An extension of Box-Jenkins transfer/noise models for spatial interpolation of groundwater head series

This paper advocates an approach to extend single-output Box-Jenkins transfer/noise models for several groundwater head series to a multiple-output transfer/noise model, The approach links several groundwater head series and enables a spatial interpolation in terms of time series analysis. Our multiple-output transfer/noise model relates the single-output transfer/noise models from individual series by taking the spatial correlation of the white noise process into account and spatially interpolating the parameters of the transfer and noise models. The parameters of the time series models and the white noise process are considered to be spatial stochastic fields, and are described geostatistically. The model parameters and the noise variance are interpolated by means of Kriging, The approach's applicability is illustrated by two cases: a point study (an observation well with measured data from seven observation screens) and an area study that requires spatial interpolation, These show the method's usefulness for examining the effectiveness of monitoring strategies, filling in missing data and for the quality control of measured data, It also enables the responses of the single-output transfer/noise models to be spatially interpolated. (C) 1997 Elsevier Science B.V.