https://github.com/cran/RandomFields
Tip revision: fab3d29ef16569604858ee648b9e1f6f7d4a7c96 authored by Martin Schlather on 21 September 2014, 00:00:00 UTC
version 3.0.42
version 3.0.42
Tip revision: fab3d29
RMmatern.Rd
\name{RMwhittlematern}
\alias{RMwhittle}
\alias{RMkbessel}
\alias{RMmatern}
\alias{Sobolev}
\title{Whittle-Matern Covariance Model}
\description{
\command{\link{RMmatern}} is a stationary isotropic covariance model
belonging to the Matern family.
The corresponding covariance function only depends on the distance
\eqn{r \ge 0}{r \ge 0}
between two points.
The Whittle model is given by
\deqn{C(r)=W_{\nu}(r)=2^{1- \nu}
\Gamma(\nu)^{-1}r^{\nu}K_{\nu}(r)}{C(r)=W_{\nu}(r)=2^{1- \nu}
\Gamma(\nu)^{-1}r^{\nu}K_{\nu}(r)}
where \eqn{\nu > 0}{\nu > 0} and \eqn{K_\nu}{K_\nu} is the modified
Bessel function of second kind.
The Matern model is given by
\deqn{C(r) = \frac{2^{1-\nu}}{\Gamma(\nu)} (\sqrt{2\nu}r)^\nu
K_\nu(\sqrt{2\nu}r)}{C(r) = 2^{1- \nu} \Gamma(\nu)^{-1} (\sqrt{2\nu}
r)^\nu K_\nu(\sqrt{2\nu} r)}
}
\usage{
RMwhittle(nu, notinvnu, var, scale, Aniso, proj)
RMmatern(nu, notinvnu, var, scale, Aniso, proj)
}
\arguments{
\item{nu}{a numerical value called \dQuote{smoothness parameter};
should be greater than 0.}
\item{notinvnu}{logical, if not given the model is defined as above.
(default). This argument should not be set
by users. See the Notes.}
\item{var,scale,Aniso,proj}{optional arguments; same meaning for any
\command{\link{RMmodel}}. If not passed, the above
covariance function remains unmodified.}
}
\details{
\command{\link{RMwhittle}} and \command{\link{RMmatern}}
are two alternative parametrizations of the same covariance function.
The Matern model should be preferred as this model seperates the
effects of scaling parameter and the shape parameter.
This is the normal scale mixture model of choice if the smoothness of a
random field is to be parametrized: the sample paths of a Gaussian
random field with this covariance structure are \eqn{m}{m} times
differentiable if and only if \eqn{\nu > m}{\nu > m} (see Gelfand et
al., 2010, p. 24).
Furthermore, the fractal dimension (see also \command{\link{RFfractaldim}})
\emph{D} of the Gaussian sample paths
is determined by \eqn{\nu}{\nu}: we have
\deqn{D = d + 1 - \nu, \nu \in (0,1)}{D = d + 1 - \nu, 0 < \nu < 1}
and \eqn{D = d}{D = d} for \eqn{\nu > 1}{\nu > 1} where \eqn{d}{d} is
the dimension of the random field (see Stein, 1999, p. 32).
If \eqn{\nu=0.5}{\nu=0.5} the Matern model equals \command{\link{RMexp}}.
For \eqn{\nu}{\nu} tending to \eqn{\infty}{\infty} a rescaled Gaussian
model \command{\link{RMgauss}} appears as limit of the Matern model.
For generalisations see section \sQuote{seealso}.
}
\note{
The Matern model called by \eqn{C(r \sqrt(2))}{C(r \sqrt(2))} equals
the Handcock-Wallis (1994) parametrisation.
The model allows further to be reparameterized by substituting
\eqn{\nu}{\nu} for \eqn{\nu^{-1}}{\nu^{-1}} setting the argument
\code{invnu=TRUE}. Note that the inversion of \eqn{\nu} does not really
make sense for the Whittle model. Due to this fact, if the argument
\code{invnu} is given, the Whittle model changes its definition and
becomes identical to the Matern model.
}
\value{
The function return an object of class \code{\link[=RMmodel-class]{RMmodel}}
}
\references{
Covariance function
\itemize{
\item Chiles,
J.-P. and Delfiner, P. (1999)
\emph{Geostatistics. Modeling Spatial Uncertainty.}
New York: Wiley.
\item Gelfand, A. E., Diggle, P., Fuentes, M. and Guttorp,
P. (eds.) (2010) \emph{Handbook of Spatial Statistics.}
Boca Raton: Chapman & Hall/CRL.
\item Guttorp, P. and Gneiting, T. (2006) Studies in the
history of probability and statistics. XLIX. On the Matern
correlation family. \emph{Biometrika} \bold{93}, 989--995.
\item Handcock, M. S. and Wallis, J. R. (1994) An approach to
statistical spatio-temporal modeling of meteorological fields.
\emph{JASA} \bold{89}, 368--378.
\item Stein, M. L. (1999) \emph{Interpolation of Spatial Data --
Some Theory for Kriging.} New York: Springer.
}
Tail correlation function (for \eqn{\nu \in (0,1/2]}{0 < \nu \le 1/2})
\itemize{
\item Strokorb, K., Ballani, F., and Schlather, M. (2014)
Tail correlation functions of max-stable processes: Construction
principles, recovery and diversity of some mixing max-stable
processes with identical TCF.
\emph{Extremes}, \bold{} Submitted.
}
}
\author{Martin Schlather, \email{schlather@math.uni-mannheim.de}
}
\seealso{
\itemize{
\item \command{\link{RMexp}}, \command{\link{RMgauss}} for special
cases of the model (for \eqn{\nu=0.5}{\nu=0.5} and
\eqn{\nu=\infty}{\nu=\infty}, respectively)
\item \command{\link{RMhyperbolic}} for a univariate
generalization
\item \command{\link{RMbiwm}} for a multivariate generalization
\item \command{\link{RMnonstwm}}, \command{\link{RMstein}} for anisotropic (space-time) generalizations
% \item \command{\link{}} for
\item \command{\link{RMmodel}},
\command{\link{RFsimulate}},
\command{\link{RFfit}} for general use.
}
}
\keyword{spatial}
\keyword{models}
\examples{
RFoptions(seed=0) ## *ANY* simulation will have the random seed 0; set
## RFoptions(seed=NA) to make them all random again
x <- seq(0, 1, len=if (interactive()) 100 else 3)
model <- RMwhittle(nu=1, Aniso=matrix(nc=2, c(1.5, 3, -3, 4)))
plot(model, dim=2, xlim=c(-1,1))
z <- RFsimulate(model=model, x, x)
plot(z)
\dontshow{FinalizeExample()}
}