Tuesday, 9 June 2015

GPS device calibration - how does my Garmin 310 XT fare against an "official" distance and time?

I cycled a 100km "sportive" two days ago.
It had an "official" route, and the rider numbers had a timing chip on them, so start and finish times were recorded automatically as we (the riders) went through the start/finish arch.
Isn't technology wonderful.

So, like many others, I looked up my performance on Strava.
I found two "whole course" segments had already been set up by other riders.
The "bench marks" I am using are the "official" route I got from the organisers as a TCX file that I loaded into Garmin Connect, and my "official" "chipped" finishing time (all finishing times are published by the organisers, and available for all to see.

The "benchmark" route was 101.77km (from the "official" TCX route, remember).

So how did the "whole course" segments measure up?

This one comes in at 100.5 km, so sounds right to someone without the "official" route TCX to compare it to - indeed, the rider might have taken a great "line" on every corner and come in just a touch under the "official" route length, so it could be right.
And the time?
It's knocked 7 minutes and 20 seconds off my "official" finish time! It clearly is not measuring all of the route, or the rider took a shortcut (!), or the GPS errors mean it has rounded off some corners etc. etc.
Verdict: Someone has a smartphone (?), and they just chose a section that sounded about the right length.
Update: A route analysis shows that the route start on a straight section on the main road. Useful for perhaps longer training, without having to go down the road to the stadium, but with about a km clipped of the start and end of the "official" route, it is clearly NOT the entire route! The missing couple of kilometres seems to explain the missing time as well.

What about the "other" route?
Well, that one comes in at 103.1 km, and the extra 1.3 km (over the "benchmark" "official" TCX file) is easily explainable if the rider went into any of the water stops, or took a slightly wider line on some of the corners - the shortest route through a corner is not always the fastest! Or, of course, it can be GPS variability.
So how does the time match up?
It's better  - but it has still knocked 2 minutes and 45 seconds off my "official" finish time. Perhaps the segment creator has trimmed his segment down a bit to compensate for all those water stops? Or there is some "moving time" vs, "real time" stuff going on.
Verdict: A decent attempt, but I wonder if I can do better.
Update: an analysis of the route shows it doesn't even start in the Stadium (!) This might be useful for training on during the year, but it is means the distance errors are increased, as the segments omits about 300 metres from both the start and end of the "official" course. It explains the missing couple of minutes, anyway.
But for "training" you might as well just start in the Stadium carpark, rather than a bit further along the route!, so it all seems a bit pointless ...

So I created a "segment" from my own GPS trace (recorded on my Garmin 310xt).
I took some photos at the event, so I know where the start/finish arch was, to a pretty high degree of accuracy.
And, how did my attempt come out?
Well on the first go, I got a distance of 102.8 km.
The "start" was easy to put on the "start line".
For the "finish", I was moving, of course, and I could put the "finish point" a little short or a little over the "finish line". I went for the next data point over.
The extra kilometre can be explained, as above by variations in my riding line (vs the "official" TCX file), or GPS variability (the first half of the course has a lot of trees!). I didn;t stop at any of the water stops, so I didn't make any little detours - just doing the "basic" route.
And the time?
On that first attempt, my Strava time was just 2 seconds over my "official" finish time (remember I chose the data point just over the "finish line", rather than the data point just short).
Verdict: To be within 1 km on the distance, and two seconds on the timing for an event of 100km is pretty accurate!

I'm glad I bought a "decent" device like my Garmin, not just a new smartphone!

Why does it matter?

Well, for a start, the other two routes have crossed over the 2nd and 3rd athletes for performance, so one of the athletes must have had a fast start or a fast finish that the other route has left out!
For a "tailender" like me, that is no biggie, but if you are remotely serious about your training plan, you want the results of your training/riding to be correctly reflected.

So if you want to ride the route, say, this weekend, and you want to see how you would have done in the "actual" event, then I suggest you use "my" segment, and knock a couple of seconds off, as it is far and away the most accurate Strava segment for the 100km course at the timne of writing - indeed, "my" segment has cut the error in timing for the "best" segment available from 2 minutes to just 2 seconds!

No comments:

Post a Comment