Ok the wind aided theory doesnt' make a lot of sense to me at first glance. I found the course map and a vimeo of the course.
http://www.adnocabudhabimarathon.com/routes/
https://vimeo.com/290669353
If you look at the course video, the first half is largely very simlar to the 2nd half of the course, these atheltes are running many of the same 3-4 5km segements overand over with 180 degree turns and the 30-35 k segement is basically the same as the 35k to 40k segement but in the opposite direction. So one would think if 30 to 35k was wind-aided that 35k to 40k would have wind in their faces.
Now the wind direction does make sense for 30 to 35k to be wind aided. At that point, they are largely running to the NE and here is the official weather from Weather Underground (Race started a little after 6 am). The weather wasn't as hot as I thought it was (Weather.com reports of over 70 must have been off). It was in the 60s like accuweather reported but it's not showing a big wind but the wind was coming out of the SE.. Does anyone know if the coastal breeze is a lot different than where the weather station is (i used the generic ABU DHABI INTERNATIONAL STATION)? They were running right on the beach from 30k to the finish.
https://ibb.co/KsQxR2j
Ignore that, thanks to technology, I actually found another weatherundergound station. The ETIHAD TOWERS T4 STATION - which is basically right at the start and finish of the course. Here is their weather report. It's very similar.
https://ibb.co/6DWyPVj
https://ibb.co/1GmnzFs
NOw the t4 station appears to be getting the weather at the top of the skyscraper so its like 700 feet higher than the other weather station. Is it possible there was a big wind on the ground but not that high up? Isn't that the opposite of what normally happens?
Length matters: Here's why I believe the Abu Dhabi course was short
Report Thread
-
-
Wind speed is usually measured 10m above ground and is usually slightly less at ground level due to friction.
-
This is what I posted in the Live AD thread earlier today, it' pretty obvious they DID NOT follow the pre-planned route. Now did they make up for that somewhere else ? I didn't check but there is clearly something that happened at 34K
Hardloper wrote:
Screenshots of what I mean
Course map:
https://i.imgur.com/3tv0vsf.png
Strava data:
https://i.imgur.com/xRjwjIc.jpg
That U-Turn was done at the 34k and it's not what had been planned apparently. It's way worse than what happened at the SF Marathon in 2017 though as that was a misunderstanding on where the cones had to be placed at the U-Turn (Before and not After like they did), here a whole section was removed ?!
Using Strava I measured the missing loop at about 400 meters so at 2:56/km that's about 70 seconds and would put the 30/35k split at a more likely 15:05. -
Not sure why the pics don't show up directly, just click on them to view.
-
RoJo, can you retract your article?
I don’t want LRC feeding slander and doubt. Looks like the course may have been accurate. -
Ncjd wrote:
RoJo, can you retract your article?
I don’t want LRC feeding slander and doubt. Looks like the course may have been accurate.
Yeah, if there's one thing this board is known for, it's being 100% free of slander and doubt -
Nice work!
-
Here is the best piece of evidence for the course being dead-on accurate: rojo is so sure that it's short.
-
You're right that Strava filters your gps data, but I don't think they actually use the heatmap data in the filtering. If they are like many folks who process these data they use a Kalman filter which estimates measurement error and uses it to make a more consistent estimate of your position. This is why a previous poster's statement that gps based measurements are always high is wrong. If you have a trace that bounces back and forth a lot in a truly random way, the Kalman filter will allow a model that travels in the middle of those bounces. This does mean if your track actually bounces back and forth a lot, then the modeled path will be shorter than the actual path. Given all this, I would take the mean of a random sample of Strava traces for the distance as pretty precise, given that the actual route seems to be fairly straight. Someone made a reasonable argument that the estimate could be biased due to strange gps readings (tall buildings and such). I don't know a lot about gps so I don't know how that would bias things. I'd be open to hearing about it.
Bullet the Blue Sky wrote:
Strava actually adjusts (I won't say corrects) GPS data based on their global heat map. Essentially, they use all of the data for a particular route to find and adjust your data to the most likely path. https://labs.strava.com/slide/
Some other notes on accuracy and adjustments ... https://support.strava.com/hc/en-us/articles/216919737-Why-is-Strava-showing-different-data-than-my-Garmin-
Garmin's page on accuracy claims, "Garmin GPS receivers are accurate to within 15 meters (49 feet) 95% of the time with a clear view of the sky. Generally, users will see accuracy within 5 to 10 meters (16 to 33 feet) under normal conditions."
https://support.garmin.com/en-US/?faq=IcyYpjUzRZ8vwH6C107CE8
The bottom line is that making any kind of conclusions based on gps data is a fool's errand. Garmin and Strava are good for logging, but not accurate or reliable enough for undeniable conclusions.
KAV wrote:
had-crut wrote:
I think it proves that Strata is almost as dubious as Global Warming in data quality.
What is Strava gotta do with anything? You know those data come from the watches.... -
Ncjd wrote:
RoJo, can you retract your article?
I don’t want LRC feeding slander and doubt. Looks like the course may have been accurate.
and
Stugotz's Game Notes 99% Written By Mike Ryan wrote:
Here is the best piece of evidence for the course being dead-on accurate: rojo is so sure that it's short.
Video proof that I'm correct.
https://youtu.be/1mjYuaDAhzM?t=99
Your welcome people. Video proof that we've nailed it again.
At Boston in 2011, no one could believe the times except for us who told you to expect the possibilty of something absurd based on the wind.
At Abu Dhabi in 2018, everyone was believing the times except for us who immediately said they were fake.
I'll admit i woke this morning a little worried about what I wrote as people were slamming me on here and I had two emails from titans in the sport making me doubt myself a bit but vindication is so sweet. -
Good job, looks right to me.
-
Here is a post I made in another thread with some more background on what info i received in an email from a guy going by the name of Hawk Eye.
Here is Hawkeye's email to me
Hawkeye wrote:
There is no way the men or the women ran the split for 30km-35km in these times. They ran it short.
The route measured was also not the route run. See attached.
Yesterday they changed the route on the Corniche to one lane from about 2km. The original route was using the two sides of the Corniche. Look at the fly through video on their website.
This meant they missed the piece at the top before the 35km mark. They turned back around only a cone before that piece at the top. So ran short. After changing the route on 5th they couldn't run that piece.
These times for 30km - 35km as they are now are faster than even Kipchoge in Monza and when he broke the WR in Berlin and Paula when she broke the WR for the same splits. Nobody in history has ever run those times for the 30km - 35km. And the times for those in Abu Dhabi are incorrect. This needs to be addressed and not ratified as a legitimate time. It is outside the allowed 1Mtr per Km.
Reading that, was there more off than just the roundabout? Here are his attachments.
https://ibb.co/LkGqTQQ
https://ibb.co/sCqKYVT -
With that said, I overlooked the most obvious bias in the Strava data is that runners likely started and/or stopped their gps devices before and after the finish line.
dingle wrote:
You're right that Strava filters your gps data, but I don't think they actually use the heatmap data in the filtering. If they are like many folks who process these data they use a Kalman filter which estimates measurement error and uses it to make a more consistent estimate of your position. This is why a previous poster's statement that gps based measurements are always high is wrong. If you have a trace that bounces back and forth a lot in a truly random way, the Kalman filter will allow a model that travels in the middle of those bounces. This does mean if your track actually bounces back and forth a lot, then the modeled path will be shorter than the actual path. Given all this, I would take the mean of a random sample of Strava traces for the distance as pretty precise, given that the actual route seems to be fairly straight. Someone made a reasonable argument that the estimate could be biased due to strange gps readings (tall buildings and such). I don't know a lot about gps so I don't know how that would bias things. I'd be open to hearing about it.
-
rojo wrote:
I'll admit i woke this morning a little worried about what I wrote as people were slamming me on here and I had two emails from titans in the sport making me doubt myself a bit but vindication is so sweet.
Do you have evidence that the course that was posted prior and during the broadcast was the one that was measured/certified? It may have been that posted course was wrong. -
2hrs 4mins is the new 2hrs 10mins - average time now
-
dingle wrote:
rojo wrote:
I'll admit i woke this morning a little worried about what I wrote as people were slamming me on here and I had two emails from titans in the sport making me doubt myself a bit but vindication is so sweet.
Do you have evidence that the course that was posted prior and during the broadcast was the one that was measured/certified? It may have been that posted course was wrong.
Or like I posted in the other thread that they modified something else yesterday to make up for the missing 400 meters at the roundabout. That can't be ruled out, however :
- there are no obvious discrepancies in the 5k splits other than the 30/35k
- GPS tracks on Strava are around 42.2 or less, which is quite unusual as everyone who runs marathons knows, you'd except something around 42.5/42.6 on average -
I'd like to see the kilo by kilo splits. I think that would make it very clear if both the lead men and women were pretty consistent most of the kilos then either 33 or 34 was 45-55 seconds faster there's your smoking gun.
-
what dey do wrote:
Proofinthenumbers wrote:
A marathon can be no shorter than 42.17 while taking the tangets.
Nope, it has to be at least 42.195 or it's not legitimate. And that's by the shortest route.
The course measurer will add 0.1% to his own measurement 42.195 meters to ensure accuracy.
As a measurer, we don't add the .1% to ensure the course is accurate to the distance advertised. We add the .1% to ensure that all competitors run AT LEAST the advertised distance. Subtle difference. With small variations in the calibration process and in how the measurer actually rides the course (assuming use of a Jones Counter for the measurement), it is to be expected that two highly qualified measurers will end up with slightly different distances when measuring the same course. The +.1% (Short Course Prevention Factor) is in place as it has been proven that by adding that amount it would be extremely rare that a course would end up being less than the advertised distance.
All this assumes of course that the course is laid out, cones placed, start and finish lines established, etc, in compliance with the measurements taken and map drawn. -
Not sure what to make of all this but the stride/cadence change that would occur between 30k/35 to go from running 14:30s to 13:50s definitely did not happen based on what I saw on the broadcast.
-
Okay about a minute off, with the hot weather and humidity yesterday, I think those 2 will better 2:04.00 minutes in the near future anyways, oh well.