A sensitivity analysis shows that the change of the variable density groundwater model ‘Waterdunen’ is highly location specific for each input parameter. In addition, it turns out that there are typically three lines of development of the magnitude of change.
Which input to care about? A sensitivity analysis in variable density groundwater modelling
Fresh-salt groundwater modelling is a very useful tool to analyse coastal groundwater systems; so also in Zeeuws-Vlaanderen, The Netherlands. Accurate and precise parameterization in variable density groundwater modelling is expensive and time consuming. Henceforth, it would be very useful if it were clear which input parameters are worth investing in. In this research, by means of a sensitivity analysis, an attempt was done to identify those parameters. First, the most uncertain physical and numeric parameters were looked up in literature. The effect of changing the input values of these parameters on the model output was then studied using the 3D-variable density groundwater model Waterdunen (MOCDENS3D-code). The results show that the influence of each parameter is highly location specific (e.g. coast, vicinity rivers, edges lenses, etc.) and that there are typically three lines of development of the magnitude of change. Regarding the latter, some input parameters have a negligible effect on the model output (1), others show a monotone correlation with the output (2) and a third category has no influence on the output above or below a certain threshold value (3). Finally, the research also pointed out that simulating 1D-observations with a 3D-model might do more harm than good.