This is a static archive of our old OpenStreetMap Help Site. Please post any new questions and answers at community.osm.org.

Coastline polygons-in-polygons, how to interpret?

0

There is a large number of polygons in polygons in the coastline data. The topology logic suggests that these should be interpreted as outer borders of lakes. Unfortunately, this is just partly correct. So, the question is - do you have a well-tested criterion to check whether these polygons are missing lakes, missing islands or just erroneous tagging cases.
If needed I can provide large number of illustrative cases/examples.

This question is marked "community wiki".

asked 19 Jun '16, 10:14

sanser's gravatar image

sanser
695383955
accept rate: 5%

1

It would be very helpful to show at least one example.

(19 Jun '16, 11:10) scai ♦

I second scai. Please link a region where the kind of data appears you talk about.

(19 Jun '16, 12:44) malenki
1

There are many 100s polygons in polygons, all in the source coastline data. Examples: 1. missing lake here http://osm.org/go/euGUE19fV- ; 2. land over land here http://osm.org/go/4lYpy4t ; 3. some more missing lakes here http://osm.org/go/S_yZx64 ; 4. missing (many) islands here http://osm.org/go/ud5K_gk-- (besides, riverbank and lake overlap)and so on. Just overlap the coastline data over the areas in links.

(19 Jun '16, 20:23) sanser

One Answer:

2

Regarding your examples:

  1. has a beach mapped additionally to the coastline, nothing amiss.
  2. coastline with reef and boundaries on them: can't see an error.
  3. the "missing lakes" are mapped with coastlines going the wrong way.
  4. is a messy combination of waterway=riverbank, natural=coastline and a multipolygon relation. I am not astonished that the renderer obviously is balky there.

On the wiki you can read for natural=coastline: land is on the left side and water on the right side
IMHO an easy-to-memorize-criterion.

answered 19 Jun '16, 20:51

malenki's gravatar image

malenki
4.7k24683
accept rate: 6%

1

Further to (1) someone's imported a bunch of OS Vector Map District and made it all coastline, when part of it (https://www.openstreetmap.org/way/255902449 ) should be a small lake, not coastline at all.

Like any import, it just needs properly checking afterwards.

(19 Jun '16, 21:57) SomeoneElse ♦

Thanks for the (quick) answer. My question was pretty explicit and it was not meant to be provocation or irritation. I meat the many, many hundreds of coastline polygons-in-polygons issues when I try to resolve missing objects in other area classes (rivers, lakes, forests ...). The problem is that these polygons are in the OSM source data but confuse the renderers and researchers. The 1-4 example analyses in the answer even add more confusion. Just look at SomeoneElse's confrontation to your #1 text (thanks SomeoneElse) or just click over the layers in the link #2, and so on.

(20 Jun '16, 08:56) sanser

I didn't think I was "confronting" anything....

(20 Jun '16, 08:59) SomeoneElse ♦

@sanser: I didn't feel provoked by you but seemingly still fail to get your point.
By looking at the existing data, comparing it to aerial imagery and – if needed – looking at the mapping suggestions there is not much left to create confusion. IMHO

(20 Jun '16, 15:12) malenki

Source code available on GitHub .