Analysis of finnland mp seabounds processing

The generation of the sea polygons with --generate-sea=multipolygon,no-sea-sectors sometimes still produce flooded areas. I have checked the finnland map and found the following things: 1. Some of the inner polygons generated for the seabounds mulitpolygon sometimes are inverted. Usually the inner polygons should denote all natural=land but some point out the sea are. This causes the flooding. I cannot precisely point to lines of code where this happens. It might have to do with self intersecting polygons. I have attached a GPX of one of self intersecting and inverted ones. Open JOSM, import the GPX, convert it to data layer, tag the polygon with natural=land. - Compare it with the real map, you will see that it is inverted. - Run the validator plugin. You will see that it has 27 duplicated nodes (I don't think that this is a problem). And there is an intersection in the east on the osmosis boundary shape. This should not create a real problem for mp. Maybe some parts of the polygon are dropped. But this might point to a problem of the sea multipolygon creation. I could easily add a warning if a way intersects itself. Don't know if that's a performance issue but will test it. 2. Some islands seem to be tagged with natural=land. They are not considered in creating the sea multipolgyon. Example: http://www.openstreetmap.org/browse/way/48666101 Hope these two things help to reduce the flooding. WanMil

Hi WanMil,
The generation of the sea polygons with --generate-sea=multipolygon,no-sea-sectors sometimes still produce flooded areas.
What are the sea sectors good for, and why are you disabling them?
1. Some of the inner polygons generated for the seabounds mulitpolygon sometimes are inverted. Usually the inner polygons should denote all natural=land but some point out the sea are. This causes the flooding.
Am I right assuming that mkgmap is generating multipolygons for sea regions? I downloaded some of the map area covered by your GPX file, and not all coastlines of islands were part of a multipolygon relation.
2. Some islands seem to be tagged with natural=land. They are not considered in creating the sea multipolgyon. Example: http://www.openstreetmap.org/browse/way/48666101
Could mkgmap ignore the natural=land when generating the multipolygon?
Hope these two things help to reduce the flooding.
How should we proceed? I can devote some time to fixing broken multipolygons if mkgmap gives me good diagnostic messages. I have already fixed all the roundabouts and turn restrictions that mkgmap complained about in Finland, over the course of some weeks. Marko
participants (2)
-
Marko Mäkelä
-
WanMil