[OSM-talk] Night of the living maps 07.02.2012 - a 'virtual' global mapping party

SomeoneElse lists at mail.atownsend.org.uk
Fri Jan 27 23:24:02 GMT 2012


Matthias Meißer wrote:
> Of course we know, that using just Bing, is just the #2 choice for 
> adding details to our database and that survey is what we all really 
> like. But in this case, we believe, that it is a good compromise (see 
> hints in Wiki). Currently there seem to be no definitive answer if 
> this technique is right or wrong, maybe we can decide on this when 
> years passed and we can take a look back. So for now, let's just do 
> what we are good at: Create together a real good map :)

As Nick has said, we need to be a little careful here. It's one thing 
for someone who's familiar with an area to add in features using the 
Bing overlay to help them, but something else entirely to add features 
to the map based only on what it looks like from overhead.  The things 
that are useful to know about countryside roads tracks and paths are 
things like access rights (Am I legally allowed to drive down there?  
What about cycling?) surface (OK, I'm legally allowed to cycle down 
there but would I have to carry it most of the way?) and all of the 
points of interest (gates/stiles/shops/pubs etc.) that you can only get 
details of by actually going there.

If you don't add any of those things you end up with something that's a 
poor facsimile of a Google map. In the UK at least, Google recently 
replaced third-party map data with their own, derived by exactly this 
method - and while it's fine at navigating from town A to town B via 
major roads it's a bit poor for anything else.  As an example, on a 
Google map go to 53.186133, -1.151462 and zoom out one notch.  Other 
than the A and B road that you can see, none of the "roads" that Google 
shows actually are.

Another potential problem is alignment - both GPS traces and imagery 
might give odd offsets in hilly countryside (and as towns tend to be 
built on the flat bits, this might not be something that town mappers 
are familar with).  Some Bing imagery is wildly offset (see for example 
http://www.openstreetmap.org/edit?lat=32.7411&lon=-16.6912&zoom=14) but 
it's often possible to align it better with a combination of multiple 
GPS traces and the local knowledge of having been there.  It can also be 
fairly out of date - if you believed Bing, you'd think that Britain 
still had a coal mining industry.

Perhaps a better approach would be a combined one of adding features 
from Bing etc. AND then going out, getting a bit of fresh air and 
collecting all the useful details that you can't get from an aerial 
photograph?  Sure some things (woods, lakes) aren't easily accessible on 
all sides and aerial imagery is really helpful filling in the gaps that 
you can't get to but things like "who owns that wood?" and "is their 
permissive access?" can usually only be found out by going there.  
Adding things in areas that are neglected locally rather than miles away 
means that there's more chance of a local OSM community maintaining the 
data once added - it won't turn into some zombie representation of how 
things were whenever the Bing aerial photos were taken.

Finally, please don't forget to put a "source" tag on every remotely 
edited object so that when someone comes along later to try and 
reconcile multiple sources of data they have at least got a chance of 
working out what came from where.

Cheers,
Andy





More information about the talk mailing list