From OpenStreetMap Wiki
Jump to navigation Jump to search

Closed Polygons

It says "Mapnik really needs closed polygons", but is that realistic? I mean are we really going to try to draw a single closed way around the entire coastline of Asia for example?

I thought it was widely agreed that we would not do that. A single way made up of many thousands, probably millions, of nodes, will cause problems for the various softwares, and for the OSM Protocol interface itself. I remember a discussion related to long motorways in the US, where it was agreed that they should actually be chopped into smaller ways if they get beyond a certain length, just to make the data easier to swallow.

Also it seems a rather fragile requirement. Would the Mapnik coastline rendering of the Netherlands suddenly break, if somebody disconnected the polygon over in India?

The tiles@home approach Tiles@home/Dev/Interim Coastline Support is a bit hacky (which I guess is why it's described as an interim solution) ...but it works, without the need for closed polygons.

-- Harry Wood 16:04, 27 November 2007 (UTC)

There doesn't need to be a single way in the OSM database in order to derive a complete polygon. As long as the adjacent ways share a common node then the shapefile can be created from the adjacent ways. As an example see the coastline of Hawaii, which exists as many ways in the OSM database, but from which a shapefile has been created for the Mapnik coastline. Dmgroom 19:09, 27 November 2007 (UTC)
Aha. Yeah it ocurred to me afterwards that might be what was meant. So a "closed polygon" doesn't mean a single closed way.
There's still a problem of it being fragile. And if Mapnik derives a polygon, does that mean it's loading in all of the ways? e.g. encircling the whole of Asia. It's a lot of data to shunt around, just to render the blue on the right side of a line. -- Harry Wood 15:06, 28 November 2007 (UTC)

Note on Coastline and Yahoo imagery

The Coastline of Conakry (Guinea) is interesting 'cause they have high differences in tides (see It can be up to 3 meters - so the yahoo image was taken on low tide. --katpatuka 08:35, 8 February 2008 (UTC)

Failure to render

Sometimes I have modified the coastline and added some land features as well, and while the land features show up, the coastline is updated at far zoom levels but remains the same at near zoom. What is happening here? CrystalWalrein 13:04, 5 March 2009 (UTC)

  • can you give an example Dmgroom 13:38, 6 March 2009 (UTC)
Here is an example. Osmarender has already accepted the coastline, but Mapnik does not, even when Metacarta Up-To-Date is used. CrystalWalrein 23:14, 6 March 2009 (UTC)
  • I've got the same problem after editing St-Laurence river (Montréal - canada) with potlatch/yahoo!. Osmarender accepted the coastline I had edited but Mapnik does not. Example. Something new about editing coastline and get the proper picture ? Daniel 04:28, 9 june 2009 (UTC)
  • Here is an other example, even the coastline checker does not work propperly. Any explanation for this? --Noframe 14:10, 13 June 2009 (UTC)
  • The coastline error checker is currently experiencing issues with the server. . Dmgroom 11:18, 16 June 2009 (UTC)
  • Thanks for your Information! --Noframe 14:33, 16 June 2009 (UTC)

So is the renderer actively buggy, or just slow? Eg. [1] here I did some major changes to the coastline of Bintan Island, and while nearby roads etc already show up on the rendered map, the coastline is still the incorrect old version. Jpatokal 13:22, 25 September 2009 (UTC)

Just slow I think. It says on the page, the shapefiles are generated "every few weeks" -- Harry Wood 14:12, 25 September 2009 (UTC)


The German page de:Coastline is much more specific. Please translate it to English. Thanks, --Markus 14:05, 5 September 2009 (UTC)

Does anyone think this is still necessary? If not, should we say it's closed? (Yes, I know this isn't a ticket tracking system...) Jeffmeyer 20:05, 6 December 2012 (UTC)

Tagging Low water

Are there any established methods of tagging the low water? PeterIto 22:24, 17 March 2012 (UTC)

There is this proposal Proposed_features/Water_cover, which is not a direct answer to your question, but obviously one edge of any tidal area would be the low tide mark. Dmgroom 19:27, 18 March 2012 (UTC)
  • Thanks, however that does seem to be stuck and to an extent taken over by wetland. On reflection I guess wetland does everything that is needed, and is already being used in some places. All it needs is some more clarity on how to tag sand, or 'beach' that gets covered at high tide and one is done. I will look more carefully at wetland. PeterIto 21:50, 18 March 2012 (UTC)
  • Do we need to propose a new clear tag for low water? How about natural=low_water_spring. One can then tag the area between high and low water with natural=beach, natural=wetland or whatever with an additional tidal=* to indicate that it isn't always dry. Would that make sense? PeterIto 09:25, 7 June 2012 (BST)
  • The coastline seems to be defined as the 'mean high water'. That will cause that most of the coastline is accompanied with a 'wetland:tidal_flat' on the seaside. Practically everywhere where the coast is not either very steep or artificial. The land-side of that area should be the coastline. Under this is is probably not very reasonable to use the coastline for the border as done for most of Great Britain. --Humbach 12:47, 2 October 2012 (BST)

Proposed Merge of Page

I am proposing to merge this page with natural=coastline which contains essentially the same information. At present mappers need to look at both pages to get all the detials of how to map coastline, and how it renders.Dmgroom 12:49, 7 May 2012 (BST)

Yeah the two pages have been around for a long time. Maybe the separation originally was to do with coastline rendering details (and all the quirks) going on the "Coastline" page. Elsewhere on the wiki we do have this idea of "Feature" pages vs tag documentation. e.g. we have Railways in addition to Key:railway. See also Wiki organisation. Feature pages are a sort of general guide mentioning many tags, while tag documentation is supposed to give more brief structured info. So we could change it to better follow that kind of idea, but yes at the moment Coastline and Tag:natural=coastline have ended up having very similar info. -- Harry Wood 04:23, 8 May 2012 (BST)
Point taken, maybe just a bit more of a tidy up of the two pages is all that is needed. Dmgroom 11:38, 8 May 2012 (BST)
  • Thanks Harry/ Dmgroom. I have just given the coastline tag article a bit of a cleanup into simpler headings prompted by the above. Please make sure I have got it right, but I would suggest that we then add any more details about coastline to this article. The Whole subject of beaches, mud flats etc need some more work. Also how to map low-water. Could I also alert people to a discussion taking place on talk:Harbour which relates in a way to the coastline and where some additional views and input could be helpful. PeterIto 12:10, 8 May 2012 (BST)
  • I have now removed the merge banner. Please put if back if you still think it is serving a useful purpose. PeterIto 11:43, 31 May 2012 (BST)
  • I have further differentiated the pages by moving nearly all the detail about mean high water to the natural=coastline article, allowing this article to now be developed with more information about the more general subject. PeterIto 08:48, 7 June 2012 (BST)
On this note, should the information about shoreline300 and processed_p be on this page? For example, if I'm interested in downloading a file of the outlines of the various continents, is that really a tag-related topic or a more general feature topic? Jeffmeyer 20:03, 6 December 2012 (UTC)
I think the distinction between "tag-related" and "feature-related" doesn't make sense for a "feature" that corresponds exactly to one tag or key. Someone who looks for tagging coastlines will likely land on Coastline, find the tags, and go off tagging without reading any of the highly relevant information from the tag site, e.g. regarding the directionality. I'm still supporting a merge. --Tordanik 12:21, 9 April 2014 (UTC)


it's not too clear to me what in this sentence the word "including" means: “The coastline is a rich and complex environment at the edge of the marine environment including cliffs, beaches, harbours, headlands, bays, wetlands and islands.”

I have seen examples of harbours being included in the land (Mergellina harbour in Napoli, now repaired, made the coastline look extremely weird to anybody who knows the city), other harbours being included in the "marine environment", but most often the coastline following all breakwaters and possibly even the piers.

if it was me deciding, I would say the coastline follows the natural coast and leaves all man_made marine items outside, thus built on the sea. so I would like interpreting the above "including" as referring to “the marine environment, which includes [cliffs,] beaches, harbours, headlands, bays, wetlands [and islands].” But the presence of cliffs and islands in the list make me seriously doubt of this interpretation. If the "including" referred to what the land should include, then my impression is that beaches, harbours and bays do not belong to the list.

As I understand it, that text has nothing at all to do with coastline tagging and is just "flavour text" written by some user to avoid having a page with no introduction. And imo adding that text was a pretty bad idea. --Tordanik 13:24, 8 May 2014 (UTC)

LAT - Lowest Astronomical Tide

Additionally to the upper coastline, we need LAT as lower coastline. Between we have in tidal areas the dry falling areaː wetland, beach, stones, rock...

How can we tag this line? --Markus (talk) 15:45, 9 July 2015 (UTC)

Absolutely agree, this low tide line is fundamental. I'm actually building multipolygons beyond the coastline indicating tidal=yes, or multipolygons like natural=beach or natural=bare_rock just as when I map the area of a river, with a way above the coastline indicating the start of the element (beach, bare rock, etc), and a way below indicating the Lowest Astronomical Tide, joined by two ways either side of the section of that multipolygon. The coastline usualy goes in the middle.
illustration of the several involved ways in river mapping

Here's an example: --Iagocasabiell (talk) 23:31, 17 June 2019 (UTC)

In this example you have mapped a bare_rock but not the LAT. In theory and practice the bare_rock can extend well beyond the LAT and it would be also nice to have it mapped. I assume tidal=yes would "fix" your example though something more general could be perhaps also used for underwater rocks. I would probably map an area tidal=yes independently overlapping the bare_rock which can extend from well above to well under water and will be "divided" by natural=coastline and tidal=yes area.RicoZ (talk) 20:35, 20 June 2019 (UTC)