Spatial-temporal models to monitor groundwater data

Spatial-temporal models are used (a) for the interpolation of hydrographs to locations without observations, and (b) for the definition of boundary values for a transient finite element groundwater flow model. If the first principal component-resulting from a principal component transformation performed on the data set-explains more than 95% of the whole variance, this can be used to analyse the spatial structure of the data set with respect to the temporal behaviour. Kriging can be used to estimate the spatial distribution and the error variance of the first principal component. Using the calculated kriging weights of unobserved grid points and the time series information from its explanatory observation wells, water table hydrographs can be estimated, using the error variance as an indicator of the estimation error. Using the proposed spatial-temporal models the definition of initial boundary values for transient finite element flow models is possible.