[Talk-us] place=neighborhood on subdivisions?

[Minh Nguyen]

Vào lúc 18:40 2020-09-22, Paul Johnson đã viết:
> 
> 
> On Tue, Sep 22, 2020 at 8:36 PM Mike N 
> <niceman at att.net 
> <mailto:niceman at att.net>> wrote:
> 
>     On 9/22/2020 9:26 PM, Paul Johnson wrote:
>      >         The extra hamlet nodes are import remainders that haven't
>     yet been
>      >     converted to landuse areas.   The general landuse zones for
>     that area
>      >     have been identified, but do not exactly correspond to the named
>      >     subdivisions.   As I get a chance to survey, I divide the
>     landuse into
>      >     subdivisions and convert the node to a named area for the
>     subdivision.
>      >
>      >
>      > Please don't expand these as landuse, please expand them as
>      > place=neighborhood instead.  Landuse polygons should be congruent
>     to the
>      > actual land use.
> 
>     That's a good point: the subdivisions often contain one or more landuse
>     basins, clusters of trees, etc.   I've been thinking of them as one big
>     blob, but it seem correct on a more micromap level to mark them as
>     place=, and identify the smaller landuse areas (which are sometimes all
>     residential).
> 
> 
> Exactly.  My rule of thumb is if you're thinking about putting a name on 
> it, and it's not a shopping center, apartment complex or similar large 
> but contiguous landuse, then landuse=* probably isn't what your polygon 
> should be.

It's common, intuitive, and IMO beneficial to map a planned, 
suburban-style residential development as a single named 
landuse=residential area. These developments have well-defined 
boundaries and are primarily residential in character. If there are some 
wooded lots in the subdivision, it's perfectly fine to map a 
natural=wood area inside of or partially overlapping the 
landuse=residential area, ideally without being connected to it. This 
approach is supported by longstanding documentation [1], old threads 
[2], and good support in both renderers [3] and search engines.

There have also been old discussions where folks have conflated the 
concept of landcover with landuse. [4] But I find this approach overly 
academic. Taking it to the logical extreme, landuse=residential would 
only be coincident to each house in a subdivision, given that the yards 
are non-dwellings.

I don't see the need for a fundamental distinction between planned 
residential developments consisting of multi-family apartments and those 
consisting of single-family houses, such that the former would be mapped 
as a coherent landuse area but the latter would be a shapeless place 
point. Where there's no such distinction, the landuse areas lend 
themselves to ab intuitive rendering that's good for navigating suburban 
sprawl. [5]

If a planned development truly is actually mixed-use, and not only in a 
garden-level micromapping sense, then something other than landuse=* 
would be reasonable, since a particular landuse doesn't characterize 
that development anyways. Named landuse=residential areas also don't 
tend to make as much sense in urban areas, older inner suburbs, and 
rural areas. But the areas in changeset 91255294 aren't mixed-use 
developments; they're residential areas in a suburban setting.

[1] https://wiki.openstreetmap.org/wiki/Special:Diff/1087300
[2] I previously wrote on this topic in 
<https://lists.openstreetmap.org/pipermail/talk-us/2013-June/011131.html> and 
it seemed like other respondents were taking the same approach.
[3] https://github.com/gravitystorm/openstreetmap-carto/issues/351
[4] 
https://lists.openstreetmap.org/pipermail/talk-gb/2017-January/019811.html
[5] https://osm.org/go/ZTVSa4OB

-- 
minh at nguyen.cincinnati.oh.us