Spatiotemporal Data Analysis
The modelling of solute distribution in groundwater is typically restricted to either the analysis of trends in individual wells or independent fitting of spatial concentration distributions (e.g. by Kriging) to data from monitoring events. Neither of these techniques satisfactorily elucidate the interaction between spatial and temporal components of the data.
GWSDAT applies a spatiotemporal model smoother for a more coherent and smooth interpretation of the interaction in spatial and time-series components of groundwater solute concentrations. A spatiotemporal concentration smoother is fitted for each analyte using a non-parametric regression technique known as Penalised Splines (Eilers and Marx, 1992, 1996).
A Bayesian methodology is used to select the appropriate degree of model smoothness (Evers et al, 2015). The fit of the spatiotemporal algorithm to the monitoring data can be evaluated.