Coastline problems Chesapeake Bay

Communities United States
1

We have a problem with the coastline again in the Chesapeake Bay. Coastlines have been changed to go through the middle of open water. I have tried raising the issue here, but I guess nobody saw that. Anybody here who cares enough to track this down and correct this? For some reason this happens every few years and I don’t have the time to track and correct this myself again. Coastline processing on osmdata has been halted until this is fixed.

2

This appears to be the result of organized editing by the Microsoft Open Maps Team by a Serbia-based editor.

See this changeset, for example, where the coastline was moved oceanward:

Here are links to their wiki page and github.

3

Struck a note. The University of Virginia has a particular interest in the bay watershed with their Computing for Sustainable Water project. Not learned if they’re in any way involved in applying their data to OSM but there’s even a game U.Va. Bay Game on the area.

4

Looks @Aleksandar_Matejevic made a series of changesets in the the Chesapeake Bay area. Here they are in order:

I don’t have time to look at them all in depth, but the initial ones seems to have mainly removed natural=coastline from ways. It’s not until some of the later changesets that natural=coastline was then added to other ways. The changeset comments all mention “fixing broken coastline” but the changes are far more extensive than that. I’m not sure if all the changes should just get reverted, or if there was some previous error that these edits resolved. I’ve commented on the first changeset, and asked Alexsandar to join the discussion.

5

Hi all. This was not an organized effort of Microsoft editing team. We use a tool to track broken coastlines and this was an effort to fix all reported issues. For few days there was an efffort from some other mapper who was removing coastline from all river maped polygons, so what I did is go and finish this in order to have all coastlines connected.

Since this is large amount of data I had to create a lot of changesets.

Since these fixes turned out to cause more damage I can revert everything to a previous state, but just give me a day since I am not near pc at the moment.

1 Like
6

I have just a question about this one Relation: ‪Chester River‬ (‪1312625‬) | OpenStreetMap how would you like it to be mapped? Since this is a river, it should not have coastline tag on its banks. Coastline in this case is not as wiki describes it but I think there is no other way to tag it correctly. Any thoughts?

7

Most of that is definitely not a river any more, but the sea. Generally a river stops being a river if it is tidal, i.e. if the water flows only in one direction, it is a river, if it goes up and down with the tides, it is the sea. I don’t know where that boundary is in this case, but I would be very surprised if it is where it is mapped in 1312625.

8

While this is somewhat true, there is not a bright line between a river and a sea. There is a transition zone that is sometimes short and other times very long. A tidal river is not the sea, but also doesn’t flow in only one direction. Putting the entire tidal section of a river on the ocean side of the coastline would lead to comical results in many cases, as it would have to go hundreds of kilometers inland, far past the point where the salt water ends.

Whatever portion of the Chester River is on the ocean side of the coastline (the estuary part), would no longer need this natural=water polygon around it. The polygon actually shouldn’t have a name in the first place, as river names go on the waterway=river way, not the water area polygon. The Chester River’s way can remain on the ocean side of the coastline.

4 Likes
9

Just to add some context here — I’m partially responsible for the Chesapeake Bay multipolygon in question. Eric and I created it a couple years ago with the intent of improving render times for updates outside of the coastline boundary.

In hindsight, and especially after reading through this thread, I realize there was a flaw in that assumption. Andy eventually rolled back all changes to the coastline, though he did preserve the Chesapeake Bay relation itself.

For reference, here’s the original tagging discussion from when we made the change: [Tagging] coastline v. water

10

To put it simply, would you say that any tidal body of water downstream of the fall line is considered to be within the coastline—at least in OSM terms?

1 Like
11

That’s a clear rule, but it’s not always how things have been mapped. I’m not convinced that it makes geographic sense to treat all tidal rivers as the sea either. The River Thames, for example, is currently mostly tagged as water, with the coastline starting at an apparently arbitrary point: Way: 1021454152 | OpenStreetMap. The Thames is tidal through London up to Teddington Lock, but it’s evidently a tidal river, not the sea. Similarly, Trenton, New Jersey is at the fall line for the Delaware River, but is definitely not on the Atlantic Ocean. The water area for the Delaware River currrently starts at an arbitrary point near Philadelphia International Airport: Way: 1319483389 | OpenStreetMap. The St. Lawrence River is tidal past Quebec City to Lac Saint-Pierre, but coastline tagging ends at the start of the middle estuary just downstream of Île d’Orléans Way: 182042433 | OpenStreetMap.
While the first two examples are slightly upstream of the described estuaries for those rivers, it seems clear (at least as clear as you can get by picking arbitrary examples, though these were the first three I checked) that the coastline isn’t currently expected to include tidal rivers.

Of course, I can’t talk about coastlines without mentioning Great Bay, which is a brackish tidal bay in my mapping area that is connected to the sea by two tidal rivers. The coastline is currently mapped to the mouth of the harbor, which is probably wrong but not wrong enough for me to mess with it.

2 Likes
12

My classification using the tides is a bit too simplistic, I agree. So is one solely based on fresh vs. salt water which also has been used. Like with so many things in OSM I believe it comes down to common sense. Does it feel more like a river or more like the open sea? Another hint is what people call it. If you’d ask anybody in the middle of London what this water thingy is, they’ll probably say it is the “River Thames”. If they call that water thingy the “Chesapeake Bay”, it is probably open water. If they call it “Thames Estuary” you still don’t know. If there is a clear place where the river widens drastically to form an estuary, maybe that’s a good point. If its more gradual, it is more difficult. I would also argue to put it somewhere where it feels natural from the topography. Let’s think of it this way: If the river would completely dry up suddenly, where would the coastline be then? Maybe that’s a good place.

There is no perfect answer here and we should allow (ideally local) mappers some leway in deciding where that spot is. In the end it doesn’t matter much where it is exactly. What matters is that it is not at some arbitrary line somewhere in the middle of a bay or so just because it is convenient to map it there.

3 Likes
13

Just to add that tidal=yes +natural=water should be used for tidal river areas. Renderers can then treat ocean water and tidal water is the same way, and so the question of where exactly to draw the line becomes less significant.

Part of the problem is that mappers are tempted to move coastline below the high water mark to “get more stuff rendered”. This problem should be reduced if and when this PR to improve rendering on inter-tidal zones gets through the mysterious Carto PR process.

1 Like
14

I have not reverted any changes so far since this is ongoing discussion. The main problem, from my point of view, is that any definition of the coastline can be problematic to map. If this is highest/lowest point of water level, salt of fresh water, etc… For example, Amazon as a big river with lot of strength pushes a lot of fresh water back in the sea area but has been mapped with some approximate way as coastline. In some cases, coastline is not always where see/ocean meets the land.

Having river polygons not overlapping with sea/ocean as it is now mapped is also valid and it is ok with rendering.

However what ever majority decides is what we should apply at the end.

15

Please, let’s not overcomplicate this. Big open water areas connected to the ocean are outside the coastline. Once this narrows down to some reasonable point, draw a line across and call it the mouth of the river. No, you don’t need an exact technical definition, we can use some common sense here.

There is broad consensus that huge open water areas like the Chesapeake and her adjoining bays are outside the coastline. The work to move the coastline inland can commence immediately.

4 Likes
16

It seems like just reverting back to the position the coastline was in prior to May 8th would be appropriate. You mentioned previous changes that broke the coastline, so naturally those would need to be reverted as well.

Can you link to these changesets?

1 Like
17

I have just finished reverting it to previous state. Currently there is no broken coastlines. I will recheck in the morning and compare to the state before May 8th to see if nothing is missed. Feel free to investigate, maybe there is something that I might overlook because of the complexity of the bay

4 Likes
18

Looks good enough that I could reset the coastline processing. There are still some large water areas that I am pretty sure are wrong, see in this area. But they were probably from before that change, so need a separate fixing effort.