An analysis carrying a scenario holds both of its landscapes, so the layer
that actually changed is whichever of them the scenario stood in for. Ask
rather than branching on scenario_kind at every call site.
Arguments
- x
A
connectivity_report_data, fromconnectivity_report_data().
See also
connectivity_report_data(), and gg_layer() to draw it.
Examples
# \donttest{
analysis <- connectivity_report_data(
habitat = example_wren_habitat(),
barrier = example_wren_barrier(),
species = "Superb Fairy Wren",
interpatch_distance = 200,
scenario = example_wren_barrier_scenario(),
scenario_kind = "barrier",
verbose = FALSE
)
scenario_layer(analysis)
#> class : SpatRaster
#> size : 1500, 1400, 1 (nrow, ncol, nlyr)
#> resolution : 9.996731, 10.00151 (x, y)
#> extent : 340888.6, 354884, 5796348, 5811351 (xmin, xmax, ymin, ymax)
#> coord. ref. : GDA94 / MGA zone 55 (EPSG:28355)
#> source : wren_barrier_scenario_rast.tif
#> name : layer
#> min value : 0
#> max value : 1
# }