going to be very accurate. At the start of your run your gps will pinpoint you anywhere within a 15m radius 95% of the time (and greater than 15m 5% of the time) It could position you at your starting point (not very likely) or in front of it, behind it, to the left, right or in your neighbors garage. At each positioning it does the same. At the end of your run it does some sort of post-hoc interpolation, and hopes that it's close, As you can see, the longer your course is the more accurate the GPS is likely to be. But that's not because of pinpoint accuracy of the individually recorded positions, it's because of the smoothing algorithm that guesses where you are most likely to be at any given positioning measurement. The beginning and ending positions will amount to a smaller percentage of the total distance traveled, and the deviations from the true path SHOULD average out to a virtual representation of your actual path,
Take a look at the pathways that were measure by the same device on the same course recorded on different trials (specifically images 6-8):
Malmo
Thank you.
I was just trying to provide the "Cliff Notes" version for this crowd! :)
To the OP,
Did you get a PR for 5 miles? If yes, congrats, you got a new PR.
a PR enroute? wrote:
To the OP,
Did you get a PR for 5 miles? If yes, congrats, you got a new PR.
Indeed! I did get a new 5 mile PR... of course, that would be according to my GPS watch. Thanks for the silver lining.
For the crazy guy, a 2 minute PR is completely reasonable for a person stepping up their mileage from 20-25 to 40-45 per week. I was expecting something around a 37:30 from my recent 5K. Do you really think a person taking their running semi-serious would be off on their fitness by 4 minutes? Insane.
Also, I can't drive the course, as part's of it are on non-vehicle access. I don't have any other fancy schmancy measuring device. So, for now, the GPS will have to do. I still say the GPS is far more accurate than people give it credit for if you don't have tree/building cover.
I did email the race organizer to find out the actual distance. I haven't gotten a response yet.
For those who think that Google maps is accurate, you need to re-think. A few years ago the pancake flat Berlin Marathon course showed phantom hills throughout. As it turned out y hey were buildings.
Malmo -
I don't doubt that there may have been situations where there were some unusual readings from Google maps but I have measured hundreds of road courses with the Jones/bicycle method - all compared to GoogleEarth, with never a deviation of greater then 10-15 meters per 10km. You are invited to come and measure with me anytime to see the comparison. I will even supply the bike and Jones Counter! :)
malmo wrote:
For those who think that Google maps is accurate, you need to re-think. A few years ago the pancake flat Berlin Marathon course showed phantom hills throughout. As it turned out y hey were buildings.
Remember, David is talking about using the path measuring tool on Google Earth, not some other measuring device on Google Maps. What David is referring to, and correct me if I'm wrong, is measuring just a path distance and not displaying nor compiling any elevation or topographical data. When I'm curious about elevation, I simply check the elevation of specific points by hovering the cursor over it.
For everyone else doubting it, if you don't have experience measuring paths on Google Earth, please try it. I recommend it all the time and very few actually sit down and take the time to plot points accurately....and accuracy, zooming in, and plotting many points on the turns is the key. Like I said earlier, I mapped a 400m track within 60cm of actual distance. I can measure a football field (straight line = easy) within inches of its actual length. Try it.
The nice thing about it is that it's easy to master and readily available. A bicycle with a Jones Counter is nearly perfect, but takes a little experience and you need access to one. With Google Earth, I can map a 10k within 15m while sitting on my couch with my laptop.
The example of the Berlin Marathon course doesn't mean that the elevation calculations were wrong it means,that the path was wrong. Instead of a measurement point being on the marathon course, the measurement point was on top of a building next to the course.
Even GPS watches have bad days
http://oi62.tinypic.com/20jfcsx.jpg
http://oi59.tinypic.com/2we99mt.jpg
Some devices are better than others
http://oi59.tinypic.com/2we99mt.jpg
long and short of it
http://oi59.tinypic.com/2n9yhj6.jpg
http://oi59.tinypic.com/ezi36.jpg
Garmin 620 issues
I did an experiment in a race several years ago where I asked runners wearing GPS watches to let me know their measured distance and also which model of watch they were wearing. The 205 (which is also in that report malmo posted at the top of the list) was one of them, and all of the different models (all Garmins) were of similar accuracy to the 205.
But now it looks like Garmin is sacrificing accuracy for whatever (probably price but maybe additional features) in their new models. Too bad.
Race director finally got back to me. Official distance was 9K, which means the course was 0.62 miles short.
And to my huge surprise (sarcasm), my GPS watch was almost exactly 1% long, just like every other race I've run without overhead cover.
Seems odd that it's exactly 9km - but could happen.
D.Katz wrote:
...
In regards to the accuracy of the various methods of measurement mentioned, I would rate them in the following order based on my experience (most accurate to least accurate):
1. Calibrated Riegel-Jones Bicycle Method by an experience road course measurer cutting all the tangents within 30cm of turns and cones.That's the world standard - The only method accepted for Certification around the world. Also, steel taping an entire course cutting tangents. This has been done in the far east with a number of races in the past.
...
I was part of a team who measured a ~3mi grass course using 6 different methods:
1. Steel Tape - the whole way with tangents and 30cm radius on turns, measurement done by On Course Rating Systems out of Virginia - the golden standard
Calibrated bicycle - Jones counter, done by USATF certified course personnel who measure for big city marathons etc
GPS - Garmin Edge 305 on bicycle ~9mph average
GPS - Garmin Edge 305 walking ~3mph average
Wheel - "inexpensive" un-calibrated
Wheel - "expensive" un-calibrated
The measurements came out to:
1. Steel tape = 2.9758 miles. gold standard
2. Cal Bike = 2.9717 miles, 0.14 % short
3. GPS bike = 2.9727 miles, 0.10 % short
4. GPS long = 2.9809 miles, 0.17 % long
5. Wheel Cheap = 2.9568 miles, 0.64% short
6. Wheel Good = 2.9502 miles, 0.86 % short
Wheels - all wheels slip on grass and bumps to some degree. Every slip reduces the the total measurement by a fraction of an inch adding up to a significantly short measurement over the entire course. Even the Jones counter which was calibrated on the grass came up 7m short, which is why they add a little extra to each course to make sure courses don't come up short if a record needs to be confirmed. An un-calibrated walking wheel is the worst way to measure a grass course. They always come up too short for official record keeping. You need to calibrate it against a steel tape for at least 300m to get close.
GPS - The measurement on the bike came up short, while the walking measurement was long. There are two sampling errors that offset each other when doing GPS measurements, turn truncation which shortens the course, and location precision which lengthens it. When you are rounding a turn, a faster speed will result in fewer samples which mathematically cuts the virtual path inside the turn as the line between each sample is a straight line which shortens the turn. As you slow down the error on turns diminishes but the location precision error gets amplified. To the measuring algorithm you appear to be bouncing from side to side for each sample position within the radius of accuracy for the device. The slower you go, the more bouncing back and forth you appear to be doing, which adds length to the measurement. So with the Garmin Edge 305, the sweet spot would have been about 7mph, on a course with around 1200 degrees of turns... which matters.
Summary:
On grass courses metal tape is the gold standard, but a calibrated Jones bike with some margin added will be good enough. All other methods should not be used for confirming records.
On pavement steel tape is the gold standard, but impractical. A calibrated Jones bike with margin added is the standard used today. A calibrated walking wheel is pretty good but should not be used to confirm records.
GPS is great for measuring workouts and informal records, but isn't good enough for official record keeping.
We did not drive a car on the course.
4. GPS long -
Should be
4. GPS walk -
Google Earth , "makes no claims as to the accuracy of the coordinates in Google Earth. These are provided for entertainment only and should not be used for any navigational or other purpose."
Nice work Disco
Just a few comments.
1.Because the terrain (even the grass) can change, the only truly accurate way to measure a xc course is with steel tape. BUT, it doesn't need to be that accurate. (There's a long thread on this somewhere on Letsrun). That's why there are only course records and not National/WR for non paved courses. In fact USATF does not Certify XC courses. If a measurer did have one certified it slipped through the cracks.
2.The one component you left out is a road course measurement with googleEarth.
3. For Certification (and ratification of AR & WR):
Tracks- EDM or Calibrated steel tape (certified by a country bureau of standards - NIST does it here in the USA
Road Race Courses- Calibrated JR Counter/Bicyle or steel tape
Every other device/software is for entertainment.
It's odd that you give a list of reasons why GPS won't give you an accurate result when your experiment showed the GPS measurements to be as accurate as the calibrated bicycle on that grass course. In the experiments I've done with a 205 my average error was larger than what you saw here, but I definitely got some measurements that close. My errors were distributed fairly randomly between -0.5% and 0.5%.
To the poster who said he always gets 1% long, that's what I saw in my race experiment. The measurements people reported averaged about 1% longer than the calibrated bicycle measured distance. When I ran the course with my 205 following the SPR (shortest possible route) I got a measurement that was about 0.2-0.3% in error. People, including you probably, don't run the SPR in races.
All of this is with the previous generation of Garmin Forerunners, 205, 305, etc. Looks like the newer models may not be as accurate.
dukerdog wrote:
It's odd that you give a list of reasons why GPS won't give you an accurate result when your experiment showed the GPS measurements to be as accurate as the calibrated bicycle on that grass course. In the experiments I've done with a 205 my average error was larger than what you saw here, but I definitely got some measurements that close. My errors were distributed fairly randomly between -0.5% and 0.5%.
To the poster who said he always gets 1% long, that's what I saw in my race experiment. The measurements people reported averaged about 1% longer than the calibrated bicycle measured distance. When I ran the course with my 205 following the SPR (shortest possible route) I got a measurement that was about 0.2-0.3% in error. People, including you probably, don't run the SPR in races.
All of this is with the previous generation of Garmin Forerunners, 205, 305, etc. Looks like the newer models may not be as accurate.
Two separate GPS passes differed by 13 meters over a three mile course under ideal conditions. Clear day, some trees but no full tree cover, no large buildings, very few altitude changes. Others have already reported the problems with GPS measurements in obstructed environments. GPS may be dead on, but it might not, and without an independent standard to judge against you can't know.
Like I said, GPS measurements would be good enough for a PR, but not
good enough to certify an official record.
One of the things we were doing that day was comparing the USATF SPR versus the NFHS "middle of the path" techniques for course measuring, so I assure you we took SPR into account. But you are right, no one runs the official SPR, they always run longer ... which is the point of SPR.
You might be interested to know that the NFHS method for measuring XC courses uses a "middle of the path" technique, and it does not specify the official width of the path. A course measurement could change based on the interpretation of "path width". This technique makes it possible to run a shorter path than the official distance by running tight on the turns which everyone does. Bad, bad, stuff. The NFHS rule book is complete crap on this issue.
