Analysing spatial point patterns in R - CSIRO
Analysing spatial point patterns in R - CSIRO
Analysing spatial point patterns in R - CSIRO
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
149<br />
In data sharpen<strong>in</strong>g [27] the <strong>po<strong>in</strong>t</strong>s effectively exert a force of attraction on each other, and<br />
are allowed to move <strong>in</strong> the direction of the resultant force. This tends to enhance tight l<strong>in</strong>ear<br />
concentrations of <strong>po<strong>in</strong>t</strong>s.<br />
> Y plot(Y, pch = ".", ma<strong>in</strong> = "data sharpened")<br />
data sharpened<br />
25 Adjust<strong>in</strong>g for <strong>in</strong>homogeneity<br />
If a <strong>po<strong>in</strong>t</strong> pattern is known or suspected to be <strong>spatial</strong>ly <strong>in</strong>homogeneous, then our statistical<br />
analysis of the pattern should take account of this <strong>in</strong>homogeneity.<br />
25.1 Inhomogeneous K function<br />
There is a modification of the K function that applies to <strong>in</strong>homogeneous processes [7]. If λ(u)<br />
is the true <strong>in</strong>tensity function of the <strong>po<strong>in</strong>t</strong> process X, then the idea is that each <strong>po<strong>in</strong>t</strong> x i will be<br />
weighted by w i = 1/λ(x i ).<br />
The <strong>in</strong>homogeneous K-function is def<strong>in</strong>ed as<br />
⎡<br />
K <strong>in</strong>hom (r) = E ⎣ ∑<br />
⎤<br />
1<br />
λ(x j ) 1 {0 < ||u − x j|| ≤ r}<br />
∣ u ∈ X ⎦ (28)<br />
x j ∈X<br />
assum<strong>in</strong>g that this does not depend on location u. Thus, λ(u)K(r) is the expected total ‘weight’<br />
of all random <strong>po<strong>in</strong>t</strong>s with<strong>in</strong> a distance r of the <strong>po<strong>in</strong>t</strong> u, where the ‘weight’ of a <strong>po<strong>in</strong>t</strong> x i is 1/λ(x i ).<br />
If the process is actually homogeneous, then λ(u) is constant and K <strong>in</strong>hom (r) reduces to the<br />
usual K function (21).<br />
It turns out that, for an <strong>in</strong>homogeneous Poisson process with <strong>in</strong>tensity function λ(u), the<br />
<strong>in</strong>homogeneous K function is<br />
K <strong>in</strong>hom, pois (r) = πr 2 (29)<br />
exactly as for the homogeneous case.<br />
The standard estimators of K can be extended to the <strong>in</strong>homogeneous K function:<br />
̂K <strong>in</strong>hom (r) = 1 D<br />
∑∑<br />
i<br />
j≠i<br />
1 {||x i − x j || ≤ r}<br />
e(x i ,x j ;r) (30)<br />
̂λ(x i )̂λ(x j )<br />
where e(u,v,r) is an edge correction weight as before, and ̂λ(u) is an estimate of the <strong>in</strong>tensity<br />
function λ(u).<br />
Copyright<strong>CSIRO</strong> 2010