[R-sig-eco] Simulating spatial point patterns that have spatial structure

[Dixon, Philip M [STAT]]

Liudas,

It sounds like you are trying to do what is called pattern reconstruction.  That is a method to simulate a point pattern that is constrained in one or more ways.  The possible ways include the intensity function and the K or L function.  The original paper is Tscheschel and Stoyan, 2006,  Statistical reconstruction of random point patterns, Computational Statistics and Data Analysis 51:859-871.  A textbook treatment is in Wiegand and Moloney, 2014, Handbook of spatial point-pattern analysis in ecology, CRC Press.  The methods are implemented in the R shar package.  Look for the reconstruct functions.

In response to your comments, I note that matching the second order properties (K function, variogram) usually results in a pattern that "appears" quite different, in the sense that locations of high intensity are in different places.  That's to be expected because matching the correlation structure does not mean matching the intensity.

I also note that when you match on a lot of characteristics, you will always simulate patterns that look a lot like your original one because patterns can only differ in so many ways.

Best wishes,
Philip Dixon


	[[alternative HTML version deleted]]