[OSM-talk] Is it land or sea: how to map a swamp?

Ulf Mehlig ulf.mehlig at gmx.net
Fri Jul 11 15:27:04 BST 2008

There is a "wetland" proposal which includes wetland=mangrove.


In the area where I am working at the moment 


(almost) all the coastline is drawn along the outer border of mangrove
forests, so it is considered as "land". I tagged mangroves at


as natural=wetland tentatively ... I would like to see the wetland tag
voted/approved/rendered -- it could help to map a not-so-small part of
tropical coastlines correctly! 

BTW, there is a mangrove island in this area rendered correctly in
Osmarender but not in Mapnik, in spite of a multipolygon relation which
is correct in my opinion:



On Wed, 2008-07-09 at 22:21 +0100, Mark Williams wrote:
> Hash: SHA1
> natural=marsh?
> (from Map Features)
> I would expect to find coastline on the open-sea border of this.
> Mark
> Mike Collinson wrote:
> > I agree with Stephen's comments and add that I follow the rule "if in doubt, map it as land" since we don't have the luxury of being able to map average high water marks or highest spring tide mark that a government agency might use.  If it is something that I can walk out and see most of the day or year, then I think it should be mapped as land as a navigation aid.
> > 
> > It might also be worth considering a natural=mangrove area tag.  Our current system is biased towards temperate climates.  I've hesitated so far as it is often very difficult, either on the ground or from imaging data, to map the inland extent.
> > 
> > Mike
> > 
> > At 03:27 AM 9/07/2008, Stephen Hope wrote:
> >> The northern coast of Australia has many Mangrove marshes at river
> >> mouths, some of them extending many kilometres away from the dry shore
> >> line.  PGS shows these areas as sea, because they are not dry land -
> >> and that is were the coastlines would have been imported from.  Note
> >> that "being submerged for half the year" doesn't mean the trees are
> >> covered with water, just the mud under them.  The tree tops would be
> >> above water all the time, I suspect.
> >>
> >> We've (mostly) tagged them as land, with the coast being on the sea
> >> side of them.  Technically they may be water covered (or partially
> >> water covered, usually about 6 inches deep), but if you can't swim or
> >> boat in them and plants and trees grow there it's land as far as I'm
> >> concerned.  They certainly are not ocean.  Marshes in the UK are also
> >> treated as land from the coastline point of view, even were they edge
> >> an ocean.
> >>
> >> See http://www.openstreetmap.org/?lat=-16.9642&lon=145.7843&zoom=13&layers=B00FTF
> >> for an example near Cairns.  More examples are further up the coast.
> >>
> >> Stephen
> >>
> >>
> >> 2008/7/9 Alan Millar <am12 at bolis.com>:
> >>> I came across an interesting area which I don't know how to map or tag.
> >>>
> >>> http://www.openstreetmap.org/?lat=22.066&lon=89.047&zoom=9&layers=B00FTF
> >>>
> >>> This is the Sundarbans mangrove forest on the border of India and
> >>> Bangladesh.  The map doesn't look like much, but look at the map with
> >>> aerial photos like in Potlatch edit mode and it starts to get interesting.
> >>>
> >>> I read that it is submerged for up to half of the year.  The Yahoo aerial
> >>> photos clearly show the forest areas, so I assume they were taken at a
> >>> low-water period.  Google Maps shows it as land.
> >>>
> >>> Our oceantiles file has it as land, but our coastlines treat it as sea.
> >>> Our coastlines stop at the farmlands which border it.  During the high
> >>> water period, I suppose our coastlines make sense.
> >>>
> >>> Does anyone have any recommendations of how to treat an area like this?
> >>> Any similar geography already mapped somewhere?  Thanks
> >>>
> >>> - Alan
> >>>
> >>>
> >>>
> >>> ________________
> Version: GnuPG v1.4.6 (GNU/Linux)
> Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
> iD8DBQFIdSv1JfMmcSPNh94RAty+AJ9voJsnb9ym6eiFMB9dNJFaHg5WpACfUWAO
> we9MgNpK8v5miRbnCw+4tU4=
> =tGCg
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> http://lists.openstreetmap.org/cgi-bin/mailman/listinfo/talk
 Ulf Mehlig    <ulf.mehlig at gmx.net>

More information about the talk mailing list