[OSM-dev] coastline polygons

[Martijn van Oosterhout]

On 2/7/07, Frederik Ramm <frederik at remote.org> wrote:
> > The main reason for storing data separately is to make use of spatial analisys
> > tools (e.g JUMP). For the south-east of the UK I used original Shoreline data
> > and 'almien' coastlines from OSM planet. I found coastlines in OSM to have
> > lots of 'unclean' (e.g. self-intersecting)  geometries, probably because they
> > went through some line simplification algorithms I don't know...
>
> As far as the coastline data in OSM goes, I also found deficiencies with
> some of the Scottish sea lochs (they didn't go as far inlands as they
> ought to), and corrected them by hand in JOSM based on Yahoo imagery.
> That's why I was keen on the ability to do manual corrections.

While the idea of being able to manage the entire coastline in OSM
sounds attractive, I think it's really an awful lot of data for very
little benefit. There are thousands of kilometres of coastline where
none s going to care about the exact placement. I would prefer as
system where the coastline were stored on a seperate layer/server.

In your case I would have left the coastline as is and added closed
polygons for the inland bits that wern't included.

One main problem with storing it in OSM is that renderers can't do
anything useful with it, since you can't tell the inside from the
outside.

Have a nice day,
-- 
Martijn van Oosterhout <kleptog at gmail.com> http://svana.org/kleptog/