[Imports] Fwd: Importing West Virginia State Forests Boundary

[Minh Nguyen]

Vào lúc 13:15 2021-08-11, Frederik Ramm đã viết:
> * There's a bunch of bubbly residential areas that nobody tracing from 
> aerial imagery would classify as such. For example, why is 
> https://www.openstreetmap.org/way/863679643 shaped the way it is and why 
> does it have this little hole in the South? If that is the "official" 
> data then maybe OSM is better off without it.

To me, the specific characteristics of the imported landuse data in 
Rhode Island are somewhat off-topic, since the proposal is to import a 
different kind of data in a very different geography. But for what it's 
worth, that little hole is apparently a pond. Reasonable people have 
disagreed over whether to exclude a pond from landuse areas when it sits 
at the edge between the two areas. I suppose the argument is that a 
manual mapper would've added the pond, but don't say that too loudly or 
someone will suggest a water import. ;-)

> * Why do neighbouring large landuse polygons duplicate tens of thousands 
> of nodes instead of being modelled as relations both sharing one way?

If Brian had done that where I map, I would've pleaded for the import to 
be reworked or reverted. Ways sharing nodes, sure, but modeling landuse 
areas as relations just because they border each other would be 
incredibly unfriendly to any mapper who has to maintain the data as 
landuse changes in the future. Even then, there's debate over whether 
landcover like woods should even be connected to landuse=residential, 
due to things like wooded lots, which are very common where I map and 
across much of West Virginia.

I suspect your point about conlation would become more relevant if 
there's any interest in importing an actual landuse dataset, given 
Attila's prolific hand-drawing of landuse areas in West Virginia over 
the past year. But anyways, back to boundaries...

> (For 
> example, since it appears that the plan is to import boundaries here, I 
> would be interested to learn of any conflation plans with existing 
> administrative boundaries if/where state forests should coincide with 
> them?)

Do you mean that the protected area relations should reuse the ways that 
are members of the administrative boundary relations, or that they 
should consist of new ways that share nodes with the existing ways? 
Either form of conflation would be a rather aggressive step that I would 
caution against even with manual mapping. Better to keep the boundaries 
separate unless you know otherwise.

In the U.S., state park boundaries correspond to property lines but, 
like property lines, often don't neatly correspond to municipal or 
county administrative boundaries. I'd imagine this to be especially true 
in states that were surveyed with metes and bounds, like Rhode Island 
and West Virginia. By contrast, boundaries are a bit neater under the 
Public Land Survey System, which is why you often see a checkerboard 
pattern in parklands out west. But just the other day I mapped an 
incorporated town in Indiana (PLSS) that sits inside a state park inside 
another state park, so the boundary situation is far from rational no 
matter the survey method.

The other mitigating factor is that West Virginia is one of the less 
densely populated states, so I'd assume most state forest boundaries 
would be somewhat far away from any municipal boundary.

I do agree that local residents should help to monitor this import, just 
in case there are any blind spots. Luckily, there are a couple West 
Virginians on OSMUS Slack, so I'd encourage keeping them in the loop, if 
for no other reason than they'll be partly responsible for maintaining 
this data going forward.

-- 
minh at nguyen.cincinnati.oh.us