[R-sig-Geo] circular window for spatial statistics (spatstat package)

[Roger Bivand]

On Tue, 13 Mar 2007, Michael Friendly wrote:

> Hi list,
> 
> A biology student I'm trying to help has done a study planting 20 seeds 
> in circular pots under varying conditions and wants to analyze the 
> degree of clustering
> (and other measures) in the seeds that germinate (0-20) in relation to
> the factors that differentiate pots (treatment, type of seed, ...)
> 
> The data structure is something like a data frame for all the 
> germinating seeds, recording (X,Y) location and maybe some other
> measure like size after xx days:
> 
> PotID Treatment SeedType  X    Y  size
> 
> Assume that the (X,Y) locations are referred to a circle of fixed radius, R.
> I'm not very familiar with analysis of spatial point processes, but it 
> seems that the spatstat package could handle this.  (Is there something
> better?)  I've read the documentation, but can't find an example of how 
> to specify a circular window for the ppp object that would contain the
> data and the window.

Maybe March is a circular month? There was a thread on this recently:

http://article.gmane.org/gmane.comp.lang.r.geo/1320

refering to John Fox's original:

http://finzi.psych.upenn.edu/R/Rhelp02a/archive/74907.html

For a single pot, you'd need an owin object in the coordinate system of 
the X,Y, best if the pot centre and radius were recorded. Then the 
locations of the germinated seeds in each pot would become a ppp object 
with a known owin (generated as a circumference sequence). Should the 
clustering measure be at the seed observation or the pot plot level?

Hope this helps,

Roger 

> 
> thanks,
> -Michael
> 
> 

-- 
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