Skip to contents

Create Spatial Strata for Soil Sampling

Source: R/stratify.R
ss_stratify.Rd

Partitions a study area into compact geographical strata using k-means clustering. This is the foundation for spatial coverage sampling.

Usage

ss_stratify(
  x,
  n_strata,
  prior_points = NULL,
  max_iter = 1000L,
  n_try = 1L,
  n_cells = 2500L,
  cell_size = NULL,
  equal_area = FALSE,
  verbose = FALSE
)

Arguments

x

An sf object representing the study area. Can be a polygon (POLYGON or MULTIPOLYGON) or point geometry (POINT or MULTIPOINT).

n_strata

Integer, the number of strata to create.

prior_points

Optional sf object with point geometry containing locations that should be used as fixed cluster centers.

max_iter

Integer, maximum number of iterations for the k-means algorithm (default 1000).

n_try

Integer, number of random initial configurations to try (default 1). Higher values reduce risk of suboptimal local minima.

n_cells

Integer, approximate number of grid cells to use when discretizing a polygon (default 2500).

cell_size

Numeric vector of length 1 or 2, cell size for discretization. If provided, n_cells is ignored.

equal_area

Logical, if TRUE, use the swopping algorithm to create strata of equal area (default FALSE). Cannot be used with prior points.

verbose

Logical, whether to print progress messages (default FALSE).

Value

An object of class ss_strata containing:

cells

An sf object with the grid cells and their stratum assignments.

centroids

An sf object with the stratum centroids.

cell_size

Numeric vector with cell dimensions.

n_strata

Integer, number of strata.

mssd

Numeric, mean squared shortest distance (objective function).

converged

Logical, whether the algorithm converged.

prior_points

The prior points if provided, otherwise NULL.

equal_area

Logical, whether equal-area stratification was used.

crs

The coordinate reference system.

Details

The function partitions the study area into compact spatial strata by minimizing the mean squared shortest distance (MSSD) between grid cells and their assigned cluster centroids.

Two algorithms are available:

  • Transfer algorithm (default): Standard k-means, creates compact strata of potentially unequal sizes.

  • Swop algorithm (equal_area = TRUE): Creates compact strata of equal size by swapping cells between clusters.

See also

ss_coverage(), ss_plot()

Examples

if (FALSE) { # \dontrun{
library(sf)

# Create a study area
poly <- st_polygon(list(rbind(c(0, 0), c(100, 0), c(100, 50), c(0, 50), c(0, 0))))
study_area <- st_sf(geometry = st_sfc(poly))

# Create 20 strata
strata <- ss_stratify(study_area, n_strata = 20, n_try = 5)
plot(strata)
} # }