A friend of mine is preparing for a 5km run. They use Komoot to research the routes and they record it with Strava and OrganicMaps. They’re finding quite large differences in distance estimates. Specifically, Strava route plan was giving 2.2km and OrganicMap/Komoot, which both used OSM as source estimate 5km. Upon registering the routes with OrganicMaps the estimate from a GPX was 5.6km for the entire route.
My initial suggestion was this may be a variant of shoreline problem, but maybe there’s another explanation? Can you please advise. And since this estimate is important to us, in the spirit of OSM core values, could you suggest good ways to improve the records in our local area?
Are all these estimates from actually walking/running the route and recording a GPS track? Or are some from drawing a route, or from asking the app to calculate a route? Or did the race organiser provide a GPS route?
E.g. you appear to say that Organic Maps registered 5km but also 5.6km - are these from two different methods?
Can you display these routes visually? I would expect the difference between 2.2km and 5km to show up as a clear inconsistency if you look at both on a map. (Such as the 2.2km route cutting off a large segment of the route).
Recently my GPS watch recorded a 10km race as 10.5km. The race included a fairly long stretch of tunnel where the GPS signal had obviously got lost, and the trace in that area was a complete mess that was recorded as spurious distance. But that was immediately obvious looking at a map of the trace.
I wouldn’t expect the shoreline problem to be an issue if you are routing along streets or paths.
Sorry, I just tried to clarify my initial description. 5km was the route plan, but the 5.6km was the reconstruction from the GPS. I am going to check the signal issue. I don’t think there was a tunnel, but we have significant hill in the area and quite large tree cover, so it’s a possibility.
The reason why shoreline paradox occurred to me, was that in our area many footpaths are not currently represented with high accuracy. The lines are very straight as compared to say Ordnance Survey.
So if I understand correctly, the official course distance was 5km, and the route plans from Organic Maps and Komoot agree with that, so that all seems fine. If you asked those apps to calculate a route along paths/roads, it suggests the underlying OSM data is reasonably OK.
The Strava route plan seems completely out of line. I don’t know how you calculated the route - is it possible you asked it to route from A to B without using enough “via” points and it took a shortcut? Or that it calculated a route “as the crow flies”, rather than following paths? As I mentioned, I would expect this to look obviously “wrong” when viewed on a map. It can’t be following anything close to the actual course, and it is unlikely that a lack of precision in the footpath data would lead to such a large difference.
Finally we have the GPS trace. I don’t think the distance here would be influenced by OSM data or by Organic Maps in particular - it would be a purely mathematical estimate from the trackpoints in GPS data. You could try viewing it using any online tool such as GPX Studio - you will probably see 5.6km as the distance there too. On that site you can see the track overlaid on various maps and also satellite imagery - that might help you to see if the GPS “wandered” off course at any stage. If you can export any of the route plans to GPX you could display them too and see if they look different.
There are many potential reasons why the official distance and the OSM and/or GPS measured one can differ, in particular if this was in any way “official” the length will be the convex hull of the course (probably measured with a Surveyor’s wheel or maybe these days with a RTK set up). If you add not running that line plus GPS issues (overrun around corners etc) 10% difference might be a bit high but still completely possible.
PS: your friend should be doing (long) intervals at their target competition pace and not worrying about the exact length of the course, just saying :-).
I second @SimonPoole , 10% difference is not something to worry about. A consumer GPS trace will always be different from something mapped from someone else’s trace or an aerial or LIDAR scan. Especially so in altitude, there the difference will be even much more pronounced.
Now for an estimate giving less than half of the GPS reading, there must be something else going wrong. I cannot comment on how Strava gets to that.
Regarding the shoreline problem: I witnessed somebody pedantically mapping all the switchbacks of a well-known trail in my area with the result, that OSM routers now stop recommending that trail and rather push their users to take another route over (disused) tracks and informal paths. But I guess that only happens when the respective lengths were not so much different from the start.
Sorry, I must clarify a few things. I did an appalling job of writing up the original problem. I think I edited it too many times and didn’t notice I was switching between half and full distances as the route was not circular.
So firstly, the 2.2km vs 5km. I should have said 4.4km vs 5km as 2.2km was the half distance only.
I will definitely check with the GPX Studio. Thanks for this suggestion, @alan_gr.
Really interesting side effect for the shoreline, @Hungerburg. Thanks for sharing.
Lastly, this was not an official race and indeed my friend does circuits, @SimonPoole. It was just a luncthime discussion for us. We were just a bit surprised about the 10% difference and started looking for reasons. Perhaps wondered off too far with shoreline problem.