# Best way to map boundaries when rivers or streets define the limits

I have found three mapping styles for a boundary area where one of its edges is legally defined by a river. I see pros and cons for each one, and I want to read your thoughts.

Suppose there is a river that connects nodes 5,1,2 and 6 as this.

``````5      1      2      6
+======+======+======+
|      |
|      |
3 +------+ 4
``````

The 3 styles are:

• Share All: the relation for the multipolygon area has at least 2 outer members: 1-2 river, 1-3-4-2. There are other segments, 5-1 and 2-6, for the rest of the river.
``````5      1 1      2 2      6
+======+ +======+ +======+  waterway=river (3 segments)

1      2
|      |
3 +------+ 4
``````
• Share Nodes. The relation has at least 2 outer members: 1-2 river, 1-3-4-2. There is another element that share nodes for the river.
``````5      1      2      6
+======+======+======+    waterway=river (1 long segments)

1      2

1      2
|      |
3 +------+ 4
``````
• Share Nothing: A different boundary element is mapped along the river, which does not reuse the nodes from the river.
``````5      7      8     6
+======+======+======+    waterway=river (1 long segments)

1      2

1      2
|      |
3 +------+ 4
``````

The last one (share nothing) will look like this on the map:

``````5      7      8      6
+======+======+======+
1 +------+ 2
|      |
|      |
3 +------+ 4
``````

I map with Share Nodes style. Using this method, if the river changes its course when modifying the river nodes, the boundary legally defined by the river is also changed. Relations are clearer because all members are only boundary=administrative+admin_level=*

Share Nothing could be appropriate when the boundary is defined by the current riverbed. But if the river changes, the boundary will keep its area. With this style, the two elements have different life cycles.

Finally, Share All could be used to reduce the number of elements in the database.

What is the best way? what is your approach?

I map in a region where some kinds of boundaries are normally defined as following roadways and waterways â€“ more specifically, road centerlines and river centerlines or thalwegs. Other kinds of boundaries are defined based on a land survey, but the roads are laid out so that the road centerline follows the boundary precisely, giving the government on either side jurisdiction over half the roadway, resulting in differences in addresses, road signage, and more. Whatever the cause, the boundaries align with these other features; thereâ€™s value in explicitly indicating that in the database.

Originally, mappers in this region used to make the roadways and waterways members of the boundary relation. But this made the boundary relations brittle and prone to breakage, as we would routinely have to split the roadway into dual carriageways or transform an intersection into a roundabout. (The boundary hadnâ€™t moved, because the metaphorical â€ścenterlineâ€ť hadnâ€™t moved.)

Sometime about 12 years ago, we switched to drawing distinct boundary ways that share nodes with roadways and waterways. This made the previous kinds of breakage somewhat more difficult to carry out by mistake, but it introduced other kinds of mistakes, like connecting a driveway to just the boundary but not the roadway, something that is very easy to do in some editors.

Thereâ€™s also the option to leave the boundary disconnected from the other features. Itâ€™s a good idea whenever weâ€™re unsure whether the alignment is intentional or a mere coincidence. Itâ€™s also a good idea if the boundary follows something other than the centerline, strictly speaking. But if we know itâ€™s intentional and correct, then keeping them disconnected means the features can diverge over time for reasons that have nothing to do with the boundary, just because mappers werenâ€™t paying attention to the boundary when refining the roadway.

Itâ€™s useful to consider whether the boundary automatically changes in response to physical changes to the roadway or waterway, but there are other considerations too. Some kinds of boundaries, like a cityâ€™s formal neighborhood boundaries, are defined by the roadway as a whole and donâ€™t need to be clarified beyond that. Some waterways arenâ€™t good candidates for sharing nodes because the â€ścenterlineâ€ť is unclear or subject to interpretation.

2 Likes

When Iâ€™m mapping a road which is also an admin boundary, Iâ€™ll admit to moving the boundary slightly, so it becomes a series of very shallow zig-zags crossing the road centreline.

Yes, thatâ€™s moved the boundary by possibly 30cm / 1â€™, but it avoids the other problems mentioned!

I would recheck are they defined by

• current left river shore
• current right river shore
• current thalveg
• natural thalveg - if course of river would be changed by human structure it does not apply and does not change boundary
• some other current river element
• river element, as it was at some specific date
• current river thalveg, but boundary to be updated needs to be resurveyed by joint commission (applies for example to some international borders)

In many cases boundary is seemingly defined by river, but changing course of river would NOT redefine border. In some cases you can see old river passage by looking at administrative boundaries.

The same applies to roads, but there realigning road is even less likely to result in legal changes to boundaries. Note that sometimes you will have clause buried in document to effect â€śchanges of road/river geometry after XXXX date does not change boundaryâ€ť

(I can give some specific examples if anyone would be interested)

Yes, I should add that the kinds of boundaries that simply incorporate a road centerline or thalweg by reference tend to be boundaries where the stakes are relatively low. As the stakes rise, the courts may be called upon to clarify further. Where I map:

• State and county lines are literally defined in terms of rivers per se, such as â€śdown â€¦ little Miami Riverâ€ť. However, courts have applied common law or forced a settlement to determine when the boundary follows the thalweg, versus one of the banks, versus one of the banks as of a certain date, versus an arbitrary reconstruction of said bank.

• County lines follow survey lines, and roads are laid out along those county lines. One side of the road has a different name, route number, traffic laws, and addressing scheme than the other side. Both the legal concept of a centerline and OSMâ€™s `highway=*` way are abstractions. They donâ€™t necessarily follow either the precise middle of the roadway cross-section or the double yellow lines separating opposing traffic, which may shift from side to side based on lane configurations. Practically speaking, distinguishing the two sides of the roadway is much more important than capturing the boundaryâ€™s precise coordinates.

• Most city limits are defined in terms of survey pins deliberately embedded in the centerline of the roadway. This level of precision matters to abutting property owners, who are taxed based on the road easement, and also to law enforcement and construction crews. But mappers in my region have very little interest in nanomapping to that degree, so itâ€™s good enough to conflate one abstraction with another.

• Most formal neighborhood boundaries are defined in terms of this or that road, without clarifying which part of the road. The neighborhood only becomes tangible somewhere past the curb in the form of signage or street furniture.

1 Like