Calibrating a model by regression results in a covariance matrix for the parameters. The eigen values eigen vectors of this matrix provide the same information. Because only a few eigen vectors matter, the bulk of the information is contained in only a couple of these vectors and values, which may be interpreted hydrologically to help better understand the behavior of the system. The paper works out the theory and applies it to a groundwater example having structure similar to the Amsterdam Water Supply Dunes. Next to this, it is applied to the full size multi layer Modflow model of the dune area to show how it works in more complicated situations.