"do we really have to map this explicitly with relations? Can’t you already see them from the waterway and ridge data?"
1) Ridges are missing in many parts of the world, partially because they are not rendered, but also because it might not be clear how they can be useful. The presence of watershed relations, mapped along ridges, would encourage other mappers to add the missing ridges. 2), while a ridge has to have a certain amount of slope to be called a ridge (perhaps at least 5 or 10% grade?), watershed boundaries are sometimes very shallow. If you look at the examples I posted, you can see that the watershed boundaries in the valley are not obvious without looking at topographical data or surveying in person (I used opentopomap and survey). Some of the high watershed on plateaus also have the same problem. But the watershed or basin divide line is verifiable by finding the highest between basins. 3) Watersheds may already be available as open source data. Ridges must be drawn with increasing elevation and cannot meet waterways, so it would be wrongt to import watershed boundary data directly as a serious of ridges. But with this relation it would be possible to import a watershed as a way or ways with the role "outer". Hopefully mappers would later add natural=ridge tags, after breaking up the ways at saddles and peaks so that they can be drawn in the correct direction (uphill). 4) In my area, aerial imagery over mountain is poor due to cloud cover, but the SRTM elevation data is pretty good. So I can easily draw the watershed boundary line by following topography, but it isn't possible to add every segement of river and stream in the mountains. I believe this would make it difficult to calculate the correct watershed boundaries Many segments of waterways are incomplete (especially outside of Europe). Sometimes this is incomplete data, but it can also be a real feature. For example, the Wolo River (https://www.openstreetmap.org/relation/8687500) goes underground in a limestone area and reappears kilometers laters. There is even a sharp ridge between the two valleys. There are at least 2 or 3 other big rivers in my area that do the same thing. Analyzing the ridges and waterways alone would miss this. I started thinking about this after seeing how Imagico used ridges to help separate waterways and find bad waterway data, and he mentioned that there are not enough ridges, and they are also sometimes unreliable ( http://blog.imagico.de/river-generalization/). The existence of a watershed relation would show that someone took the time to check all the ridges. Also, the most prominent ridgeline is not always the same as the watershed boundary. When a mapper is making a watershed as a complete boundary, they will have to complete the gaps, and will be encouraged to find out the basin divide line even in flattish areas. Joseph Eisenberg PS: Any thought about the name of the key and value, and the type of relation that is best; boundary or not? On Thu, Sep 13, 2018 at 4:06 PM Martin Koppenhoefer <dieterdre...@gmail.com> wrote: > do we really have to map this explicitly with relations? Can’t you already > see them from the waterway and ridge data? IIRR, people have been rendering > maps for these in the past by just analyzing existing waterway data, no > need for explicit watersheds. > _______________________________________________ > Tagging mailing list > Tagging@openstreetmap.org > https://lists.openstreetmap.org/listinfo/tagging >
_______________________________________________ Tagging mailing list Tagging@openstreetmap.org https://lists.openstreetmap.org/listinfo/tagging