prioritizr (version 0.1.2)

gaussian_field: Generate spatially autocorrelated random field

Description

Generate spatially autocorrelated random fields based on a Gaussian process model. This function outputs a RasterLayer or RasterStack object representing the random field. These fields can be used as semi-realistic distributions of species or other spatial variables. Building in spatial autocorrelation ensures that these random fields obey Tobler's first law of geography: "everything is related to everything else, but near things are more related than distant things".

Usage

gaussian_field(r, range, n = 1, mean = 0, variance = 1, nugget = 0, coef = c(0, 0), prop = NULL)

Arguments

r

Raster* object; the template for the random field.

range

numeric; the range of the underlying variogram, the distance beyond which autocorrelation is negligible. This argument defines the spatial scale of the autocorrelation.

n

integer; number of random fields to generate.

mean

numeric; mean value of the random field.

variance

numeric; the sill of the underlying variogram model, which can be thought of as the variance of the random field.

nugget

numeric; the nugget component of the underlying variogram model

coef

numeric vector; if a linear trend in the x and/or y direction is desired, this a vector with 2 components representing the slopes in the x and y directions, respectively.

prop

numeric value between 0-1; if pct is supplied, the field will be reclassified into binary values, where pct represents the proportion of ones. Useful for generating presence/absence layers.

Value

RasterStack

Examples

e <- raster::extent(0, 100, 0, 100)
r <- raster::raster(e, nrows = 100, ncols = 100, vals = 1)
gf <- gaussian_field(r, range = 20, n = 4)
raster::plot(gf)

# generate binary rasters with different ranges
gf_5 <- gaussian_field(r, range = 5, n = 1, prop = 0.5)
gf_20 <- gaussian_field(r, range = 20, n = 1, prop = 0.5)
s <- raster::stack(gf_5, gf_20)
raster::plot(s)

# add a linear trend
gf_linear <-gaussian_field(r, range = 20, coef = c(0.05, 0.05))
raster::plot(gf_linear)