[R-sig-Geo] Marginless plot output for georegistration of output graphics

[Roger Bivand]

On Wed, 17 Jan 2007, David Forrest wrote:

> On Wed, 17 Jan 2007, Roger Bivand wrote:
> ...
> >> What do I need to to do turn the key off for spplot(SpatialPointsData) ?
> >
> > I can't see it, I thought it might be auto.ket=FALSE, but it doesn't seem
> > to be that. To be honest, I would feel more comfortable with base
> > graphics and setting my own class intervals and colours. Lattice graphics
> > seem more worth the trouble when conditioning, which isn't the case here.
> 
> Hmm.  I was thinking that the spplot() was the best-practices method for 
> spatial data.

Yes and no. It is certainly best when the graphic is conditioned, and 
otherwise has the apparent advantage of automatic legends and class 
intervals. But it is difficult to use incrementally, so what you choose 
does depend on what you need to do.

> 
> > The aspect doesn't seem happy yet, and for NA and projected CRS, you get
> > "iso" back instead of the aspect from mapasp.
> 
> Is this a criticism of spplot() or the resultant png from the previous?
> 

Just a remark that in the GE context, we're dealing with geographical 
coordinates anyway, but in other contexts, mapasp() will return the string 
"iso", which fails in arithmetic operations.

> I've been working on a toGoogleEarth(spDf,zcol,title,filename) function 
> with someone off-list that writes a png and its KML file.  It seems to
> be working pretty well, and I'll see if he thinks its ready for sharing.
> 

The more the merrier! Auto-generating the KML to call the PNG would be 
very helpful, and would add to what we know about KML (there are traces on 
the Japanese R wiki too, but writing shaded polygons to text files is 
burdensome for larger data sets; PNGs are flexible).

Roger

> Dave
> 

-- 
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