What drives spatial variance of hydrological behaviour? An analysis of the regional groundwater-stream continuum

Abstract. A sound understanding of regional scale hydrological processes and influencing factors is indispensable for sustainable water resources management. It requires differentiation between natural heterogeneities, direct anthropogenic effects, and climate change impacts. In addition, an integral perspective is required comprising both surface and groundwater bodies. This study aimed at determining the key drivers for the spatial variance of hydrological behaviour, at gaining an understanding why long-term trends of observed behaviour often differ in spite of spatial proximity and similar boundary conditions, and at investigating the added value of merging stream discharge and groundwater head data for the analysis. A set of 292 time series of stream discharge and groundwater head from a 36,000 km2 region in South Germany, covering a 43 years period, was subjected to a principal component analysis. The first six components were analysed in more detail. All together they explained 77.8 % of the total variance. The first component grasped mean behaviour. Three components reflected various facets of climate patterns. Land use effects were not found to be significant when the common dependence of land use and hydrology on climate patterns was factored out. Two further components described the damping of the hydrological input signal in the subsurface. One of these differentiated between porous substrates and fractured or karstified hardrocks. Damping of the input signal was very closely related to direction and strength of long-term trends. Trends were the most clearly visible in deep groundwater time series which are suggested to be used as early warning indicators with regard to climate change rather than shallow groundwater or stream discharge. In general, the combined analysis of stream discharge and groundwater head proved to be very efficient, benefitting from complementary sensitivities toward single processes and effects, and is highly recommended for future analyses.