Skip to contents

Create raster stack representing SDM predictors

Source: R/create_predictor_stack.R
create_predictor_stack.Rd

Prepare for fitting SDMs by transforming a landscape raster into a stack of rasters representing the required predictors.

Usage

create_predictor_stack(x, SDM, classified = FALSE, fill = TRUE)

Arguments

x

SpatRaster; can only have 1 layer

SDM

The name of intended species distribution model: "riparian", "waterbird_fall", "waterbird_win", or "tima"

classified

logical; see Details

fill

logical; see Details

Value

SpatRaster with separate layers for each land cover class included as a predictor in the selected SDM representing the presence (1) and absence (0)

Details

This function is called by python_focal_prep() and is not intended to be called directly. Segregates a landscape raster into separate layers representing each land cover class. Also calls on internal datsets to create broader grouping variables as required for the selected SDM. The input raster should already be encoded with the land cover classes listed in the key.

If classified = FALSE (the default), this function calls classify_landcover.SpatRaster() to reclassify the input raster according to the land cover classifications expected by the selected SDM. (See documentation from that function for information on warning messages.) If classified = TRUE, it will be assumed to already be classified correctly, which may be of particular use for the 'riparian' models. See Vignette for details.

A warning is given if land cover classes expected by the model are absent from the provided landscape. In that case, if fill = TRUE (the default), additional layers will be created with all zero values for each missing land cover class. However, the input landscape should be carefully reviewed to ensure they are truly absent and have not been excluded from the landscape unintentionally. If needed, the resulting layers can be replaced manually before proceeding with python_focal_run().

See also

classify_landcover()

Examples

r <- terra::rast(matrix(sample(c(11,19,71,72,90), size = 100, replace = TRUE),
         ncol = 10, nrow = 10))
r = suppressWarnings(create_predictor_stack(r, SDM = 'riparian'))
#> AG RICE IDLE GRASSPAS URBAN SALIX MIXEDFOREST INTROSCRUB SALIXSHRUB MIXEDSHRUB PERM BARREN WOODLAND&SCRUB
#> Because fill = TRUE, creating missing rasters with all zero values, butconfirm they are truly absent from the landscape.