CODEWITHSUNDEEP
spatialaggregationr-packagespatstat
geminis6612
geminis6612

Reputation: 1

Can I use spatstat to measure spatial aggregation of the environment?

Currently I have individual raster data that represent the suitable environment for 18 species based on MAXENT predictions. I would like to know if the suitable environment for each species is either aggregated or not. I know that usually the R package spatsat is used to test aggregation of the spatial point pattern, but It seems that I can't not test that for the environment itself. Is that actually the case? Does any one of you would be able to direct me to a package that I could use to test aggregation of the environment? Thanks in advance!!

TO FOLLOW UP THE QUESTION ABOVE WITH MORE DETAILS AND FIGURES

I have attached two images that I hope it makes my question more clear. So I would like to be able to quantify if the green cells (suitable environment) in figure A are more aggregated that the cells (suitable environment) in figure B. Green cells have a value of 1 and white space around then have a value of zero. I do not want to use the point locations of individuals, since I am not trying to test if the individual points are aggregated. What I was doing is using the X and Y coordinates of each green cell, but if I calculate the Clark Evans it shows is not aggregated for both. I think is because if I used the X and Y coordinates of the green cells for the Clark test all are part of a continue pattern as on figure 3. I hope this extra information is able to offer some help because I think I hit a wall now.

Potential aggregated environment

Potential no aggregated environment

Green cells X and Y coordinates used for Clark test

Upvotes: 0

Views: 91

Answers (1)

Adrian Baddeley
Adrian Baddeley

Reputation: 2973

You can use spatstat to estimate the covariance function of each environment, treating each environment as a random set. Suppose G is a window (class "owin") representing the green cells; R is another window representing the red cells; and W is the containing window in which the environments are observed. To estimate the covariance function of the green cells you could do 

cW <- setcov(W)
pG <- area(G)/area(W)
cG <- setcov(G)/(pG * cW)
cG[cW == 0] <- NA
fG <- rotmean(cG)

Then pG is the coverage fraction and fG is the (isotropic) covariance function. You could now do the same thing for R instead of G, and compare the two plots. Higher values of the covariance suggest a more aggregated environment.

Upvotes: 1

Related Questions