Skip to content
Prev 753 / 29559 Next

External Drift Kriging

Hi Luke
OK. This is one option indeed, to modelling covariance after getting 
"residuals" from some kind of model to take auto the effect of known 
sources of variation.
And indeed the usual variogram will only make sense for stationary data.

In the max. lik. estimation you can do both together, getting estimates 
for a (linear) model for the mean and for the covariance parameters.

An example call would be something along the lines
provided foo is a geodata object with covariate information on altitude

Later for kriging you would need a call of the type:
trend.l = ~altitude.gr), borders=NULL)

where gr is a matrix with coordinates of prediction points and
altitude.gr is the value of the covariate at those prediction points,
So you can only krige within the borders, unless you extrapolate your 
covariate information somehow.
Kriging with external trend (w/ covariates) requires values of the 
coavriates at all prediction locations.


Hope this helps, and if the problempersists feel free to contact me 
directly

best
P.J.
Paulo Justiniano Ribeiro Jr
LEG (Laborat?rio de Estat?stica e Geoinforma??o)
Departamento de Estat?stica
Universidade Federal do Paran?
Caixa Postal 19.081
CEP 81.531-990
Curitiba, PR  -  Brasil
Tel: (+55) 41 3361 3573
Fax: (+55) 41 3361 3141
e-mail: paulojus at est.ufpr.br
http://www.est.ufpr.br/~paulojus