Check out the info at: http://www.usatf.org/routes/view.asp?rID=268226

Says the Gate River Run 15 km course in FL is 15.19 km long. Can somebody explain what appears in that space for me? Thanks.

The distance stated in these course maps is dependent on the skill of the person who plots the course. Evidently "Doug" plotted the certified route, and his methodology resulted in the calculated value - which isn't 15 km.

This bears no relation to certification, and USATF, beyond providing the mapping site, had nothing to do with it.

This post was being composed before Pete's concise explanation above appeared, so forgive my longwinded reply.

This appears to be a database of Google mapped routes which anyone can map using Google and upload to the USATF.

The distance is determined by the accuracy with which the mapper places the route defining points on the Google maps.

The FAQs warn that the distances may be wrong!

Examination of the web page code suggest that it may have been written by one Keith Lively.

Google mapping does not replace the calibrated bicycle wheel, but it can be done rather more accurately than was done by the poster named "Doug" for this race.

I several years ago tried out mapping a course to be measured and found it worked quite well, although was not as fast as a large scale map and a map measuring wheel, or even a piece of cotton. Trying it agin I think the googlemaps software has been improved. So I tried it on one of my accurately measured routes.

Using the satellite view at maximum resolution, I could position the points within a metre or so of the SPR, except where the route goes under trees and the kerb locations are hard to guess. It took less than ten minutes to plot the course. I then uploaded it to the USATF website. You can see it at Abingdon Loop.

Google measured my points as 4.54 km. I did not quite join up the loop since I could not plot the finish on top of the start, so my google route ends about 3 metres short of the complete loop.

Measurement of the loop with a steel tape along the SPR yielded 4533m, and many calibrated bikes give 4533+/_ 3m without the SCPF. So I am now quite impressed with the result from the Google maps.

I have also recently started measuring this loop with a GPS unit carried on the handle bars of my bike. I will report these results in full when I have carried out a number of experiments to determine the best way to measure with my GPS. But I can give a preview now of some results:

ETREX H track recorded 1/sec (no smoothing applied to track points): 4659m
ETREX H track recorded 1/sec (no smoothing applied to track points): 4637m
ETREX H track recorded auto, high resolution: 4533m
ETREX H track recorded auto, high resolution: 4530m

In auto the internal Garmin software selects points to record attempting fit the course accurately. The track distance is the distance between these, already smoothed, points calculated by pythagorous.

Mike,

It's interesting that you see such a big difference when changing the settings on your Etrex. In results I posted quite a while ago I found that changing the settings on a Garmin 60csx or a couple different Garmin Forerunners had very little effect on the measurements they recorded. All my GPS measurements, regardless of the settings I used, were within 0.4% of the Jones measured distances. The average difference was about 0.2%.

http://measure.infopop.cc/eve/...9510622/m/8301003542

Mark,

Yes, I am trying to invesigate the internal Garmin software, by experiment and "reverse engineering" or at least guessing how it works. It is work in progress. I will post more when I have more experimental data.

Please note the distances I have posted here are not from the Garmin odometer function but from the data points read out and then calculated by pythagorous. Since the data points at 1/sec are noisy that noise adds in to the distance when calculated by this method.

I am also collecting data with the Garmin odometer function, and I am trying to smooth the downloaded points using GPSU software to reduce the noise.

Overall I am very impressed with the Garmin ETREX H on the bike, and it fully supports the data you have published here.

Ah, maybe the better question is why is the 'calculated distance' even used? Too many runners will not get further than what they read at the top & figure that's THE race distance. And, since it appears on the official governing body's website...what appears must be right! The GRR must be 15+ km.

My suggestion: we (RRTC) should request the distance in that area reflect the actual and not some 'calculated distance'- avoid confusion.

Ths site is used for tons of courses that are not certified, and for estimating lengths of courses yet to be measured. How would USATF know how to edit the certified courses from those that are only someone's estimate of a new course, or for someone's personal training run?

It's a very useful site as it stands. I'd say leave it alone. I've not heard of anybody getting confused yet.

Note: As the site now stands, there is no mechanism keeping anyone at all from calling their plotted course the "USATF XX km Championship," and plotting a course that comes out to XX plus or minus a huge value.

It's not the only mapping program out there, and policing is a bigger job than I would want to undertake.

This message has been edited. Last edited by: Pete Riegel,

It would be nice though, to have a special location within Running Routes that included only maps of certified courses that are "USATF-approved." USATF-approved would mean only that the map was submitted by the person who measured the course, and the final distance shown is within say, 0.5% of the advertised distance.

There would be no requirement for measurers to do this, but if they wanted to they would have a place where only they could put an electronic map of their course.

I have heard this complaint before. I have approached USATF, but have not found a solution. Hence, I say as Pete leave it alone.

Ah, mapping with satellite photos. Inexact.

One could map (I use MapMyRun.com, instead of the USATF site, as you can edit the track) their precise Jones/Riegel-measured course, and not have the displayed distance correct. The reason is that the satellite photos are "rubber-sheeted" to fit to one-another.

As the photos are stitched together to make a continuous image, the images are, by necessity, skewed a bit. Many of the courses I measure are really darn close to the distance mapped. But, I have a few courses that are off by 150 meters on 5k course that start and finish in the same area. These are likely in the area that images were stitched together.

My point is, even on courses that we certify, the accompanying satellite image may not plot out close enough to keep some people happy. And, the mile points are usually off by a noticeable amount. I always explain the rubber-sheeting to people, and that seems to explain the variance sufficiently.

Duane,

Are those courses you talk about in areas with poor map resolution? I have seen areas with bad resolution where a road doesn't even line up between two images that make up the map.

But just about every course I've ever Jones-measured (all in high resolution map areas), I've also measured with Google Earth, and I've never seen errors anywhere near as large as you mention. As long as I have been able to clearly identify the start and finish on the maps, the measurements have always been within about 1%, or about 50m per 5k, and usually a good bit closer than that.

No, these are in the Denver area, with very good resolution. I looked at one pretty closely, and could see where the images were stitched together. I haven't studied the other one that closely, but they are the only two that were way off on distance.

On longer courses, I see variance more often, but it normally isn't more than 100 feet, or so. Point-to-point are more-pronounced than a loop.

But, a great example of how the math and programming behind the satellite mapping programs is not precise is: http://www.mapmyrun.com/run/un...-denver/525880721378

Look at the west end of that course. Mile 13 is just before the turnaround. Looking at Miles 12 & 14, they are almost on top of one another. But, look at Miles 11 & 15, and they are very far apart. Straight line each way, same road. Look a little east at Miles 10 & 16, and 9 & 17, and they are closer together. The variance is much more than one would expect on a straight-line, out-and-back course.

That is why I don't rely too heavily on the accuracy of any mapping program. They get close, but I always have a contingency for adjustment, both shorter and longer.

I think the issue has more to do with where MapMyRun decides to put mile marks rather than inaccuracies in the imagery. Even if the imagery is bad, it makes no sense that measuring from east to west(10 to 11) would give a different answer than measuring west to east(15 to 16).

I zoomed in on the mile marks on that map to locate where MapMyRun says they are relative to local streets and buildings. I then went to Google Earth and measured between those points. Google Earth says "Mile 10" to "Mile 11" is 1794 meters, and "Mile 15" to "Mile 16" is 1417 meters.

I think you should send that map to the folks at MapMyRun and ask them what's up.

Duane's probably getting better resolution because he's so much closer to the satellites.

I actually got a call from MapMyRun yesterday, for another reason. But, I let her know about the mile points. She pulled-up the map and was surprised. She said she will scold the programmers, and let them know they have a problem.

Stu, are you saying I am high???

No, no, no Duane, I'm sure Stu was just inferring you're a little closer to the stars!