setCgram {gmGeostats}

Generate D-variate variogram models

Description

Function to set up D-variate variogram models based on model type, the variogram parameters sill and nugget and a matrix describing the anisotropy of the range.

Usage

setCgram(type, nugget = sill * 0, sill, anisRanges, extraPar = 0)

Arguments

type

model of correlation function. The function expects a constant, e.g. the internal constants 'vg.Gau' for Gaussian model or 'vg.Exp'. for exponential models. See examples for usage.

nugget

(DxD)-matrix for the nugget effect. Default is a muted nugget (0).

sill

(DxD)-matrix for the partial sills of the correlation function

anisRanges

2x2 or 3x3 matrix of ranges (see details)

extraPar

for certain correlation functions, extra parameters (smoothness, period, etc)

Details

The argument type must be an integer indicating the model to be used. Some constants are available to make reading code more understandable. That is, you can either write 1, vg.sph, vg.Sph or vg.Spherical, they will all work and produce a spherical model. The same applies for the following models: vg.Gauss = vg.Gau = vg.gau = 0; ⁠vg.Exponential = vg.Exp = vg. exp = 2⁠. These constants are available after calling data("variogramModels"). No other model is currently available, but this data object will be regularly updated. The constant vector gsi.validModels contains all currently valid models.

Argument anisRange expects a matrix $M$ such that

h^2 = (\mathbf{x}_i-\mathbf{x}_j)\cdot M^{-1}\cdot (\mathbf{x}_i-\mathbf{x}_j)^t

is the (square of) the lag distance to be fed into the correlation function.

Value

an object of class "gmCgram" containing the linear model of corregionalization of the nugget and the structure given.

Examples

utils::data("variogramModels") # shortcut for all model constants
v1 = setCgram(type=vg.Gau, sill=diag(2), anisRanges = 3*diag(c(3,1)))
v2 = setCgram(type=vg.Exp, sill=0.3*diag(2), anisRanges = 0.5*diag(2))
vm = v1+v2
plot(vm)