A more systematic way yet would be to make a map where you pick a grid of points, say 1 every mile, and then calculate a route from each point to a spot 10 miles east and 10 miles north of that starting point and then sum the distances and store that sum as the value for that point. Anywhere you get points with values much over 20 miles will be places with routing gaps. Some will be natural like rivers, but many will be gaps in our data.
I think there was some work done like this when people were connecting up tiger data along US county lines. But I don't remember who did it or where it was. I don't have the tech skills to code this up, but in theory it is not all that tricky. If someone did make such a map, it would be interesting to see the result. -AndrewBuck On 11/03/2017 07:51 AM, Joseph Reeves wrote: > Hi John, > > I like things like this; I'm not sure if I've got anything useful for you, > but happy to talk about it. I think that routing would be a good place to > start testing. > > For example, in 2009 OSM reported that it was a 7298 km drive from Cape > Town to the Kenya / Ehtiopia border. In 2010 this was reduced to 6666 km: > http://blogs.thehumanjourney.net/finds/entry/route_finding_across_a_continent > > Now it comes back as 6051 km: > http://www.openstreetmap.org/directions?engine=graphhopper_car&route=-33.9157%2C18.4257%3B3.5190%2C39.0547#map=4/-22.55/19.42 > > > According to OSM, the driving distance is roughly 17% less today than in > 2009; I would imagine that the reduction in distance represents both the > building of new physical infrastructure and an improvement in OSM data. > > Routing is a really easy way to look for problems in OSM, such as: > https://twitter.com/iknowjoseph/status/567719622070525953 > > So if you're interested in specific areas, I would suggest creating a route > and looking to improve mapping along it. > > Long term, we could think of ways to conduct this analysis more > systemically. I would like to experiment with something along the lines of: > > > 1. Daily dump of OSM's Africa data > 2. Create 100 start points and 100 end points > 3. Daily create routes from each start to each end point > 4. Log / graph distance of each route > 5. Display each route on a map, coloured by change in daily route > distance > > > Does anyone have a spare server? :) > > Cheers, Joseph > > > > On 2 November 2017 at 21:27, john whelan <jwhelan0...@gmail.com> wrote: > >> If you look at many parts of the map in Africa you can see squares of >> activity. Often HOT projects but there are many many places where a >> tertiary or higher classified highway is fifty meters or less from >> connecting to another highway. >> >> I seem to recall in Ghana they had a project to just map the major >> highways. This is pure infrastructure, it isn't exciting like mapping >> buildings but I suspect it would make the map a lot more usable if we could >> connect up a few more highways and I'm not even sure how you could detect >> them. >> >> Thoughts? >> >> Thanks John >> >> _______________________________________________ >> HOT mailing list >> HOT@openstreetmap.org >> https://lists.openstreetmap.org/listinfo/hot >> >> > > > > _______________________________________________ > HOT mailing list > HOT@openstreetmap.org > https://lists.openstreetmap.org/listinfo/hot >
signature.asc
Description: OpenPGP digital signature
_______________________________________________ HOT mailing list HOT@openstreetmap.org https://lists.openstreetmap.org/listinfo/hot
