On Tuesday 12 March 2013, Jochen Topf wrote:
>
> OSMCoastline first generates the land polygons, them splits them,
> then creates the water polygons as inverse from the lang polygons. If
> the land polygons are broken, so are the water polygons. So I don't
> think you can reduce the chance of breakage that way.

Possibly a better way to prevent large errors in the coastline would be 
to maintain a simplified version known to be free of large errors and 
use it to create bounds for the real up-to-date coastline.  As you said 
the quality of the OSM coastline is quite good these days globally so 
at least on the land side it can be said with very high confidence that 
a few kilometers from the OSM coastline there is only land.  So using a 
correspondingly buffered old and verified land polygon in addition to 
the real one would prevent large errors either by accident or by 
vandalism and would not hamper real improvements of the data.

I don't know how often you discard the new data based on comparison with 
previous day data but the above would be an alternative to that.

Greetings,

-- 
Christoph Hormann
http://www.imagico.de/

_______________________________________________
dev mailing list
dev@openstreetmap.org
http://lists.openstreetmap.org/listinfo/dev

Reply via email to