Skip to contents

Create Multiple Landscapes

Source: R/landscape_create.R
create_landscapes.Rd

Generates n binary landscapes from the requested patterns, sampling pattern-specific parameters independently for each landscape. This supports the construction of training sets.

  • Control patterns have no spatial structure and differ only in vegetation cover: "bare", "random", "dense".

  • Ecotone patterns have a vegetated and a bare zone separated by a transition: "sharp" (abrupt), "diffuse" (gradual), "fingers" (finger-like extensions), "clustered" (scattered clusters), "bands" (sinusoidal bands).

  • Patch patterns are self-organized, without a boundary: "spots" (vegetation patches), "gaps" (bare gaps), "labyrinth" (maze-like bands).

Usage

create_landscapes(
  n = 50,
  patterns = c("random", "bare", "dense", "sharp", "diffuse", "fingers", "clustered",
    "bands", "spots", "gaps", "labyrinth"),
  width = 100,
  height = 100,
  rotation = c(0, 360),
  params_list = NULL,
  pattern_probs = NULL,
  max_retries = 3
)

Arguments

n

Integer. Number of landscapes to generate (default: 50).

patterns

Character vector. Patterns to sample: "random", "bare", "dense", "sharp", "diffuse", "fingers", "clustered", "bands", "spots", "gaps", or "labyrinth" (default: all patterns).

width

Integer. Width of each landscape in pixels (default: 100).

height

Integer. Height of each landscape in pixels (default: 100).

rotation

Numeric. Angle in degrees (default: c(0, 360)). A single value applies to every rotatable landscape. A length-2 vector gives the bounds of a uniform range sampled as whole degrees. Only "sharp", "diffuse", "fingers", "clustered", and "bands" are rotated; other patterns ignore this argument.

params_list

Named list of pattern parameters (default: NULL). Each name must match a pattern and each element must come from its pattern_*() constructor, for example list(spots = pattern_spots()). A single value is fixed across the batch; a length-2 vector is sampled once per landscape. Omitted patterns use their default sampling ranges.

pattern_probs

Numeric vector of sampling weights, one per element of patterns (default: NULL). NULL creates balanced pattern counts. A vector of the wrong length issues a warning and uses equal weights.

max_retries

Integer. Maximum retries after a failed landscape generation (default: 3).

Value

A named list of landscape objects, each as returned by create_landscape. Landscape names are "<pattern>_<index>", with "_rot<angle>" appended for rotated landscapes. The list holds fewer than n landscapes if generation still fails after max_retries; a warning reports the shortfall.

See also

landscape to wrap an existing raster, for example a real map, into the same object type; plot_landscapes to plot the result.

Other landscape creation: create_landscape(), pattern_bands(), pattern_bare(), pattern_clustered(), pattern_dense(), pattern_diffuse(), pattern_fingers(), pattern_gaps(), pattern_labyrinth(), pattern_random(), pattern_sharp(), pattern_spots()

Examples

# Generate 20 landscapes
landscapes <- create_landscapes(n = 20)
#> ✔ Successfully generated all 20 training landscapes

# Access a landscape
landscapes[[1]]
#> Landscape: "clustered_1_rot94" [pattern: clustered]
#> -----------------------------------------
#> Dimensions: 100x100 (10000 cells)
#> Resolution: 1.0x1.0
#> Extent    : xmin=0.0, xmax=100.0, ymin=0.0, ymax=100.0
#> Values    : min=0.0, max=1.0
#> Parameters: width = 100, height = 100, boundary_position = 0.471359662851319, n_clusters = 8, cluster_radius = 8, cluster_zone = 0.714724025502801, elongation_x = 1.37432571477257, elongation_y = 1.36488283867948, rotation = 94 

# Check the pattern
landscapes[[1]]$pattern
#> [1] "clustered"

# Get all landscape patterns
sapply(landscapes, \(x) x$pattern)
#>  clustered_1_rot94   diffuse_2_rot221           random_3            spots_4 
#>        "clustered"          "diffuse"           "random"            "spots" 
#>            spots_5            dense_6   fingers_7_rot306             bare_8 
#>            "spots"            "dense"          "fingers"             "bare" 
#>             bare_9       labyrinth_10    bands_11_rot325            gaps_12 
#>             "bare"        "labyrinth"            "bands"             "gaps" 
#>    sharp_13_rot265    bands_14_rot228  diffuse_15_rot300  fingers_16_rot352 
#>            "sharp"            "bands"          "diffuse"          "fingers" 
#>           dense_17     sharp_18_rot74          random_19 clustered_20_rot85 
#>            "dense"            "sharp"           "random"        "clustered" 

# Custom parameters, as a single value or as a range sampled per landscape
landscapes_custom <- create_landscapes(
  n = 12,
  patterns = c("spots", "sharp"),
  params_list = list(
    spots = pattern_spots(n_spots = 15, spot_radius = c(8, 12)),
    sharp = pattern_sharp(boundary_position = c(0.4, 0.6))
  )
)
#> ✔ Successfully generated all 12 training landscapes