Facing a couple of dry and warm years which are consistent with climate change scenarios, there is now increasing need for advanced diagnostic tools for drought risk assessment at the scale of years or decades. Commonly models are used for that purpose. However, they often suffer from a lack of data at sufficient spatial resolution, resulting in substantial uncertainties when applied beyond the bounds of single case studies. On the other hand, recent experience showed that simple extrapolating of trends of observed behaviour would not be adequate due to substantial changes of boundary conditions.
A new method has been developed and tested at a regional scale (about 105 km2). It has been shown recently that most of the variance of groundwater head dynamics at that scale can be ascribed to differing degrees of damping of very similar input signals, that is, groundwater recharge dynamics, depending on the thickness and the texture of the overlying vadose zone. The degree of damping can easily be determined by a principal component analysis of a set of groundwater head time series. The stronger the damping the more pronounced is the memory. It could be shown that for wells with pronounced memory groundwater heads have been decreased for about 40 years in Northeast Germany. Thus the backbone of landscape hydrology has been exhibiting continuous weakening, resulting in increasing drought risk in the mid-term, although intermittent recovery at other sites seems to suggest the opposite.