acorr: Calculates the full spatial autocorrelation on a raster using...

In AdamWilsonLab/rasterAutocorr: Quickly calculate spatial autocorrelation on 2D rasters.

Description

Applies the Wiener-Khinchin theorem to extract spatial autocorrelation using Fast Fourier Transform techniques. This results in an extremely fast way to calculate a complete correlogram (correlation as a function of distance) for a raster image.

Usage

acorr(x, padlongitude = T, verbose = T, ...)

Arguments

x	A raster* object. Missing values are indicated by NA.
file	File to write results to as in writeRaster. If NULL a temporary file is written as in the raster package.

Value

The spatial autocorrelation matrix

References

en.wikipedia.org/wiki/WienerKhinchin_theorem

Xianlin Ma, Tingting Yao, A program for 2D modeling (cross) correlogram tables using fast Fourier transform, Computers & Geosciences, Volume 27, Issue 7, August 2001, Pages 763-774, ISSN 0098-3004, http://dx.doi.org/10.1016/S0098-3004(01)00007-3.

http://www.johnloomis.org/ece563/notes/freq/autoself/autoself.htm

http://www.seas.upenn.edu/~ese502/NOTEBOOK/Part_II/4_Variograms.pdf

Examples

# packages used for the data generation
require(raster)
require(fields)
library(dplyr)

## first show the examples in Marcotte, Denis. 1996. "Fast Variogram Computation with FFT." Computers & Geosciences 22 (10): 1175–86. doi:10.1016/S0098-3004(96)00026-X.
m1=raster(matrix(c(3,6,5,7,2,2,4,NA,0),ncol=3,byrow=T))
m2=raster(matrix(c(10,NA,5,NA,8,7,5,9,11),ncol=3,byrow=T))

ac=acorr(m1,padlongitude=T,verbose=T)

## confirm nobs == nh11 on top of page 1179
10^as.matrix(ac[["nobs"]])

## comfirm acor= gh11 on top of page 1179
round(as.matrix(ac[["acor"]])/10,3)

AdamWilsonLab/rasterAutocorr documentation built on March 19, 2020, 12:01 a.m.