Timeseries modeling of piezometric heads: determination of precipitation response in the presence of other stresses and hydrogeologic heterogeneity

Time series of piezometric head implicitly contain information about groundwater system. This information is reflected in the impulse response function of precipitation. In order to use these functions as a signature of the system, they must be determined with appropriate reliability.

In order to investigate the effect of other influences on the determination of the precipitation response, piezometric data have been studied from a transect running from dunes along the Dutch North Sea coast with natural vegetation into a polder area with agricultural and urban land use. In the dune area, there are several drinking water abstractions with known abstraction rates or levels.

Each time series of piezometric head has been modeled with precipitation, evaporation and up to 4 extractions as explanatory variables. The precipitation impulse response functions from these models have been evaluated per piezometer and compared to neighboring piezometers. Over-parametrization gives more problems for the determination of the precipitation impulse response than the presence of other influences as long as these are not too correlated with the precipitation. The reliability of the precipitation response can be assessed using spatial correlation with neighboring piezometers and the local hydrogeology.

Reliable impulse response functions can be used for interpolation and prediction of heads and for characterisaton of the hydrogeological settings.