[R-sig-Geo] Handling invalid geometries from the ‘mapdata’ package

[Roger Bivand]

On Mon, 28 Mar 2011, Karl Ove Hufthammer wrote:

> I have discovered what looks like a geometry error in data from ‘worldHires’
> data set. Here’s an illustration:
>
> library(rgeos)
> library(rgdal)o 
> library(mapdata)
> library(maptools)
>
> x.map = map("worldHires", "Norway", fill=TRUE, col="transparent", plot=FALSE, xlim=c(0,40), ylim=c(50,73))
> x.sp = map2SpatialPolygons(x.map, x.map$names, proj4string=CRS("+init=epsg:4326"))
>
> plot(x.sp)            # *Looks* OK
> gPointOnSurface(x.sp) # Error: TopologyException: side location conflict at 3.97911 0.0769463
> gIsValid(x.sp)        # Error: Self-intersection at or near point 27.9364 70.0864
> points(27.9364, 70.0864, col="red", pch=19)
>
> I guess this is to be be expected, as the ‘worldHires’ data isn’t of very
> high quality, and is not guaranteed to be consistent. My questions are 
> 1) is it possible to automatically ‘fix’ this problem inside R, and 
> 2) what other problems may I expect from continuing to use the invalid data?

To start with, it is always advisible to promote the sp SpatialPolygons 
object Polygons components to GEOS compatibility:

pls <- slot(x.sp, "polygons")
pls1 <- lapply(pls, checkPolygonsHoles)
slot(x.sp, "polygons") <- pls1

which adds a comment attribute to each Polygons object assigning hole 
interior rings to their containing exterior rings.

Your problem cannot be solved automatically, but here from zooming in on 
the problem area:

gIsValid(x.sp)
plot(x.sp, xlim=c(27.9, 28), ylim=c(69.9, 70.1))

and eyeballing the bounding boxes of the Polygons objects:

lapply(pls1, bbox)

it looks as though 26 is the problem:

plot(x.sp[26], add=TRUE, border="orange")

x.sp1 <- x.sp[-26]
gIsValid(x.sp1)
gPointOnSurface(x.sp1)

Fresh releases of rgeos and maptools are on their way to CRAN, but source 
may be checked out from R-Forge now (the package builds there are of 
yesterday's version).

Hope this helps,

Roger

>
> -- 
> Regards,
> Karl Ove Hufthammer
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo

-- 
Roger Bivand
Economic Geography Section, Department of Economics, Norwegian School of
Economics and Business Administration, Helleveien 30, N-5045 Bergen,
Norway. voice: +47 55 95 93 55; fax +47 55 95 95 43
e-mail: Roger.Bivand at nhh.no