Thai-Laos border

In Khong Chiam at the confluence of Mun and Mekong, the border is - for a very short distance - on the right side of the Mekong (i.e. on the “Thai” side). Look at following node:
Isn’t the middle of the Mekong the official border there?

That node certainly looks to be displaced. It’s sitting inside what we presume to be Thailand. Paul_12 did the import so we’ll have to see how he responds.

This post raises some important concerns about rivers and boundaries. I spent some time re-aligning the Moei River a couple of weeks ago. Long ago, major portions of the boundary and the river were made to share the same way and that way wanders all over both banks in places. Clearly that is not correct. I realigned the river in the areas I worked and for a while was taking care to separate the boundary line from the river when possible but it’s a lot of extra work to do that and it got me wondering just what the correct approach should be. I reasoned that these boundaries are only guides. They need not be located with a surveyor’s precision. Indeed, we don’t have the tools to do that. After coming to that conclusion, I began to move both boundary and river to where I thought the center of the river actually was.

In similar fashion, many town and borough (county) boundaries in the State of Alaska follow the course of rivers and streams. All waterways, but especially those in Alaska, inevitably change location over the years. Does the boundary move with them? What is the mapper’s responsibility in such cases? Seeing as we cannot know with certainty where the boundary actually lies, is it really worth the effort it takes to separate boundary and river ways so one can be moved independently of the other?

In the case you’ve cited here Bernard, my tendency would be to move the node and its associated boundary line to the middle of the river. You may disagree. In any case, how can one ever determine with certainty exactly where that node should be?

I follow your thinking … a lot of rivers were very poorly plotted in the early days of landsat, and often I tidy the map by realigning and adding extra nodes when I come across them. Its twice the work where a boundary should follow the river centre but has been plotted as a separate way.
To save a bit of time, in cases like these, I’ll split the boundary, join it to the river and follow the river way as long as necessary. Then I can do the precision/realignment moving the shared border/river nodes at the same time.

But I note that my approach is at odds with Daves - so is it better for boundaries and rivers to share the same way, or plot one one parallel to the other (with a fractional offset) ?

I really dislike working with shared boundaries but it’s often the best way to map certain things. For example, I often use shared nodes to draw a wooded area that borders a pond or lake. When adding a river or stream that is also a boundary of some sort, I’ll copy it so it’s a separate way, offset it slightly and then add the waterway tags to the copy. Both methods have advantages and disadvantages.

Just checked a few river borders of Germany (Rhine, Inn, Danube, Oder). Not all of them use the line of the river, so that seems not so uncommon.
By the way, also Google Earth seems to use the same data: also there Laos reaches the right side of the Mekong at that place.

I understand that the LSIB data used for the import is a few decades old. Some parts may not be 100% accurate, and the course of the river may indeed have changed.

I mentioned this a bit in the thread discussing the import. It appears that legally, the boundary changes with the river. There have been several cases of towns changing hands between Mexico and the US, but I’m not aware of any involving Thailand.

There’s also another point to consider. The international river boundaries between Thailand and Laos (I haven’t checked the others) are defined to be along the “thalweg” (deepest channel) of the river. So technically, it should be mapped separately from the river way, which is supposed to follow the middle line between the riverbanks. (Provincial boundaries follow the middle line, though, and I usually use the river ways as the administrative boundary for these.)

Where the imported boundary is obviously wrong, like here where the line cuts into land, I think we should go ahead and fix it. It’s OSM’s position to follow what’s on the ground anyway. One should keep in mind, though, that as demonstrated in the Preah Vihear case, it’s officially recognised survey maps that end up determining the result of such disputes. (Though mountains obviously move around much less than rivers, so maybe this isn’t really relevant.)

A Wikipedia article says:

Also that indicates that we should have separate lines/nodes for the river and the border.
Overall, the data of that import seem pretty good. It was just such a strange experience when the map showed that I was about to enter Laos while I expected it on the other side of the river (and according to the quote from Wikipedia, the border should really be just a few meters away, because of many small islands were visible there already in December).