[R-sig-Geo] Ordinary Kriging vs Kriging with external drift

[José Luis Rotundo]

Dear All,

I have kind of a conceptual question.

Have been working on a data set of a continuous variable across the
US. I managed to model the spatial dependency by fitting different
model variograms and did ordinary kriging to extrapolate the values. I
find out that for some years there is a significant relationship
between latitude and the variable I am interested in model. However,
in no case the R2 was higher that 0.10. My thinking was that the drift
was low enough in terms of predictive value, so the drift could be
overlooked. Do you agree with this?

Thanks for any hint on this questions.

Best,
Jose

-- 

José L. Rotundo
CONICET
Facultad de Ciencias Agrarias
Univ. Nacional de Rosario
Zavalla, Santa Fe
Argentina