gdistance (version 1.1-2)

transition: Create an object of the class Transition

Description

Create a Transition object from a RasterLayer or RasterBrick object. Transition values are calculated with a user-defined function from the grid values.

Usage

transition (object, transitionFunction, directions, ...)

Arguments

object

RasterLayer or RasterBrick (raster package)

transitionFunction

Function to calculate transition values from grid values

directions

Directions in which cells are connected (4, 8, 16, or other), see adjacency

...

additional arguments, see methods 1 and 3 below

Value

• Transition object

Details

Users may use one of three methods to construct a Transition* object with this function. 1) TransitionLayer from RasterLayer transition(object, transisitonFunction, directions, symm) When a symmetric transition matrix is required, the user should supply a transitionFunction f that obeys f(i,j) = f(j,i) (a commutative function). The function transition does no commutativity check. To obtain an asymmetric transition matrix, a non-commutative function should be supplied and an additional argument `symm' should be set to FALSE. 2) TransitionLayer from RasterBrick transition(object, transitionFunction="mahal", directions) This method serves to summarize several layers of data in a single distance measure. The distance between adjacent cells is the normalized reciprocal of the Mahalanobis distance (mean distance / (mean distance + distance ij). 3) TransitionStack from RasterLayer In contrast with the above methods, this method produces resistance matrices by default. a) Continuous variables - barriers transition(object, transitionFunction="barriers", directions, symm, intervalBreaks) This method creates a TransitionStack with each layer containing a discrete boundary between areas in object. Areas are defined by intervals in object. The argument intervalBreaks is a vector of interval breaks corresponding to the values in object. If between a pair of cells i and j, min(i,j) < break AND max(i,j) > break, then the value ij in the transition matrix becomes 1. All other values in the transition matrix remain 0. The package classInt offers several methods to define intervals. If symm is changed from the default (TRUE) to "up" or "down", it will give either only the upslope (symm="up") or downslope (symm="down") barriers. b) Categorical variables - barriers transition(object, transitionFunction="barriers", directions) In this case, areas are defined as categories in the input raster. A raster with a categorical variable can be created with asFactor(). The layers of the resulting TransitionStack contain all possible combinations of categories. Which layer contains the combination of categories i and j out of n categories, can be determined with these formulae: if symm is TRUE: layer(i,j) = n*(j-1) - j*(j-1)/2 + i-j. if symm is FALSE and i>j: layer(i,j) = ((n*(j-1) - j*(j-1)/2 + i-j) * 2) - 1. if symm is FALSE and itransition(object, transitionFunction="areas", directions) Here, areas are also a categorical variable (see under 3b). The layers in the resulting TransitionStack represent each one area. Connections between two cells which are each inside the area are set to 1. Connections between a cell inside and a cell outside the area are set to 0.5. Connections between two cells outside the area are set to 0.

Examples

#Create a new raster and set all its values to unity.
  r <- raster(nrows=18, ncols=36) 
  r <- setValues(r, runif(ncell(r)))

  #Create a Transition object from the raster
  tr <- transition(r, transitionFunction=mean, directions=4)
  tr #the sparse matrix is of class dsCMatrix (symmetric)

  #Create an asymmetric transition matrix
  #first, an arbitrary non-commutative function 
  f <- function(x) max(x) - x[1] + x[2] 
  tr2 <- transition(r, f, 4, symm=FALSE)
  tr2 #the sparse matrix is of class dgCMatrix (=asymmetric)
  
  #Barriers - interval breaks
  tr3 <- transition(r, "barriers", 8, intervalBreaks=c(0.25, 0.5, 0.75))
  nlayers(tr3)

  #Barriers - categories
  r <- round(r * 2)
  r <- asFactor(r)
  tr4 <- transition(r, "barriers", 8)
  nlayers(tr4)
  plot(raster(tr4[[2]]))
  
  #Areas
  r <- round(r * 2)
  r <- asFactor(r)
  tr5 <- transition(r, "areas", 8)
  nlayers(tr5)
  plot(raster(tr5[[2]]))