[Tagging] Peak-based mountain range proof of concept

[Joseph Eisenberg]

I don't think this option is better than drawing a mountain range as a way
which follows the highest ridgelines between the highest peaks. Relations
made up of only nodes are rare and somewhat hard to visualize. Using a way
(or a relation made up of ways, e.g. natural=ridge ways) is easier for
mappers to maintain and see in the editor. A way is the more verifiable
option because  it simplifies the mountain range to the highest ridge
rather than encouraging including all possible peaks, and for linear
mountain ranges the main peaks are most likely to be agreed upon while the
outer limits are often vague or arbitrary.

Whether you pick some peaks or follow the ridges, database users will need
to do some processing to place a label or estimate the area of the mountain
range.

This is fine: cartographers who are making quality maps still need to place
labels by hand, since algorithms still produce lousy results. If you do
want automated label placement, then it's not much more work to create a
convex hull (a bulging shape) around the peaks or line of the mountain
range, or for possibly better results you can use a digital elevation model
to find the contours of the mountains and estimate the shape and size of
the mountain range. For example, OpenTopoMap already processes a DEM to
rotate the natural=saddle icons to match the shape of the terrain. (
https://github.com/der-stefan/OpenTopoMap/issues/54)

Some other issues: it appears that this relation is mostly the same as the
set formed by "all natural=peak features in the State of Vermont", with
some omissions which are hard to explain.

Why doesn't the mountain range continue south into Massachusetts or north
into Canada? Is this a Vermont-only concept? Why is it only including the
peaks in the center of Vermont, but not the hills farther east or west?

Why not include https://www.openstreetmap.org/node/356555077 when 3 peaks
just a kilometer to the west are included? (Perhaps that is just because
this is an experiment?)

-- Joseph Eisenberg

On Mon, Dec 28, 2020 at 7:24 PM Brian M. Sperlongano <zelonewolf at gmail.com>
wrote:
>
> In light of the ongoing "how to tag a mountain range" discussion, I
created the following object which maps the Green Mountains in Vermont,
USA.  As a flatlander, I had substantial help from a local Vermonter, who
understood the topography and which peaks should be included.  It is
modeled simply as a relation of all the mountain peaks in the range:
>
> https://www.openstreetmap.org/relation/12102399
>
> I offer this as a proof of concept of one possible way to map a mountain
range, hopefully to help further the discussion.  I am not necessarily
advocating for this scheme or the specific tagging that I have on that
object (and I'm not planning to write a proposal around it), but I thought
it would be useful to demonstrate a concrete example that we could all look
at.
>
> The relation contains a bit over 200 peak nodes, which is a completely
manageable size.
>
> Advantages of this approach:
> 1. Point cloud still yields the approximate location of the range as a
whole
> 2. There is sufficient data to algorithmically determine the location and
size of a label
> 3. Does not introduce polygons into any editor
> 4. Eliminates disputes over where the boundary of the "range" lies
> 5. Whether or not a peak is in a mountain range is reasonably verifiable
(based on local knowledge, mountaineering club definitions, etc.)
>
> Disadvantages:
> 1. Still need to make decisions about whether peaks are in the range
based on geography.
> 2. Need to decide whether all peak objects are included or only the most
prominent ones.
> 3. No definitive way to determine whether a point is "inside" the
mountain range geography (though one might imagine an algorithm based on
the point cloud)
> 4. More work to map than a rough, fuzzy polygon
>
> _______________________________________________
> Tagging mailing list
> Tagging at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/tagging
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/tagging/attachments/20201229/ba328b1e/attachment-0001.htm>