gmGeostats (version 0.11.3)

KrigingNeighbourhood: Create a parameter set of local for neighbourhood specification.

Description

Create a parameter set describing a kriging neighbourhood (local or global) for cokriging and cokriging based simulation. This heavily relies on the definitions of gstat::gstat(). All parameters are optional, as their default amounts to a global neihghbourhood.

Usage

KrigingNeighbourhood(
  nmax = Inf,
  nmin = 0,
  omax = 0,
  maxdist = Inf,
  force = FALSE,
  anisotropy = NULL,
  ...
)

Value

an S3-list of class "gmKrigingNeighbourhood" containing the six elements given as arguments to the function. This is just a compact way to provide further functions such as predict_gmSpatialModel

with appropriate triggers for choosing a prediction method or another, in this case for triggering cokriging (if alone) or eventually sequential simulation (see SequentialSimulation()).

Arguments

nmax

maximum number of data points per cokriging system

nmin

minimum number of data points per cokriging system

omax

maximum number of data points per cokriging system per quadrant/octant

maxdist

maximum radius of the search neighborhood

force

logical; if less than nmin points are found inside maxdist radius, keep searching.

anisotropy

currently ignored; in the future, argument to specify anisotropic search areas.

...

further arguments, currently ignored

Examples

data("jura", package="gstat")
X = jura.pred[,1:2]
summary(X)
Zc = jura.pred[,7:10]
ng_global = KrigingNeighbourhood()
ng_local = KrigingNeighbourhood(maxdist=1, nmin=4, 
                                omax=5, force=TRUE)
ng_local
ng_global
make.gmCompositionalGaussianSpatialModel(data = Zc, coords = X,
                                         V = "alr", ng = ng_local)