[OSM-talk] What Streets are in what Places

[Peter Childs]

2009/11/13 Andrew Errington <a.errington at lancaster.ac.uk>:
> On Fri, November 13, 2009 16:43, Peter Childs wrote:
>>
>> Any ideas.
>
> You could calculate node density (nodes/km2) and assume that node density
> will decay from the centre of a town to the edge.  This would work for the
> nodes in ways, since 'in town' will have more streets than 'out of town'.
> A rural area with winding roads might have an increased number of nodes
> (to get smooth curves) but it would have fewer roads.  You could also use
> POI density, on the assumption that there are more shops, hospitals, pubs
> and restaurants in a town, and the density drops off out of town.  Set a
> threshold, and mark the border between above threshold (in town) and below
> threshold (out of town).

Sounds as good method as any other. Might be able to use where
Churches, Market Squares, Pubs, Shops, Schools etc to work out where
town centres are, If we can do that we can have a list of possible
town centres without names. after all if somthing has the name
"Wainscott Primary School" and we have a village of "Wainscott" nearby
one would tend to presume that the school must be within the bounds of
the village.

>
> Places like London would be tricky, as the node density would be high
> across the whole area, so maybe you can't pick out the individual towns.

However, land use might help. Commercial Zone tend to be near the
middle and residential/industrial on the outskirts, however then they
go and build that big new out of town shopping centre.

Historically towns appeared at major road junctions, river crossing
etc. However in recent years they tend to mean the town ends, after
all who crosses that major dual carriageway that circles most major
towns these days. (After all we have to build our bypasses (Hitch
Hikers Guide))

>
> Anyway, please do not give these ideas any more credence than those from
> any other random internet source. :)
>
> Best wishes,
>
> Andrew
>

Of course not.

Peter.