[Imports] Administrative Boundaries and Divisions in Portugal

Christoph Hormann chris_hormann at gmx.de
Mon Apr 13 17:41:02 UTC 2015

On Monday 13 April 2015, Petr Morávek [Xificurk] wrote:
> >
> > when the river is the boundary you will want to move both together
> > and you want them to be linked together to determine what is in
> > which entity (eg the river, the road), and to keep the fact that
> > they are linked
> this is not quite true. In Czech Republic I know a lot of places
> where historically the boundary was defined by a small stream or
> river. The course of the stream or river has changed, but the
> administrative boundary stayed at the old place.
> AFAIK even parts of country boundary that are defined by
> international treaties as going by the centerline of a river are in
> fact fixed at a given place and periodically reviewed and
> renegotiated to reflect possible environmental changes.

Since the myth that all or nearly all political boundaries are defined 
through coordinates is frequently brought up in OSM context a few 
things to keep in mind:

There are three common types of boundary definitions: 

* definition through point coordinates and geodetic lines between 
* definition through physical geographical features like rivers or 
* definition through a mathematical formular based on either coordinates 
or physical features - this is primarily for maritime boundaries.

I made a few random tests some time ago for various international 
borders based on information available on the net and the (possibly not 
very representative) results were that most borders that roughly follow 
a river are actually defined through the river itself (i.e. not through 
a list of coordinates).  In case of ridges the situation is not quite 
as clear - there are probably at least as many international borders 
following ridges that are defined through peaks and other locations 
with straight lines in between as there are such that are defined as 
the ridge line itself (i.e. watershed) without coordinates.  This 
probably has to do with the fact that positions on a ridge are better 
to survey than positions in a river.

Needless to say that even if a certain border following a river is 
formally defined via coordinates the de factor line of control (which 
is what is mapped in OSM) is usually the river.

And the fact that an authoritive data set contains line strings or 
polygons cannot be used as an indication this is the actual definition 
of the border of course.  Most boundaries in OSM have at some point 
been imported and many from authoritive sources.  Still many are 
clearly wrong.  When you see a border roughly but not exactly following 
a river in OSM this is usually an indicator for inaccurate data, not 
for a strange border definition.  Famous and blatant example:


If the data set you are planning to import contains information on the 
actual definition of the boundaries it would be great to maintain this 
in OSM.  There are currently very few borders in the database with such 
information, AFAIK all of them maritime borders, like:


Christoph Hormann

More information about the Imports mailing list