Skip to main content

Ken Trombatore, a friend and RRCA-certified coach, produced an outstanding article on how to use a GPS device in running. He demonstrates with clear diagrams with GPS data how GPS measurements are rarely if ever accurate. This brief article goes a long way, in my mind, to putting to rest the "Your course was long" lament of GPS-wearers.

How To Use A GPS Watch For Running and Racing
Original Post

Replies sorted oldest to newest

Most of the reason runners get GPS measurements in races that are longer than the true distance of the course is because they do not run the SPR, and not because their GPS units are inaccurate.
https://measure.infopop.cc/eve/...9510622/m/9611072113

I was glad to see most of his article is about the inaccuracies caused by GPS units not having a clear view of the sky. There is no doubt that GPS units will not report accurate distances when used in areas where there are tall buildings.

In the other parts of his article he makes two assumptions that many people make that are almost certainly not true.

1) The track points that the GPS saves to later show you your path on the computer are the only ones that it uses to calculate distance.

2) The GPS calculates distance by simply "connecting the dots." It's possible Garmin's algorithm is this simple, but I think it's more likely they do a bit of filtering of the data. But one thing is for sure, none of us have any idea what their algorithm is.

Because there are many unknowns about Garmin's method of calculating distance, the only way to determine how accurate their GPS units are for measuring distance is to test them.

https://measure.infopop.cc/eve/...9510622/m/8301003542
Good points, Mark.

One of these days, I am going to Google Earth and record a local 5K with two tracks - one going on the tangents, the other going down the middle of the road the entire way. I'll choose a curvy course to highlight the difference. I know intuitively the difference will be non-trivial. It will be interesting to quantify this difference on a sample course.

Maybe I will post the course here and hold a little competition. Participants will guess the difference. Closest guess gets a box of PK nails.

Oh, and it is clear to me that few if any runners around my part of the country are anywhere near as well-informed as you about GPS use or as competent in using for measurement, Mark. They read the manufacturer's claims that "The Veeblefester 600XC records your precise running distance" and then they compare the distance this thing calculates with our measured courses. Ken's article at least does a fair job of demonstrating the futility of this.
Last edited by pastmember
I finally got around to doing this comparison of a 5K measurement. First, the 5K certification map showing the SPR. Next, a careful Google Earth measurement going down the middle of the neighborhood roads, just as most runners in race do:

The certified course: Ben's Run 5K Certification Course Map

The Google Earth Measurement: Ben's Run 5K Down The Middle

Any guesses as to the difference (realizing this is not scientific - but I did a good job with the Google Earth track) in measurement between these two paths?

The answer is - I think I would like to see how many guesses we get in the next few days. So, a little contest. Closest to the "actual" difference wins. Winner gets a free race name change for any certification map they send me. No fair using any on line tools. Just guesses. In feet, meters, miles, kilometers, millimeters, or inches. No percentages of the overall distance are acceptable. Convert your percentage to distance units.

Post your answer here and I will post the winner on Sunday, February 17th, 2013. If your answer isn't wildly off, I will post that too. If your answer IS wildly off, I will post it with no name. If you wish to withhold your name, well, OK, I guess. I would rather you include it with your entry.

OK, here goes. Start sending them in.
Last edited by pastmember
quote:
Originally posted by Mark Neal:
Ben's Run 5K Down The Middle doesn't look very middle of the road.


Yeah, I noticed that when I posted it, Neal. It is an artifact of saving the image. I could give you a link to file on Google Earth, but then you could easily see the distance. In the KMZ, I magnified the map just to the point where the GE path cursor was the same width as most of the roads. It was then easy to line up the cursor so that I was measuring right down the middle. When we get in some more guesses, I will send a link to this map on GE so you can see for yourself.

SO..., what is your guess?
I had a tough time and just now gave up. First I was unable to find the light pole for the TA. Then GE kept distorting the view. When I used to use GE frequently the view was directly overhead, but now it seems to shift to various oblique views which I'm unable to cope with. My measurement ran out of gas at about 4800 meters.

I'm disappointed at my performance.

By the way - very nice course map!

I tried again and sent a private message to Lyman.
Last edited by peteriegel
Lyman, I would like to know if you got my guess not sure if I sent it right. Also I am looking at your map and I am wondering how you made it? Did you use illustrator or some other software? Very nice map that I would like to make for my race directors. Can you give me a few pointers I would really appreciate any help you could give me.
Also I believe the certification number isn't a valid number that you have on the map. Is this a real run and if so how would I look it up on the web site without a valid number?
new puzzle based on Ben's Run Map

Here’s a puzzle that was generated when I tried to locate the turnaround point on Ben’s Run. I found some inconsistencies in the description of the TA when I looked at the course in Google Earth.

Find the inconsistencies and post them here. First to find them may win, unless a better answer arises.

Entries close Thursday, February 21, at 4AM.
quote:
Originally posted by Pete Riegel:
Mark,

I stand corrected, and am at a stand trying to make an intelligent guess with nothing to guide me regarding the nature of anything on the course.

So, the estimate I provided was in contravention of the rules, and I hereby withdraw it.


No worries, Pete. The idea behind this is to illustrate what Mark Neal points out about GPS use in races: that GPS users get a longer reading because of how they run the course much more than how GPS devices perform in areas with a good sky view. I just want to use our guesses to show, as a group who are educated in this, whether we tend to intuitively grasp the difference in this one instance. Nothing rigorous or scientific about this. I am hoping for a good anecdotal object lesson from this exercise as we accumulate more guesses.
quote:
Originally posted by J. A. Wilhelm:
Lyman, I would like to know if you got my guess not sure if I sent it right. Also I am looking at your map and I am wondering how you made it? Did you use illustrator or some other software? Very nice map that I would like to make for my race directors. Can you give me a few pointers I would really appreciate any help you could give me.
Also I believe the certification number isn't a valid number that you have on the map. Is this a real run and if so how would I look it up on the web site without a valid number?


J.A., I got your guess. Thank you. I appreciate the compliment on the map. Since learning how to submit color maps in <500Kb PNGs via Corel, I have been working on developing better color maps because I find that users say they are easier to read and to use. In doing this, I have incorporated good ideas from Duane Russel, Jim Gerwick, and others. I use Corel Draw, which is essentially the Canadian version of Adobe Illustrator. I got started with Corel because Phil Quinn, who has certified many courses, gave me a copy of an ancient version of Corel to use to get started making electronic maps. I now use the latest version of Corel Draw.

As of this writing, it seems that Illustrator does not offer the same export options that Corel does. That is, I am told that Illustrator does not currently support exporting the native file as a color .PNG of acceptable-to-USATF size. Corel gives you thousands of options to do this. Fiddle with the "Export To Web" command in Corel Draw for a while, and voila, a color map that meets USATF standards pops out. Now, having said this, the map version you saw was >500Kb - I forget how much greater. There is of course some loss in resolution and detail when you compress an image from 1.1 MB, as with Ben's Run, to ~ 450 KB. But the compressed result still looks and prints much better than hard copies scanned by USATF.

Justin Kuo's clever GIMP solution allows versions of color maps in many file formats to be converted to .PNGs of acceptable size and resolution. I looked at it, but I have not tried it because I already know how to do this in Corel. As a matter of fact, Corel can take many different formats in and output them as .PNGs, as well.

I apologize for posting an incomplete version of Ben's Run initially. I did not look at this version, which I should have purged from my files after updating it. I went back and posted the final version, with the certification number and the turnaround description. Check it out.

Corel Draw is massive. It has a ridiculous number of features that are not needed to create maps. so, it is useful to know what to ignore and what to focus on. If anyone is interested enough in Corel Draw, which I think is a fine program, to want to try it, I may have the ability to provide the key for one or two seats. Let me know and we can give it a try. I will also spend up to an hour with anyone who has the software and who wants a brief tutorial in how to get started making maps.
quote:
Originally posted by Pete Riegel:
new puzzle based on Ben's Run Map

Here’s a puzzle that was generated when I tried to locate the turnaround point on Ben’s Run. I found some inconsistencies in the description of the TA when I looked at the course in Google Earth.

Find the inconsistencies and post them here. First to find them may win, unless a better answer arises.

Entries close Thursday, February 21, at 4AM.


Uh-oh. Have I now exposed myself? Do I need to re-write this turnaround description?
quote:
Originally posted by Mark Neal:
14401 Cutstone looks to be south of both 14512 and 14514 Cutstone? But maybe not since they are on opposite sides of the road. Locating addresses on google earth is not always accurate either.


It is on the opposite side of the street, Mark. But I can tell you this could make only a relatively small impact on the overall difference between the two measurments
quote:
Originally posted by Duane Russell:
Lyman, doesn't "Export for Web" save the png as a 72 dpi image? We are to be at 300 dpi. I don't have Corel on this computer, so I can't check to see what options are available when exporting for the Web, but most exports for Web convert the image to 72 dpi.

Just checking.


Duane, in Corel, "Export to Web" gives you countless options. Seriously. It allows you to set the X and Y pixel count to whatever you want, the resolution to 300 or anything else, 8.5 X 11 page size (you have to add an invisible 8 1/2 X 11 frame so that the margins do not get clipped), what kind of dithering you want and to what degree between 1 and 100, whether you want anti-aliasing, and an interlacing option. Most remarkably to me, it allows you to specify whatever number of discreet colors you want in your .PNG from one to 256. It allows you to start with one of several common color palettes, and then to specify which colors go where in your image, so you can reduce the size if you need to. It allows you to delete redundant colors for the same reason.

This was a non-intuitive bear for me to noodle this out in Corel. Once I did, by trial and error,I was amazed at the flexibility and the convenience. I have produced a few color .PNG maps in the 250 Kb range with this feature that look fine on screen, and look surprisingly clear when color printed.

I set the controls in Export to 300 DPI, 2550 x 3300 pixels, paletted 8-bit < 500Kb PNG, 8.5 X 11, Optimized palette, Ordered dithering @ 100%, anti-aliasing,and no interlacing. I then compare the preview of the map in the tool which shows this image next to the original. You can then easily tweak any of the colors or the settings while watching the effect on the image size. Sometimes, I try to cram as much quality in as possible and keep it just below 500KB.

Corel then allows you to save any final selection of the myriad options, and it then uses this custom set as a default for your next map. Then, when you are finished with your next map, you hit the "Export" command, and, voila, a finished .PNG to spec.
Lyman,
I was just guessing that if the turn-around is between 14512 and 14514 Cutstone, it might not be only 10 feet from the 14401 driveway. But I'm just guessing at what inconsistency Pete might be referring to. There is no way to tell for sure from a google earth map, especially when you're looking at addresses.
But anyway, you made the smart move of giving two descriptions of the turn-around, so even if you have something not so perfect in one description, it's made clear by the other one.
Right, it is tough with the available imagery. Thanks, I have done a lot of course layouts in my time. I have a lot of feeling for someone out there in the frigid pre-dawn darkness trying to locate timing points on the road.

A couple of years ago, one of he biggest races in the D.C. area laid out a mile marker facing the wrong way, on the wrong side of the road, not far from a turnaround. The marker was then way off. When I looked at the certification map, I noticed it did not specify which side of the road, or which direction runners would be going at this point. The layout folks still should have known, of course, yet... I think a clearer map might have prevented that error.
Pete’s Puzzle Answer

When Lyman proposed his puzzle I hopped on it like a duck on a Junebug. Using Google Earth I located the S/F, and found the TA right at the lightpole on Cutstone, right where Lyman’s diagram showed it to be. I sent my answer to Lyman. Then Mark Neal pointed out that I had violated the rules of the puzzle when I used Google Earth to measure the course. I withdrew my answer from contention.

I noted that Mike Sandford had used a novel approach. Since I was out of contention I wrote to him and found that his method was not a new one, but probably new to Lyman. I used Mike’s method and obtained an answer very close to Mike’s. My disqualified answer using Google Earth measurement was quite a bit larger.

Then Lyman added the TA description to the mix, slightly changing the location of the TA. Using Google Earth I tried to reconcile the description with the reality as shown by Google Earth. I found:

1) The lightpole is not between 14514 and 14512 Cutstone. It is south of both. A lightpole shadow can be seen on Google Earth, nicely tying down its location.

2) I had originally mistaken a wrong driveway for that of 14001. Later examination was not successful in location this driveway because of trees and shadows.

I thought that perhaps the Google Earth image was an oldie, but the imagery date was shown as 10/12/2012, fairly recent.

I concluded that my “puzzle” was not much of a puzzle at all, but Mark Neal was the first and only solver.
Well, the last entry came in just last night, from Bob Thurston.So, I extended our deadline. We have a total of six entries. All answers were rounded up to the nearest meter. Remember, we are comparing a 5,000 - meter certified course with a path down the middle of these same roads and we are guessing how much longer the runner going down the middle of these 20-foot wide roads travels. The second question asked how many seconds would a 6-minute per mile runner using the SPR finish ahead of a runner starting at the same time, running the same pace, down the middle of the road.

We were just looking for guesswork here. Closest to my careful but unscientific measurement wins. Some of the responses came from a careful methodology.

OK, so now the results:

  • Duane Russell: 5049 meters
  • Mike Sanford: 5100 meters, 23 seconds
  • Mark Neal: 5110 meters, 25 seconds
  • J.A. Wilhelm: 5130 meters, 29 seconds
  • Bob Thurston: 5145 meters, 32 seconds
  • "Cheatin'Pete": 5187 meters

And, the winner is ... (drum roll)..."Duck On A June bug" Pete Riegel!!

Wait. "Cheatin' Pete" used G.E. to measure down the middle. Not exactly guesswork, Pete. Sorry, you are D.Q.'d. Too bad.

My results from Google Earth: 5,182 meters. Our new winner is: Bob Thurston. The SPR runner would run 18:37. The middle of the road runner would traverse 3.64% more distance, thus reaching the finish in ~ 19:19 on this course, an astonishing (to me) 41 or 42 seconds behind his savvier friend.

Well, I get two things out of this little exercise: 1) It tends to support Mark Neal's point about GPS users and the distances their devices record in races: GPS devices are reasonably accurate where they have a good sky view; and how their users run the course likely has much more to do with their "You course was long" claims than GPS performance; and 2)On a course with a lot of curves and turns, a savvy SPR-runner will beat his equally speedy but less SPR-savvy competitor by more than any of us might have guessed.

All good guesses, gentlemen. Thanks. This was fun. Bob, you are the winner of a free map-changing job (don't let Gene know) haha.
Lyman,

When you measured with Google Earth I understand that you measured in the centre of the roads.

I assume that would be in the centre of the satellite image of the road not the centre of the road overlay.

What route did you take at junctions? Which of the following?


I assumed that even the most stupid runner would not stay exactly on the centre line and do a sharp right angle turn where the centre lines met, but would round off the junction corner by remaining 10 foot from the kerb. If that assumption is not correct and you measured with the route with sharp right angles, then on the corner shown above it adds about 17 feet. There are 11 such junctions on the course so that would mean that I would have to add 57 metres to my result.

I am not sure if my quick eyeball estimate of the amount of turning on the course was accurate - I took a value of 6 full 360degrees of turning and calculated as follows

quote:
6X360degrees of turning
Roads seem to be 20 feet wide from google so runner in centre on bends is 9 feet from the SPR

So 6 x 2 x pi x 9 feet = 339 feet

I make 339 feet is 123 seconds that the runners on the SPR runs faster than the runner in the road centre.

Note just a rough calc - for simplicity I assume the centre road run stays 10 feet from the inner kerb at road junction.

Next morning I corrected my answer as follows:
quote:
I was just going to bed last night when I did the quick calculation. Afterwards I worried that I had not treated the 180 degree turn round correctly since the road here would be narrowed. This morning I have checked the satellite view of the turn round road. It is 20 feet wide like the other roads. So on the approach to the turn round in the right half of the road the available width is 10 feet.
So if running in the centre of half the road you are 5 feet from the kerb where as the SPR will go within 1 foot on bends. The turn round is 180 degrees. So to modify my result for this turn round for 5 feet from kerb instead of the 10 feet which I used, the excess distance has to be reduced by pi*4 feet= 12.6 feet. 339-12.6 = 326 feet which is my corrected answer. Time difference also reduces and is now 22 seconds.
(By the way I have just noted a typo in the post above it should say 23 seconds not 123 seconds. I never checked my tying last night!)
Last edited by mikesandford
I was thinking along some of the same lines as Mike on this-- does a "middle of the road" runner really go right up to the dead center of the intersection and then turn? I was also wondering about those messy little turns where the course was restricted, etc. In the end I just thought these things could be ignored, and since almost all real street corners are arcs of a circle, you can just assume the MOR runner stays half the road width away from the inside on turns.
I must have made a really lucky guess on that road width. But I was also surprised by the total amount of turning on this course-- I think I figured something like 5.1 x 360 degrees but that was not including the little turnaround in one corner of the map so maybe Mike's estimate of 6 x 360 is closer.
Lyman, thanks for all this entertainment from your challenge. I'll call you to work out the prize deal!
I knew I would be blown away by the scientific techniques and analysis of this little puzzle by some of our brethren.

I did not attempt to re-create the likely path that a typically inattentive runner (including some highly competitive athletes in my experience)would take on this course, so Mike, Bob, you are justified in going with a smaller number than my measurement. My strategy was non-meticulous, but my measurement technique was careful. I can provide the G.E. link if anyone is really interested (if you are, you need to get out more often Razzer). In G.E., I set the magnification of the map so that the rectangular drawing cursor was equal to the width of the 20-foot wide roads. Using this as a guide, I inched down the roads, keeping the cursor lined up between the sides of the roads.

This creates a path almost dead down the middle of the road. If I am looking for something to do later this year, maybe I will come back to this exercise to inject a little more realism: I will enlarge the map even more, and try to measure using a path that I think is one that might more likely in a real race. That is, cutting a little bit of a tangent here and there at turns, but still running mostly in the middle of the road, as we commonly see in races.

In big races, lots of people run longer than down the middle for parts of the race as well as shorter. So, I do not think the additional distance I measured is too far out of whack.

I intend to use this exercise as one more arrow in my quiver of responses to the "Your course was long" refrain heard at almost every race these days in these parts.

This was fun for me. Thanks to all who participated.

Add Reply

Post
×
×
×
×
Link copied to your clipboard.
×