I've been using osm4routing to filter and formate OSM data as a graph and process it. However for my application (see 1) I don't care about the "insides" of cities (/towns/villages/..), all I need is the fact that there is a city. Filtering out residential streets from osm4routing output is easy (see 2) but this still leaves nodes which represent road junctions and similar. How can I merge all the nodes inside a city to a single node? This node should be incident to all the edges which were connected to the city. I was hoping that there was something which would help me distinguish which node belongs to a city. I tried looking at the addr key but I am not sure whether all objects inside a city have this properly assigned. The boundary information seems also plausible but again I don't know about the reliability and also this seems to be harder to do. Due to the size of the data I am of course looking for a general rule (since I can't do this manually).

1) Analysis of road network on state, or continental level, not city infrastructural level.

2) just select the lines of the edges file s.t. they have number >1 in the fifth column

asked 15 Apr '15, 18:36

ondrejsl's gravatar image

ondrejsl
31112
accept rate: 0%

edited 15 Apr '15, 20:56


This is an interesting question about how to generalise OSM data. In practice there is no easy way.

You could use one of the following ways to try & identify built-up areas:

  • Merge certain landuse classes (residential, commercial, industrial, retail,etc) together with some amenity areas (hospital, university, college, school) and some leisure areas (park, recreation_ground) to create a proxy for built-up areas. This will only work when landuse has been mapped, and it becomes harder when landuse has been mapped in detail.
  • Use the density of residential streets. This has been done using RapidMiner by the GIS group at Heidelberg University.
  • Simply use a standard size buffer around place nodes (say 5km for cities, 2km for towns & 1 or 0.5 km for villages). This could be refined by using voronoi triangles to handle cases where places are distinct but their buffer zones would overlap.

I presume you are ignoring residential streets, so once you have some approximation of built-up areas linked to the actual OSM element describing the place, you need to clip the road network using the built-up areas. It is important that you identify which place a given road belongs to before clipping. This can be done by identifying which roads intersect the perimeter of each built-up area.

The dangling ends of the roads are then connected to a single node representing the city (typically I'd use the existing OSM node or centroid of other elements for this).

This represents a considerable amount of post-processing of OSM data, which requires knowledge of geometries. Personally I'd do this in a database such as PostgreSQL with PostGIS. An additional problem then arises because you may want the data in OSM XML format for building your graph. Mainly this involves re-working the way_nodes table for truncated ways. All additional artificial path links are not really a problem as these can be created with negative ids outside the range of ids in use.

For other post-processing of OSM data for routing take a look at some other applications.

Note the process as described may create inaccurate graphs (specifically a road which loops into a 'city' area and out again may be incorrectly connected to that city.

permanent link

answered 16 Apr '15, 08:30

SK53's gravatar image

SK53 ♦
21.4k44221336
accept rate: 20%

Your answer
toggle preview

Follow this question

By Email:

Once you sign in you will be able to subscribe for any updates here

By RSS:

Answers

Answers and Comments

Markdown Basics

  • *italic* or _italic_
  • **bold** or __bold__
  • link:[text](http://url.com/ "title")
  • image?![alt text](/path/img.jpg "title")
  • numbered list: 1. Foo 2. Bar
  • to add a line break simply add two spaces to where you would like the new line to be.
  • basic HTML tags are also supported

Question tags:

×15

question asked: 15 Apr '15, 18:36

question was seen: 1,500 times

last updated: 16 Apr '15, 08:30

powered by OSQA