[R-sig-Geo] Kernel density map of severe weather events

[Rolf Turner]

On 05/08/14 02:02, Adrian Antonescu wrote:
> Hi all,
>
> I am trying to create a kernel density estimation map of points
> representing locations of severe weather events (attached script). The
> map and the event (point) data are imported from .shp files with LongLat
> projection. First the map and the point data were converted to a Albers
> Equal-Area Conic projection and then the kernel density was calculated
> using: density.ppp(ppp.regular, kernel = "gaussian",
> sigma=bw.diggle(ppp.regular)). The resulting map looks OK, but I have
> one question: what are the units of the density plot? Since I am using
> Albers Equal-Area Conic with units in meter, the resulting density
> should be events per square meters, but I am not sure if this is right.

The units will be events per square "unit" where "unit" is whatever the 
unit designation is for the pattern to which density.ppp() is being 
applied.  So if the units of "ppp.regular" are metres, then density 
estimate produced by density.ppp() will be in events (points) per square 
metre.

The only assumption made by density.ppp() is that the distance metric 
for the points under study is Euclidean.

cheers,

Rolf Turner

P. S. You can save yourself a few key strokes by typing

     density(ppp.regular,sigma=bw.diggle)

* density() dispatches to (the method) density.ppp()
* sigma can be a *function* (such as bw.diggle)
* the Gaussian kernel is the default

R. T.

-- 
Rolf Turner
Technical Editor ANZJS