I am currently updating the coastline on the west coast of Scotland. Several times I have come across the practice of piggybacking on to the level 4 administrative boundary to add an additional tag for the coastline. This does produce a better coastline than the original PGS sourced coastline, but I think it is a bad practice. I think that the coastline and admin boundaries should always be represented by their own separate ways. Thoughts?
If the boundary and the coastline are actually in the same place, then I think its fine for them both to use the same ways. Note that boundaries are usually mapped as relations, so they may not have a tag on each way. So you could have one boundary relation with sections of the coastline as members.
Though it is worth checking whether the coastline and boundary are actually the same.
I assume you are referring to the council area boundaries, should be admin_level 6 ?
It seems these boundaries usually follow the low water mark, plus cutting across some estuaries etc. You can check the OS OpenData BoundaryLine to see where they actually go.
The coastline in OSM should be mapped as the high water mark. So it is often some distance away from the administrative boundary.
I think its OK to map the coastline and boundary in the same place (and using the same way) to start with, as a rough approximation, or if no better data is available. But where possible they should both be mapped accurately in the correct locations.
If people were only using the admin boundaries for the coastline as an improvement on the PGS data then it wouldn’t be so bad. However, I have seen several instances of where an coastline was piggybacked onto an admin boundary way, and the ‘shared’ way was then moved to the MHWS line. This results in the coastline being in the right place, but the admin boundary being in the wrong place. It would have required less effort to have created a new way for the coastline following the MHWS line, than trying to fit the existing way to the MHWS line.
Piggybacking onto the admin boundary ways is leading to tagging errors in admin boundaries being introduced. Consider this way:
http://www.openstreetmap.org/browse/way/83116148
This coastline way was clearly originally an administrative boundary way (it contains relation information that are administrative boundary related). The administrative boundary tag for Highland has been deleted from this way, and a coastline tag has been entered. The way has also been moved to align with the MHWS line on the OS OpenData Streetview map. It would have been easier to just create a new way for the coastline and leave the administrative boundary as it was. There is now no Level 6 administrative boundary way for this part of Skye. That is not a big issue for the main slippy map, but anyone producing a map using the Level 6 admin boundary ways will have sections of admin boundary missing.
If you look at boundaries on OSM Inspector (Admin Level > 4) then you will see that the mainland part of Highland has been ‘knocked out’ due to a missing link in the boundary relations. You cannot tell from OSM Inspector where the missing relation is. I found the same problem when I was updating coastline in Argyle and it was down to someone using admin boundary ways as coastline ways. It took me several hours to get the tagging back to the way it should be.
Another problem with piggybacking the coastline onto Admin boundary ways is that you cannot see them in Potlatch when editing. Potlatch only displays the way as an admin boundary, and you cannot tell that a way is also a coastline without inspecting the tags. This will likely lead to coastline errors being introduced.
No, I think it was the other way around. That was originally just a coastline way, but it later became part of the boundary relation. It does not need any boundary tags if it is part of a boundary relation.
So the coastline has always been in the correct place (though may have been inaccurate). But the boundary was always in the wrong place.
I added most of the admin boundary for Highland Council. It is a lot quicker/easier to just use the existing coastline ways, instead of importing/drawing new ways for the entire boundary. Yes, the boundary is in the wrong place, but its pretty close. And it is a lot more useful than having a missing/incomplete boundary.
I will try and add more sections of the correct boundary sometime, though it is a fair bit of work to do it correctly, and avoid errors.
Yes, its true that both the coastline and boundary relations can get broken, by people who don’t realise what they are doing. Having a shared way may make this more likely. And maybe the editors could be improved to make it clearer what’s happening.
I would suggest using JOSM instead of Potlatch for these kinds of edits, as you can use the Validator to find coastline or boundary errors.
Also I don’t think it is very helpful to just trace coastline from OS Streetview, it is not always accurate. The Bing aerial imagery is often more useful and up to date.