[R-sig-Geo] About map projection of world map

Barry Rowlingson B.Rowlingson at lancaster.ac.uk
Thu Nov 20 15:49:48 CET 2003


>  I've tried to use map projection of world map shape file(which has latitude
> & longitude coordinates) in many setting( +proj=merc +lon_0=0 etc.).
> 
>  However I couldn't get good results of projections of world map, all points
> on edges of map extent of the world map were linked and broke maps in map
> drawing.
> 

 Yes, this is a known problem with Rmap and projections, and nothing I
can do about it at the moment.

 Because the map data is just a vector with two end-points, for example:
 (179E, 10N) to (179W,10N), there's no way of knowing if that line is
the short line that goes over the 180E line or goes the other way around
the world. This is because there are two straight lines between any two
points on a sphere.

 You could take a heuristic approach, and draw the shortest of the two
lines, and this would work for world country data. The plotting code
would still then have to split the line into two parts, broken at the
edge of the mercator projection. But this might not be what you want if
you are mapping arbitrary vectors on a sphere.

 What we really need is a way to disambiguate the geometry. One possible
way is to specify an origin point and a distance and direction vector on
the sphere. This then uniquely identifies the destination point and the
set of locations on the line. However, most geo data sets are stuck in
the origin coord-destination coord paradigm...

 I've only ever wanted Rmap to draw maps of small areas, up to continent
size, and so I've not been motivated to fix the meridian problem. Sorry!

Baz




More information about the R-sig-Geo mailing list