internal function to compute cokriging (simple, ordinary, universal, with trend)
gsi.Cokriging(
Xout,
Xin,
Zin,
vgram,
Fin = rep(1, nrow(Xin)),
Fout = rep(1, nrow(Xout)),
krigVar = FALSE,
tol = 1e-15
)
(M,g)-matrix with coordinates of desired interpolation locations
(N,g)-matrix with coordinates of conditioning locations
(N,D)-matrix with condtioning observations
D-dimensional variogram model of class "gmCgram"
either NULL or a (N,p)-matrix with the base functions of the trend evaluated at the conditioning locations (defaults to a vector of $N$ 1's)
either NULL or a (m,p)-matrix with the base functions of the trend evaluated at the conditioning locations (defaults to a vector of $M$ 1's)
logical or NA, should kriging variances be returned? FALSE=no, TRUE=give point-wise cokriging covariance; NA=return the whole covariance matrix of all points
tolerance for the inversion of the cokriging matrix (in case of near-singularity)
A (M,D)-matrix of predictions, eventually with an attribute "krigVar"
containing the output defined by argument krigVar