On 7/27/2016 5:11 AM, Kevin Kenny wrote:
The immediate question: I have the boundary multipolygon for a large
state park. The park has several stretches of waterfront. In some
places the boundary of the park follows the high tide line. In others,
it's set back from the shore (and the waterfront may have another
owner). And in other cases the boundary extends far offshore (which
may have implications for boaters). How best to divide and tag it so
that the park exists as a unified entity, but does not result in
rendering land or trees in the water?
----
Now from some gratuitous ranting, because I'm getting discouraged:
:-)You are not alone! My ideas below, feel free to disagree.
ACCESS RESTRICTIONS
My last question here, regarding how to tag public lands for which
permission is required (but routinely granted) got answers that left
me in a deeper state of confusion. The general consensus seemed to be
"there is no difference between those and private lands other than the
personality of the landowner, and they therefore must be tagged alike:
access=private". That answer did not satisfy - I want a map that
renders those cases differently, and things tagged alike cannot be
rendered differently." Moreover, I don't hold out much hope that a
formal proposal, wikified and voted, would end any differently; the
voters are mostly on this list. (Also, nobody answered my question
about how to initiate such a proposal.) I'm leaving Long Island
mistagged with "access=yes" and not touching the "access=permit" on
the New York City watershed parcels that I imported a few months ago
(without a peep on "imports" about that detail of the proposal).
So that particular aspect of the project is "on hold" for now.
If you want finer detail then consider adding a sub tag e.g.
access=private
private= ... umm just how would you distinguish between a wilderness are
where access permission is very restrictive and native lands where
access is just a paper application that is easy to get and native lands
where permission takes, say, 3 months of processing? You would need to
consider what values to have for the 'private' key. Once you do it and
have some practicae at it .. add an OSM wiki page on it describing what
it is so others know. I have a few wiki pages to add ..
sport=hammer_throw, discus_throw, long_jump etc.
FOREST BOUNDARIES
I'm also trying hard not to resurrect the argument about "forests."
The general consensus is that there simply is no way to tag the case,
important in the US, of "a tract of land legally managed for wild-land
resource production (wood and other products)." In this community,
that idea simply cannot be separated from "land covered with trees".
There are also other confusing ideas such as a "natural wood". The
last, it appears, means either also "covered with trees" or else
"virgin stands of old-growth forest", and also appears to connote
"unmanaged" - which is a contradiction, since our few remaining tracts
of wilderness are managed intensively to keep them that way. I've come
to accept that any correct tagging will not render,
I and others take this view
natural=wood ... any area covered by trees ('natural' or not, managed or
not)
landuse=forest .. any area of trees used to produce wood products -
lumber, sap, oils etc.
and most nearly correct taggings will suffer from rendering gaffes
like trees in water. (The concept of "a pond in the forest" apparently
is sufficiently foreign that the phrase, on this forum, is nonsensical
to the point of being meaningless: "surely you mean a pond SURROUNDED
BY the forest?") So I do the best I can to tell as few lies as
possible while still choosing a tagging that will be visible on the
renderer, recalling that "boundary=protected_area" does not render. I
don't expect, given the amount of progress toward rendering it in the
last two or three years, that I'm going to see rendering of protected
areas on any maps I don't produce.
Lets concentrate on the tagging - rendering is a separate issue with
many choices.
A pond/lake in a forest - don't tag the water area with an area of trees.
Don't confuse an administrate area (state park) with land cover areas
... these are separate features and should have their own separate OSM
existence - and those OSM existences should be independent.
That's fine, I can live with doing my own rendering, although it
increasingly means that I have to keep my own data on the side because
there's no way to represent it semantically in OSM's tagging
structure. It's at worst an inconvenience.
STATE PARKS (and many other types of public land)
New York, like many US States, has a system of "State Parks," which
are land managed primarily for the purpose of public outdoor
recreation. (Some of them have secondary purposes such as resource
conservation. In particular, the large parks near the New York-New
Jersey border exist at least in part to protect watershed for the
cities of New Jersey.)
Many, if not most of these parks, particularly the larger ones, are
multiple-use areas. They correspond roughly with "national park" in
the IUCN system - but I'm reluctant to use that terminology, since
they are not administered at the Federal level. "National Park" is a
specific term in the US, and it does not apply to State Parks. In any
case, "boundary=protected_area protect_class=2" seems made for them,
and IUCN appears to allow for the case where a government other than
the national one could designate such a thing. (On the other hand, on
their site, they accord New York's wilderness areas a protection class
of VI - while they enjoy virtually the strictest protection of any
wilderness areas in the country, and in my opinion are class Ib, and I
tagged them thus.) So, I'll accept that "state park" <=> "protected
area". That doesn't help me with rendering, of course. It'll just be a
blank spot on the map.
You are lucky. In Australia there are 'National Parks' administered by
the Federal Government and 'National Parks' administered by State
Governments! These can be tagged with the appropriate operator= tag for
identification of that.
So, what do to to make them show? "leisure=park" doesn't feel right.
The state parks I'm working on aren't small green spots in the city.
Some have large tracts of backcountry. A typical trekker will spend
2-3 nights in the woods on a trip from Greenwood Lake to the Bear
Mountain Bridge through the parks. "landuse=forest" sort of works,
except that every documented meaning of that tag is a lie. The land is
not managed to harvest forest products; in fact, the taking of trees
is forbidden. Nor is the land, in most places, entirely covered by
trees; the parks encompass lakes, marshes, scrublands, and even
developed sections.
But all right. I'll settle for either "landuse=forest" or
"leisure=nature_reserve" as something that at least does not misrender
horribly, and note that such tagging is retained for the benefit of
legacy renderers.+
Yuck.
Your 'state parks' are administrative boundaries. Don't tag for the
render.
PARKS AT THE SHORELINE
But now I trip over another issue. A fair number of state parks lie on
the shoreline, be it the ocean front, the Hudson River, one of the
Great Lakes, or any number of smaller waterways. In many cases, their
legal boundaries extend for quite some distance offshore. This is
significant - it has implications for boaters, for instance - and so
we want to keep the boundary (which militates for
leisure=nature_reserve, which at least shows SOMETHING).
If we were to tag landuse=forest or leisure=park or most such things,
we would run afoul of the fact that the current renderer will render
such areas as land or at least overprint patterns such as trees on them.
The argument could be advanced that the "human" perspective of the
park is that it consists of the land portions, plus some offshore
regulated area that's a different animal. And I suppose that I could
therefore carefully carve out water from the legal boundary, and
create two different multipolygons that share most of the ways, one to
carry the "protected area" designation and the other to carry the
"landuse". That's starting to get into some really detailed work for
what I hoped would be an initial sketch (but see below for some
philosophy discussion).
SO THE IMMEDIATE QUESTION ABOVE: "How would other people divide and
tag a state park whose property line extends offshore?" (The
particular case I have in mind is fairly complex; there are places
where the park's boundary is coterminuous with the high tide line,
other places where it's set back some distance from the water, and yet
other places where it extends far down the foreshore or out into
permanent open water.)
Tag the state park boundary as an administrative boundary, don't include
any landcover tags on it.. the landcover should be a separate area/entity.
TOP-DOWN vs BOTTOM-UP MAPPING
It gets tremendously more complicated if detailed land use and land
cover are needed. Then there will be complex webs of multipolygons,
sharing some but not all of the ways. As a matter of fact, I have a
strong preference for NOT sharing ways among, say, preserve boundaries
and things such as "natural=wood" because they make editing really
complicated and seldom describe the situation in the field. Trees grow
where they will unless humans remove them. They are no respecters of
property lines.
Is that what everyone else does? Do even the roughest sketches
(drawing the boundaries of parks, for instance) with webs of
multipolygons sharing many of the ways, so that one set of
multipolygons can be tagged for the protected area, another for the
land use, and still another the land cover? If so, it seems to make
for an editing nightmare.
Yep.. I'm on the side of separating the landcover and landuse ways ..
they only follow one another for things like 'state forests' that are
used for lumber production (primary use, secondaries of recreation and
conservation)
Is there really no way to make these things approximately correct
without metre-by-metre analysis of land cover and land use? (Without
detailed exploration, which may not even be allowed, I surely cannot
tell a "natural" wood from a "managed" one. I can see indications,
such as the predominance of a single species of tree, or a narrow age
distribution of the trees, but these cases can arise naturally as
well. And I don't even know what "managed" means! Is a
highly-protected wilderness area 'managed'? Or does 'managed' imply a
plantation? I get conflicting answers.)
If I cannot tell then it is natural=wood, covers the existence of all
trees - a landcover issue. . what they get used for is a landuse issue.
I stay well away for the 'managed' thing .. I think that is just confusing.
What I tend to hear in this community is, "you're going at it all
wrong." Apparently, the only way to approach it is to start with your
own back yard and map only fine details, and then aggregate the
details into coarser structures. I'm getting the feeling that mapping
large objects such as parks, and then filling in the details of what
is in the parks, is something that the community deprecates. I do seem
to hear a subtext that I shouldn't care about where a park is until
I've mapped its interior in detail.
If that's really the One True Way of Mapping, then the work that I've
done trying to get public land boundaries in New York sorted out is
largely in contravention to it. I hope that's not the case, and I'm
just running into the sort of special cases that might be expected in
a state whose largest park is larger than the nation of Slovenia.
I suppose, otherwise, I could stick to detailed mapping - the sort of
thing that I did last weekend with
https://www.openstreetmap.org/changeset/41001386#map=15/42.8139/-74.1317&layers=N.
But I really want my map to show where the state land begins and ends,
so that I know where I'm allowed to travel before I get there. And it
would be nice if I don't have to go slogging through beaver swamp to
survey ponds and wetlands before I can have a polygon for the State
Forest - which, after all, is mostly wooded and ostensibly managed for
forestry. In practice, the land there is resource-poor enough that
there's little timber harvest, so it's more a recreational area. (And
in practice, few people want to recreate in that swamp. Now that I've
mapped the major features, I probably shan't be back. The only
important aspect was the long trail that traverses it, and I've got that.)
There are those who will only accept precise data and prefer a blank map
if precise data is not available. The 'there be dragons' approach to
mapping. I take the view that indicative data is better than no data!
So I would map what I suspect to the best of my knowledge is the
situation on the ground. Then when I go there I would add data -
increase the precision of the existing data (mine or others). The
important thing about this 'indicative data' is to include a good source
statement ... so people can see how vague or accurate it is.
I won't even get into asking about how to tag seasonally-varying
wetlands, or ones that vary on a longer cycle. Beavers return to the
same areas fairly predictably. They may be grown to meadows or alder
swamps, but won't grow to mature forest before they're flooded again.
It's nice to have these cyclic areas mapped, since they will be muddy
but negotiable most years, and inundated and impassable in some. But
that's way more advanced than I'm trying to get to at present. I'll
settle for parcel boundaries, open water, and trail alignment.
There are also intermittent things - not to be confused with seasonal.
And then there are a few things that are both seasonal and intermittent.
_______________________________________________
Tagging mailing list
[email protected]
https://lists.openstreetmap.org/listinfo/tagging
