Uncertainty evaluation in functional kriging with external drift

We predict a curve at an unmonitored site taking into account exogenous
variables using a functional kriging model with external drift. To evaluate
uncertainty of the predicted curves, a semi-parametric bootstrap approach for
spatially correlated data is adapted to functional data. Confidence bands are obtained
by ordering the bootstrapped predicted curves in two different ways. The
proposed approach is illustrated on pollutant functional data gathered by the
monitoring network of Piemonte region (Italy).