[mkgmap-dev] Sea 'bleeding' into land in northern France

Fri Aug 5 18:02:10 BST 2022

Hi John

Adding to what Thorsten has written:

It can be impossible to reliably deduce what should be sea when just
using the coastline data from within an OSM area extract. The extract
area is expanded to be within a set of rectangular tiles and, as a
result of this, coastline data might just stop within a tile; coastline
is defined as having sea on one side and land on the other so having an
"end" renders this ambiguous.

mkgmap has various --generate-sea options to control what happens in
this situation, but all can fail given particular circumstances.

For this reason it is common for mkgmap map developers to use a world
representation for sea and mkgmap has the option to use this rather
than the extract coastline data (--precomp-sea=sea.zip) - see:

https://www.mkgmap.org.uk/doc/options
 and
https://www.mkgmap.org.uk/download/mkgmap.html

Ticker

On Fri, 2022-08-05 at 16:45 +0200, Thorsten Kukuk wrote:
> Hi,
> 
> On Fri, Aug 05, john510185 wrote:
> 
> > Hi, I hope someone can help with this. I'm a big fan of
> > OpenMapChest and am
> > amazed by Ben Konrath's lovely maps. There's a slight problem (of
> > the first
> > world type) in that in his map of Western Europe (specifically the
> > Calais tile)
> > the sea appears to have breached the sea wall and the land is
> > inundated with
> > blue! Ben says he can't help because it's due to a bad change
> > someone has made
> > to OSM. There's also something funny further west, around Le Havre.
> > He says the
> > problem will be solved by whoever made the change, but it's
> > persisted now for
> > quite a few iterations of his map.
> 
> Yes, the coastlines are very often broken in the OSM database, but
> for
> the last days I couldn't find any serious problem in that area. There
> are the usual wrong directions of islands, currently in the near of
> spain.
> Depending on how he builds his maps and handles the coastline, this
> could be a really big problem or no problem at all.
> 
> So it heavily depends on how he builds his maps.
> 
> > There does appear to be a conflict between his British Isles map
> > and the one
> > for Europe. However, the problem isn't replicated in OsmAnd, or in
> > any of the
> > other Garmin-related maps I use, which would suggest that OSM
> > itself is OK.
> 
> Your assumption is wrong. The coastline is most of the time broken in
> OSM, that's why people building maps have checks or use verified
> coastline data to avoid this flooding. E.g. by using pre-build sea
> data
> from here for garmin maps: https://www.thkukuk.de/osm/data/
> 
>   Thorsten
> 
> > Any wisdom greatly appreciated!
> > 
> > One of the replies I received was as follows:
> > 
> > Those maps are generated with mkgmap, and this sort of problem
> > happens
> > occasionally when coastlines get broken in OSM. You might look at
> > the
> > archives for mkgmap-dev at lists.mkgmap.org.uk or perhaps post there
> > directly.
> > 
> > Any help with this would be much appreciated. I have MapSource
> > screenshots
> > available, but didn't attach them in case I messed anything up!
> > 
> > Cheers!
> > 
> 
> > _______________________________________________
> > mkgmap-dev mailing list
> > mkgmap-dev at lists.mkgmap.org.uk
> > https://www.mkgmap.org.uk/mailman/listinfo/mkgmap-dev
> 
> 