[R-sig-Geo] [R] how do we sample in spatial statistics?

[Adrian Baddeley]

Zhijie Zhang writes:

> Do we also use sample() function to do the spatial sampling?

> In spatial statistics, do we need two-dimentional sampling or some other
> sampling techniques? Which functions could be used in spatial samping in R?

  It depends on what kind of data or population you are sampling.

  (a) If your data or population are stored in a vector, a list or a data frame,
  then you can sample some of the items in the list, simply by using
  the R function sample() to select values of the list index.
  For example 
	mydata <- list(.............)
	index <- sample(seq(mydata), 3, replace=FALSE) 
	sam <- mydata[index]
  will select three items at random from the list.

  In package 'spatstat', a pattern of points or a pattern of line segments
  can also be indexed by numerical values. For example
	data(cells)
	index <- sample(cells$n, 3, replace=TRUE)
	sam <- cells[index]
  takes the `cells' dataset, a spatial point pattern, 
  and selects three of the points at random.
	
  (b) If your data are spatially referenced (for example a map of
  temperatures across the ocean), you may want to sample these data
  at `spatially random' locations. In principle you can use any
  random pattern of points to do this. 

  Package 'spatstat' contains a large number of functions for generating
  random patterns of points. Type help(spatstat) for a list of them.
  These functions can be used to sample spatially-referenced data in any package. 

  In spatstat, for example, if ocean temperatures are stored as a pixel image, you can 
  sample them using a point pattern:
	temperature <- im(........)
	samplepoints <- rstrat(temperature$window, 10, 10, 3)
	sam <- temperature[samplepoints]
   creates an image called 'temperature', then generates a stratified random pattern
   of points, and extracts the ocean temperatures at these points.

Adrian Baddeley