[Tagging] Way inside riverbank

Christoph Hormann chris_hormann at gmx.de
Tue Apr 14 10:52:36 UTC 2015

On Monday 13 April 2015, Torstein Ingebrigtsen Bø wrote:
> Hi,
> I'm currently importing topological data of Norway to OSM. From the
> data set we have riverbanks; however, we do not have the deepest
> middle way as required by the wiki [1]. This middle line is therefore
> drawn manually. This is a time consuming (and dull) job. For one
> municipal it takes around 5-10 hours to draw all these lines, in
> Norway we have 428 municipals. Drawing all these middle lines will
> slow down the time to import everything dramatical. I am therefore
> curious of what is the benefits of this line. Is it really necessary
> or is it a "nice to have"?

It is the other way round - the riverbank polygon is optional and 'nice 
to have'.  The waterway line is what actually defines a river in OSM, 
it also gets the name tag and other attributes.  The primary reason for 
this is to map the water flow structure.  Also it is much easier to 
verify and fix structural errors in waterway line mapping than for 
water polygons.

Generating a waterway line when you only have polygons is fairly simple 
via straight skeletons (http://en.wikipedia.org/wiki/Straight_skeleton)  
If the source data set does not contain continuous line features for 
rivers generating those should be part of import preparation.  And i 
disagree with Janko here - if the source data does not contain certain 
information that is required according to OSM mapping conventions you 
should not import the data unless you produce the information in some 
way (either through manual mapping, computing the missing data or 
getting it from other sources).  Otherwise the data becomes dead mass 
in the OSM database.

Christoph Hormann

More information about the Tagging mailing list