blockCV (version 2.1.1)

spatialAutoRange: Measure spatial autocorrelation in the predictor raster files

Description

This function provides a quantitative basis for choosing block size. The spatial autocorrelation in all continuous predictor variables available as raster layers is assessed and reported. The function estimates spatial autocorrelation ranges of all input raster layers. This is the range over which observations are independent and is determined by constructing the empirical variogram, a fundamental geostatistical tool for measuring spatial autocorrelation. The empirical variogram models the structure of spatial autocorrelation by measuring variability between all possible pairs of points (O'Sullivan and Unwin, 2010). Results are plotted. See the details section for further information.

Usage

spatialAutoRange(
  rasterLayer,
  sampleNumber = 5000L,
  border = NULL,
  speciesData = NULL,
  doParallel = FALSE,
  nCores = NULL,
  showPlots = TRUE,
  degMetre = 111325,
  maxpixels = 1e+05,
  plotVariograms = FALSE,
  progress = TRUE
)

Arguments

rasterLayer

A raster object of covariates to find spatial autocorrelation range.

sampleNumber

Integer. The number of sample points of each raster layer to fit variogram models. It is 5000 by default, however it can be increased by user to represent their region well (relevant to the extent and resolution of rasters).

border

A sf or SpatialPolygons object to clip the block based on it (optional).

speciesData

A spatial or sf object (optional). If provided, the sampleNumber is ignored and variograms are created based on species locations. This option is not recommended if the species data is not evenly distributed across the whole study area and/or the number of records is low.

doParallel

Logical. Run in parallel when more than one raster layer is available. Given multiple CPU cores, it is recommended to set it to TRUE when there is a large number of rasters to process.

nCores

Integer. Number of CPU cores to run in parallel. If nCores = NULL half of available cores in your machine will be used.

showPlots

Logical. Show final plot of spatial blocks and autocorrelation ranges.

degMetre

Numeric. The conversion rate of metres to degree. This is for constructing spatial blocks for visualisation. When the input map is in geographic coordinate system (decimal degrees), the block size is calculated based on deviding the calculated range by this value to convert to the input map's unit (by default 111325; the standard distance of a degree in metres, on the Equator).

maxpixels

Number of random pixels to select the blocks over the study area.

plotVariograms

Logical. Plot fitted variograms. This can also be done after the analysis. It is FALSE by default.

progress

Logical. Shows progress bar. It works only when doParallel = FALSE.

Value

An object of class S3. A list object including:

• range - the suggested range, which is the median of all calculated ranges

• rangeTable - a table of input covariates names and their autocorrelation range

• plots - the output plot (the plot is shown by default)

• sampleNumber

• variograms - fitted variograms for all layers

Details

The input raster layers should be continuous for computing the variograms and estimating the range of spatial autocorrelation. The input rasters should also have a specified coordinate reference system. However, if the reference system is not specified, the function attempts to guess it based on the extent of the map. It assumes an unprojected reference system for layers with extent lying between -180 and 180, and a projected reference system otherwise.

Variograms are calculated based on the distances between pairs of points, so unprojected rasters (in degrees) will not give an accurate result (especially over large latitudinal extents). For unprojected rasters, the great circle distance (rather than Euclidian distance) is used to calculate the spatial distances between pairs of points. To enable more accurate estimate, it is recommended to transform unprojected maps (geographic coordinate system / latitude-longitude) to a projected metric reference system (e.g. UTM or Lambert) where it is possible. See autofitVariogram from automap and variogram from gstat packages for further information.

References

O'Sullivan, D., Unwin, D.J., (2010). Geographic Information Analysis, 2nd ed. John Wiley & Sons.

Roberts et al., (2017). Cross-validation strategies for data with temporal, spatial, hierarchical, or phylogenetic structure. Ecography. 40: 913-929.

Examples

# NOT RUN {
# load the example raster data
awt <- raster::brick(system.file("extdata", "awt.grd", package = "blockCV"))

# run the model in parallel
range1 <- spatialAutoRange(rasterLayer = awt,
                           sampleNumber = 5000, # number of cells to be used
                           doParallel = TRUE,
                           nCores = 2, # if NULL, it uses half of the CPU cores
                           plotVariograms = FALSE,
                           showPlots = TRUE)


# run the model with no parallel
range3 <- spatialAutoRange(rasterLayer = awt,
                           sampleNumber = 5000,
                           doParallel = FALSE,
                           progress = TRUE)

# show the result
summary(range1)
# }