Spatial cross-correlation patterns of European low, mean and high flows

Low and high flows are associated with different hydrological processes. High flows correspond to the direct response of catchments to water input, whereas low flows occur in pronged dry periods and are governed by depleting storages. Therefore, the inter-annual dynamics of high and low flows are often considered to be independent. To shed light on this assumption, we analysed a pan-European dataset of 615 streamflow records, summarized as time series of annual streamflow percentiles (5th, 10th, ... , 95th). The analysis was based on comparing the spatial cross-correlation patterns derived from the different percentile series. Their interrelation was visualized by projecting them into a low-dimensional space. We found that large parts of the cross-correlations of the percentile series can be summarized by one dominating component. This component represents geographical continuous regions in Europe of correlated streamflow. Departures from this mean pattern occurred for low and high flows and were characterized by the corresponding spatial correlation functions. Generally, spatial correlations appear to be stronger for high flows than for mean flows, particularly for short distances (< 400 km). Low flows, on the other hand, have the lowest spatial correlations for short distances. For longer distances (> 800 km), this pattern reverses and the spatial correlation of low flows become largest. This discrepancy between low and high flows suggests that hydrological systems are more homogeneously linked to climatic fluctuations under wet conditions. Under dry conditions, local catchment properties appear to play a larger role in translating climatic fluctuations into hydrological response. Copyright. (c) 2010 John Wiley & Sons, Ltd.