I’d like to improve the mapping of a public park near me which consists of different areas.
Within the current boundaries, there are:
1/3 landscaped areas with trees, paved footways, play areas for kids, unpaved open spaces
1/3 buildings which are not accessible to the public.
1/3 is a large paved square which looks exactly like a city square. Under this is an underground car park. The square doesn’t have a distinct name on the national topo map, it’s just part of the park.
The boundaries themselves are partly fences, partly walls with gates, partly open to the street.
What are some best practices to map this?
Should there be one large polygon for the whole thing, or distinct polygons for the square, for the play area, or both?
Should the polygon for the park include the private buildings?
How do I ensure that routing apps can connect the neighbouring streets, the paved square and the footways?
If the whole area is considered the park I would make a big polygon and then map smaller things like the play park and buildings inside it. Leave it to the renderers to figure out what to render on top of what.
Have a look at New York City’s Central Park, it has all sorts of things in it, buildings, roads and even a zoo…
Both. Overlapping or nested polygons are A-OK in OSM!
In my opinion, if the private buildings are within park boundaries, then yes. It’s good to also tag them as such!
Routing generally relies on ways sharing nodes. You should use the usual features such as footways, crossings, parking aisles, etc. to build out a network of routable ways.
If the park boundary is defined of multiple disconnected separate areas, you should use a multipolygon. You shouldn’t use a multipolygon if the park boundary could be traced without lifting your pencil, so to speak.