Simulate a three-level occupancy + cover hurdle data set — simulate_occu_multiscale

Draws cell occupancy z_c ~ Bernoulli(psi_c), plot availability a_cj | z_c=1 ~ Bernoulli(theta_cj), visit detection y_cjv | a_cj=1 ~ Bernoulli(p_cjv), and the cover hurdle cover_cjv | y_cjv=1 ~ f_pos, with a shared areal ICAR field on the occupancy (sigma) and cover (alpha * sigma) arms. The data-generating process for occu_multiscale_cover() recovery tests.

Usage

simulate_occu_multiscale_cover(
  n_cells = 60L,
  plots_per_cell = 4L,
  visits_per_plot = 2L,
  beta_psi = c(0.4, 0.6),
  beta_theta = c(0.2, 0.5),
  beta_p = c(0, 0.5),
  beta_pos = c(log(0.1), -0.4),
  positive = c("lognormal", "beta", "gaussian"),
  phi = 0.35,
  adj = NULL,
  sigma = 0.7,
  alpha = 1,
  trend = FALSE,
  sigma_trend = 0.7,
  alpha_trend = 1,
  seed = NULL
)

Arguments

n_cells: number of areal cells (graph nodes).
plots_per_cell: plots (the availability units) per cell.
visits_per_plot: detection replicate visits per plot.
beta_psi, beta_theta, beta_p, beta_pos: length-2 c(intercept, slope) coefficients for the four arms (one covariate each).
positive: "lognormal" or "beta" cover arm.
phi: cover dispersion (lognormal log-scale SD, or beta precision).
adj: n_cells x n_cells adjacency; default a 1-D chain.
sigma: areal-field marginal SD on the occupancy arm.
alpha: cover-arm field scaling (alpha * sigma on the cover arm).
trend: add a second, spatially-varying-coefficient (SVC) areal field weighted per cell by a cell-level covariate tcov (added to data): tcov_c * sigma_trend * f_trend[c] on occupancy and tcov_c * alpha_trend * sigma_trend * f_trend[c] on cover. Default FALSE.
sigma_trend, alpha_trend: trend-field marginal SD and its cover-arm scaling (used only when trend = TRUE).
seed: optional RNG seed.

Value

A list with y, y_pos ([n_plots x visits_per_plot]), the plot-level data (cell id cell, covariates), adj, and truth.

Examples

sim <- simulate_occu_multiscale_cover(n_cells = 10, plots_per_cell = 2,
                                      visits_per_plot = 2, seed = 1)
dim(sim$y)