How important is an accurately measured race?

RoyMcC

Krusty_Clown wrote: »

I'm just pointing out that watch-based GPS receivers offer accuracies of at best 95% (5% margin for error) and that's if and only if you follow a perfect race line. QUOTE]

You surprise me a bit there Krusty. My impression is that they are more accurate than that, or at least that's been my experience.

Example 1 - although I would normally Garmin a new training route only once, if I do it a second or third time the distances invariably match up to with a few metres.

Example 2 - Adamstown 8k and Dunshaughlin 10k I both found to be smack on the button as regards distance. Surely that would be a coincidence if your statement is correct.

Also, the grumbles about under or over-distances at some races usually bear some similarity as between the runners themselves.

Not necessarily disputing what you say though - is that '95% at best' a proven statistic?

Gringo78

W.B. Yeats wrote: »

Question: is there a margin of error that is acceptable at all? (4.9/5 is 2% out)

Condo feel free to correct, but I think an AAI Measured course will be measured to be 100.1% of the distance i.e a 5k course will be measured to be 5005m in length. This suggests a jones counter accuracy of 99.9%. So I think anything less is unacceptable. 2% on a 5mile race is 30sec error on the winning time..not on. 1% on a marathon and it'll be 400m short...again not on.

There are plenty of course measurers out there....for true charity races, many of these will give their time for free, for commercial races, theres no excuse not to pay one. A Jones Counter costs less than 100euro so even if they want to not go the certified route, no excuse not to be accurate.

Condo131

W.B. Yeats wrote: »

Question: is there a margin of error that is acceptable at all? (4.9/5 is 2% out)

SCPF (Short Course Protection Factor) in measurement is 0.1%. This is added, after calibration, to ensure that the course is not short. So...in 5 miles......the amount added is ...8.04 metres. 2% would amount to 150 metres = between 25 and 40 secs, depending on the athlete.

Btw, the vast majority run nowhere near the SPR (Shortest Possible Route). I'm constantly amazed that, in races, people take the long way round bends. On a twisty route, that can add a hell of a lot!

Burgman

Condo131 wrote: »

SCPF (Short Course Protection Factor) in measurement is 0.1%. This is added, after calibration, to ensure that the course is not short. So...in 5 miles......the amount added is ...8.04 metres. 2% would amount to 150 metres = between 25 and 40 secs, depending on the athlete.

Btw, the vast majority run nowhere near the SPR (Shortest Possible Route). I'm constantly amazed that, in races, people take the long way round bends. On a twisty route, that can add a hell of a lot!

This is why I believe that the "shortfall" announced by the race organisers of the 5 mile race last Saturday (149 metres) still does not resolve the issue. I am glad to see Krusty Clown (the resident expert on Garmins on this site I believe) expressing scepticism about the accuracy of GPS watches. I recorded 4.88 miles on Saturday. Add on 0.0926 mile and you still get only 4.9726 miles. I would expect that a Garmin watch should give 5.0x miles. I am certain that I did not run the most efficient line, even though I try to ensure that I do. Has anyone ever read the IAAF/AIMS document on race measurement? If you did, you would not have any confidence in the accuracy of a Garmin, usually worn by someone keeping to one sided of the road on a road with lots of bends. Certified course measurers have to go to great lengths to ensure an accurate course length.

shels4ever

Gringo78 wrote: »

There are plenty of course measurers out there....for true charity races, many of these will give their time for free, for commercial races, theres no excuse not to pay one. A Jones Counter costs less than 100euro so even if they want to not go the certified route, no excuse not to be accurate.

From what I've heard there isnt a huge amount of them to covery all the races that are now running.

Slogger Jogger

I measured the plod half marathon at 21k yesterday which is pretty damn good for a race which is offroad and over manys a hill. And it was a bargain. As the race director said in advance it was never going to be a PB-able route.

You pay a premium for most races these days. The least that you expect is that it is accurately measured and marked. Especially when some people will have trained specifically for them and may have nailed a PB - only to have it dismissed due to inaccuracy :mad:

robinph

W.B. Yeats wrote: »

Question: is there a margin of error that is acceptable at all? (4.9/5 is 2% out)

I think there is an acceptable margin of error for XC races, for road though anything less than the full claimed distance is unacceptable.

2% out for a marathon would be about 25 and a half miles, I don't think anyone would accept that. Missing any of the extra 385 yards would be too wrong.

lecheile

There are some very good points made about the other elements that are enjoyed and appreciated in taking part in races, but foremost in people's list of must-haves is a correct distance. I don't accept that race directors or organisers are prevented in terms of cost or access to the right resources, in getting a course properly measured and certified.

If you are arranging a race as a club or through a club, the local county board will have qualified course measurers available, many of whom provide the service at minimal cost. If you are arranging a commercial event, you need to plan for spending a few euro to have the course measured properly - and look for help early on! The course does not need to be re-certified unless some element of the course changes. Where a race carriers an AAI or other recognised label, this is on the basis of a course certificate being produced by a suitably qualified individual - this clearly is not happening in many cases.

MisterAdam

Burgman wrote: »

Add on 0.0926 mile and you still get only 4.9726 miles. I would expect that a Garmin watch should give 5.0x miles.

GPS was first launched in 1978. In terms of technology, it's a dinosaur - a dinosaur in polyester flares and platform shoes. It's favorite band is probably the Bee Gees.
Not only is it hopelessly out of date, but we're not even using its full potential. We only recieve the low resolution civilian signals. The US mlitary, who operate GPS, keep the really accurate measurements for themselves.
There are a few other, more technical, reasons why GPS is inaccurate. But I don't even almost understand them.
However, the biggest source of error has to be this - GPS calculates your position periodically and then assumes you ran in a straight line between each two points. If you don't run in a straight line, it'll underestimate the distance you ran. Anyone who has seen their estimated speed plummet when they round a sharp corner can testify to that.
The point is that the inaccuracy is skewed. Generally you'll have ran slightly further, instead of shorter, than your watch says. And it's entirely possible that, if you stick enough close to the corners, you can end up with a reading slightly shorter than the actual race length.

Krusty_Clown

RoyMcC wrote: »

You surprise me a bit there Krusty. My impression is that they are more accurate than that, or at least that's been my experience.

You're right RoyMcC. I went back and had a look through the 405 Garmin user manual (where I thought I'd seen the reference before) and didn't find it there. I actually found it in one of the older Garmin GPS devices I used to have (Garmin Quest).

Here's what it had to say (for that device):
GPS Accuracy Position: <15 meters (49 feet) = 95% typical accuracy
So the Forerunner is a little better than the 5% I previously quoted. Typically when I go running, the accuracy of my 405 would be around 25-45 feet, which based on Garmin's reckoning is probably around 3%-4%. Again, remember, this is the best case scenario. That's why the overall distance/pace is just a curiosity when I run a marathon. I still use a pace band and stop-watch. Garmin's are a valuable aid for training and racing, but they're just an aid.

Now, let's go back to the 1,000 mile challenge, and knock 3-4% off of all the Garmin users.

robinph

Each individual point recorded may be off between 25-45ft, but unless the very next point recorded is off by 25-45ft in the completely opposite direction it doesn't matter as far as the speed or distance recordings are concerned.

If your locked onto 5 sats and they give you a particular position, but that is 45ft to the left of where it is in reality that is fine. The next position recorded once you move 10ft to the right and are still locked onto those same 5 sats will still be incorrect by the same 45ft to the left of reality on the ground. The GPS will have recorded you moving 10ft, you did move 10ft, just the long/lat positions for the start and end points are both wrong by the same amount. You still moved the distance that the GPS said.

Now during a race you will loose connection to some sats and gain a lock on others which will effect the individual point accuracy along the way, but overall that would be insignificant and from what I'd seen of looking at track logs the one really odd point out of the thousand plus is just ignored in the distance and speed calculations. I've not looked closely at a trace from the 405 in a while though admitedly.

Unless you are doing laps around a couple of cones on a straight road I don't think that the bends are too much of an issue either as none of us are running fast enough that we'd get round those bends in the park in a second such that the line between the points comes up shorter than actuall run.

Krusty_Clown

We're going way off-topic, and this is probably a better discussion for the GPS forum, but I wouldn't necessarily agree. At each recorded point, you can be anywhere within a 'x feet radius'. So there is no reason that your trackpoints will be consistently 10 feet to the left (irregular pattern could be down to anything, like additional tree cover, buildings, etc). Take for example:

Where the blue line is the true route and the red line is the Garmin recorded route. Perhaps the difference in route length is only a couple of inches over three trackpoints (up to 5 seconds), but over the course of a long race, these all add up (.1 of a mile, over a three hour marathon, if the pattern persisted (unlikely though it is)).

Condo131

shels4ever wrote: »

From what I've heard there isnt a huge amount of them to covery all the races that are now running.

Imho, there aren't enough anyway. The same people are 'caught' for measuring all the time...sometimes measuring two or more courses in one day...plays havoc with the training :eek:!

This is the current AAI list of approved measurers.....it's a bit dated, so there may be more trained & approved now. Note that there are large areas of the country with, apparently, no approved measurer.

Measuring established courses isn't too much hassle..apart from traffic and generally starting around 6 am on a Sat or (more likely) Sunday, to avoid it.

The big problem with new courses include preconceived ideas like "it must start here" and, at the same time, "it must end here" - fine if the course is 100% accurate before anyone measures at all. The other main problem with new courses is the pre-selected course may be significantly "out" - have found a 10 mile course "measured" with 4 wheels was actually almost 0.5 mile short. All this can result in a load of 'pfaffing around' looking for alternative routes, or even remeasuring another, alternative, course, on another day, along with doing all the calibrations again.

PS: We've been meaning to hold a measurers training course in Cork for some time, but long fingering has...well..emm...long fingered it . Probably will arrange it in the next three months.

lecheile

Condo131 wrote: »

Imho, there aren't enough anyway. The same people are 'caught' for measuring all the time

Agreed! In many cases though, those guilty of not providing properly measured courses would not be aware of such limitations as they may never have sought a qualified measurer - "Shur aren't dem clocks on de cars that do kilometers fierce accurate now"

Condo131 wrote: »

This is the current AAI list of approved measurers.....it's a bit dated, so there may be more trained & approved now.

There was a course held in Tullamore in April(?) of this year which was filled with a whole new batch of enthusiasts keen to apply their newly honed skills. We'll just need to wait a few years for that to be reflected on the AAI website. Best bet is to contact local county board in the first instance.

plodder

I tested my Garmin on a track once and it measured 401.2 metres, which was reasonably accurate. Maybe, it was a fluke, but even though, each individual point was metres off on the map, the overall path followed the shape of the track and was quite accurate for measuring distance. As I understand it, a significant amount of the error is caused by atmospheric conditions, and that tends not to vary so much over short distances or times.

By the way, when GPS came out first, it was restricted by the military so only they had the most accurate measurements, but that hasn't been the case for over ten years. High end surveying equipment based on GPS can be accurate down to centimetres.

I think it'd still be better to rely on the old technology for course measuring though, as it'd be more consistent as well as accurate (for this purpose). But, a GPS watch would be a useful sanity check, and if someone even cycled that 5 mile course with a watch, they'd definitely have seen it was short.

robinph

plodder wrote: »

I think it'd still be better to rely on the old technology for course measuring though, as it'd be more consistent as well as accurate (for this purpose). But, a GPS watch would be a useful sanity check, and if someone even cycled that 5 mile course with a watch, they'd definitely have seen it was short.

Yep, someone heading backwards round the course during the setup on a bike and they would have spotted the mistake alright.

TFBubendorfer

Krusty_Clown wrote: »

You're right RoyMcC. I went back and had a look through the 405 Garmin user manual (where I thought I'd seen the reference before) and didn't find it there. I actually found it in one of the older Garmin GPS devices I used to have (Garmin Quest).

Here's what it had to say (for that device):
GPS Accuracy Position: <15 meters (49 feet) = 95% typical accuracy
So the Forerunner is a little better than the 5% I previously quoted. Typically when I go running, the accuracy of my 405 would be around 25-45 feet, which based on Garmin's reckoning is probably around 3%-4%. Again, remember, this is the best case scenario. That's why the overall distance/pace is just a curiosity when I run a marathon. I still use a pace band and stop-watch. Garmin's are a valuable aid for training and racing, but they're just an aid.

Now, let's go back to the 1,000 mile challenge, and knock 3-4% off of all the Garmin users.

The most my Garmin has ever been out on a properly measure course was this year's Cork marathon that came up as 26.5 miles on my Garmin - that's still only 1% difference.

I'm absolutely sure your 3-4% estimation is way out (it's inaccurate ). After 2.5 years of using my Garmin I'm absolutely convinced the actually displayed accuracy is within 1 %.

Btw., when I first started using it I found to my disgust that a lot of my standard training courses had actually been slightly short, not long. They had mostly been "measured" using one of mapmyrun's predecessors. If anything, the Garmin now has me run longer, not shorter.

oldrunner

plodder wrote: »

I tested my Garmin on a track once and it measured 401.2 metres, which was reasonably accurate. Maybe, it was a fluke, but even though, each individual point was metres off on the map, the overall path followed the shape of the track and was quite accurate for measuring distance.

I run Santry track every week using Garmin 405 for sessions. It consistently over-estimates the length of the track. In two recent 1500m races (last week and week before), it measured the races as 1610m and 1580m. I believe this occurs because of the infrequency of recording track points. On an oval track this causes problems. However, on long runs with a lot of straight lines, the sum of the errors tends to add up and i get very consistent results both in terms of total distances and mile check-points.

Before people provide the 'obvious' answer - I run close to the inside line in lane 1 - some of the overage will be due to not holding this line but most is down to a technical issue:

Standard 400m track.
Each lane's width is 1,22 m, the measurement is made at 30 cm from the inner border for lane 1, 20 cm for the others:

Lane 1: 400.00 m
Lane 2: 407.04 m
Lane 3: 414.70 m
Lane 4: 422.37 m
Lane 5: 430.03 m
Lane 6: 437.70 m
Lane 7: 445.36 m
Lane 8: 453.03 m

A Garmin will sample track point data typically between 1 to 5 second intervals on a track. There is a sphere of accuracy around each track point reading (determined by sat coverage etc etc). The smarts inside the Garmin try to assume you are running in a straight line unless it can determine otherwise by a couple of consecutive readings. Garmins have a tendency to overshoot on corners usually yielding long distance readings because of this.

The slower you move around a track the more sample points there will be, less overshoot, hence a higher accuracy.

Gringo78

TFBubendorfer wrote: »

The most my Garmin has ever been out on a properly measure course was this year's Cork marathon that came up as 26.5 miles on my Garmin - that's still only 1% difference.

I'm absolutely sure your 3-4% estimation is way out (it's inaccurate ). After 2.5 years of using my Garmin I'm absolutely convinced the actually displayed accuracy is within 1 %

I have 5 x 3km races measured on my Garmin all over the same accurately measured course. Distances measured were 2.99 to 3.08km giving an accuracy of -0.4% to 2.74%. Average is 1% long. On any given day though, could easily be up to 3% out! I've run 31 races with the Garmin on certifed courses and the most its measured short has been 0.4% though. Its measured short 3 out of 31 races. I consider myself to run a pretty good SPR race line.

ManFromAtlantis

i would'nt run a race that wasnt advertised as accurate measure. i know for a lot of people they mightnt care but i do.
but when its advertised as such then it shoudl be. i would be very annoyed if it was too short.

just a small point on the 400m track. someone mentioned that its measure 300mm from track edge. if you wear gps watch on left hand then it might come up a bit short (ok only a tiny bit but still !)

Krusty_Clown

For those of you who think your Forerunner is very accurate, can I suggest a test: borrow someone else's Forerunner and wear it on your other wrist and compare the results at the end of your run. If Forerunners are very accurate, the results should be very consistent. Do you think they will be?

0.2%, 2.5%, 4%; it's all largely irrelevant. The important point is, it's accurate enough for our training needs, however, it is not accurate enough to measure a course where the expected outcome is a correct (accurate) measurement. For an accurately measured course you need to invest in a Condo131 (and it goes without saying, you then have to use the course that was measured). If you are advertising an approximate measured distance (and the keyword is advertising) then by all means, measure the course a couple of times using a number of different GPS devices, and make sure you follow the racing line. But don't forget to inform your customers that the course has not been accurately measured.

shels4ever

Just an idea on how it records a track race

http://connect.garmin.com:80/activity/embed/11947547

Don't think i ran in lane 8 during the race

Garmin are a great guide and training aid but stop watch and jones counter for races

ManFromAtlantis

shels. what was the race 3km?

plodder

Krusty_Clown wrote: »

For those of you who think your Forerunner is very accurate, can I suggest a test: borrow someone else's Forerunner and wear it on your other wrist and compare the results at the end of your run. If Forerunners are very accurate, the results should be very consistent. Do you think they will be?

Not disagreeing with you on the main point here, but I'd say two identical watches would be very consistent, if you have an open view of the sky and you wear them on the same wrist (covering myself with enough caveats there )

shels4ever

ManFromAtlantis wrote: »

shels. what was the race 3km?

Yep 3k

MaroonTam

From my point of view, the distance of a race course is the single most important part. I choose my races based on the distance and where it falls in my training, so I expect it to be the distance I was planning for.
The time part I tend to record using my own watch, so chip timing would not be as high a requirement.
I ran the 5 mile last year, this year was to be on the same course. This should have allowed a direct comparison of progress....

Condo131

shels4ever wrote: »

Just an idea on how it records a track race

http://connect.garmin.com:80/activity/embed/11947547

Don't think i ran in lane 8 during the race

Garmin are a great guide and training aid but stop watch and jones counter for races

What I'd like to know is what you were doing by the lamp post near the 100m start, where you appear to have left the track altogether?

lecheile

Krusty_Clown wrote: »

The important point is, it's accurate enough for our training needs, however, it is not accurate enough to measure a course where the expected outcome is a correct (accurate) measurement

Krusty - I believe that this is the crux or the issue. There is a requirement to use the correct tools (i.e. Jones' counter) to verify the course distance, but it needs to be used by a suitably qualified operator before a course can be confirmed as accurate - (that is of course until the condo131 becomes a commercially available measurement device ). Without the correct skill being applied to the task, you are likely to be back to the same margin of error to that provided by a GPS watch.

The AAI (and others who issue permits) have a role to play in verifying the existence of course certificates before any permit is issued. And those who are arranging properly measured races should be asn explicit as possible about the course measurement (name the course measurer if he/she is happy to be named) and put it up to the self-measured / roughly measured events.

Krusty_Clown

plodder wrote: »

Not disagreeing with you on the main point here, but I'd say two identical watches would be very consistent, if you have an open view of the sky and you wear them on the same wrist (covering myself with enough caveats there )

It was actually a rhetorical question. I don't think they will be.
I'll see if I can borrow Mrs Garmin's watch this evening for my 15 mile run. It's identical to mine in all but colour (though I should check the bios/gps versions). We may end up running at the same time though, so I'll have to try my experiment some other time.

shels4ever

Condo131 wrote: »

What I'd like to know is what you were doing by the lamp post near the 100m start, where you appear to have left the track altogether?

Thats where my supprot crew were for water stops and massages...,

What i liked was my 4:09 pace.. think i should be my race pace now for all races...