[R-sig-Geo] simulation of random fields of vectors

[Kjetil brinchmann Halvorsen]

The title says it all: I have been searching on CRAN and the spatial
task-view, but itb seem that all the simulation packages
only simulate random scalar fields. Any ideas on how simply simulating
a random field of vectors, for the gaussian case.

The model could be intrinsic, that is, the spatial correlation is the
same for eachy component of the random vector, or
some more complex model, which we can write usefully in the following way:

The joint distribution of Y(x) and Y(x+h) is multinormal, dimension 2p
where p is the dimension of the individual  vectors, with
variance-covariance matrix
the 2px2p-matrix $C \otimes \Sigma(h)$, where \otimes denotes the
kronecker product of matrices.
This means that the spatial correlation matrix is \Sigma, and the
correlation matrix of the components of Y(x)
is C. Here, C would be constant, while \Sigma would depend on the
separation distance h.
C would have dimension p\times p while \Sigma would be 2x2.

Kjetil

-- 
"If you want a picture of the future - imagine a boot stamping on the
human face - forever."

George Orwell (1984)