spatial_bootstrap builtin

spatial_bootstrap(
    coords,
    values,
    variogram,
    *,
    weights=None,
    n=100,
    seed=0,
    quantiles=Ellipsis,
    cutoffs=None,
)

Uncertainty in global statistics of spatially correlated data.

Each realization draws unconditional Gaussian values at the sample locations with the correlation of variogram, turns them into uniform ranks and reads them through the declustered distribution of values. The statistics of the resampled values then spread as far as the spatial correlation allows: with a pure nugget they match the classical bootstrap, with ranges much longer than the data extent the samples act as one.

Parameters:

Name	Type	Description	Default
coords	(array_like, PointSet or BlockModel)	(n, 2) or (n, 3) sample coordinates, or a container.	required
values	array_like or str	(n,) values, or their column in coords.	required
variogram	Variogram	Normal-score variogram; only its correlation (the variogram over its total sill) is used.	required
weights	array_like or str	Declustering weights, or their column; default equal.	None
n	int	Number of realizations.	100
seed	int	Seed; each realization draws from a seed mixed from it.	0
quantiles	sequence of float	Probabilities of quantile columns P10, P50, ...	Ellipsis
cutoffs	sequence of float	Cutoffs of proportion columns above 1.5, ...: the fraction of values above each cutoff.	None

Returns:

Type	Description
Table	One row per realization: mean, then the quantile and proportion columns.

used in

• spatial bootstrap