[mkgmap-dev] [PATCH v5] sea polygons

Wed Aug 26 09:37:12 BST 2009

Dear Chris,

I also get similar problems for some tiles (the result may depend on the 
splitter parameters).
The problem is that there are several shoreline segments that do not end 
at the map boundary. Two examples are:
- Rügen: The shoreline ends at the border to Poland. The current patch 
closes this segment since it's "nearly closed" (the distance between 
start and end is less than 0.1 * segment-length). This should result in 
the Polish part of the island (Swinoujscie) beeing flooded. This seems 
to be the best the algorithm can do since he doesn't know the Polish 
part of the shoreline cut off by osmosis.
- At the border to the Netherlands, the shoreline is also cut off. 
Currently the algorithm tries to create sea polygons for this shoreline 
segments by adding a "corner point" on the boundary of the map. However, 
this seems not to produce good results in all cases.
- I have noticed problems due to partly missing shoreline segments for 
example at the Elbe. This, however, seems to be a problem in the 
splitting of the data.

If anyone has a good idea how to handle an incomplete shoreline, please 
let me know. Right know, it might be best to do the following:
- get a country extract like germany.osm.bz2 from geofabrik,
- calculate the boundary box,
- use xapi or same other source to get the complete shoreline in this 
boundary box,
- mix in the shoreline using osmosis,
- use the result as input for splitter and mkgmap.

I have not tested this yet, but it should give better results.

Best wishes
Christian

Chris-Hein Lunkhusen schrieb:
> Robert Joop schrieb:
>   
>> christian, thank you very much for working on this!
>>     
>
> Yes, great to see that the sea is going blue.
>
>   
>> here are some current observations with my big bbox cut from the planet:
>>     
>
> For germany (geofabrik extract) :
>
> Inner Land tiles w/o coast are ok.
>
> The northern sea area is "inverted", so islands are blue.
>
> The algorithms don't seems to get the "water is on
> the right side" rule right.
>
> Chris
>
> _______________________________________________
> mkgmap-dev mailing list
> mkgmap-dev at lists.mkgmap.org.uk
> http://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
>   