TY - JOUR A1 - Brunner, Manuela I. A1 - Gilleland, Eric T1 - Complex High‐ and Low‐Flow Networks Differ in Their Spatial Correlation Characteristics, Drivers, and Changes Y1 - 2021-09-15 VL - 57 IS - 9 JF - Water Resources Research DO - 10.1029/2021WR030049 PB - N2 - Hydrologic extremes such as floods and droughts are often spatially related, which increases management challenges and potential impacts. However, these spatial relationships in high and low flows are often overlooked in risk assessments and we know little about their differences and origins. Here, we ask how spatial relationships of both types of hydrologic extremes and their potential hydro‐meteorological drivers differ and vary by season. We propose lagged upper‐ and lower‐tail correlation as a measure of extremal dependence for temporally ordered events to build complex networks of high and low flows. We compare complex networks of overall, low and high flows, determine hydro‐meteorological drivers of these networks, and map past changes in spatial relationships using a large‐sample data set in Central Europe. Our network comparison shows that low flows are correlated more strongly and over longer distances than high flows and high‐ and low‐flow networks are strongest in spring and weakest in summer. Our driver analysis shows that high‐flow dependence is most strongly governed by precipitation in winter and evapotranspiration in summer while low‐flow dependence is most strongly governed by snowmelt in winter and evapotranspiration in fall. Finally, our change analysis shows that changes in connectedness (i.e., the number of catchments a catchments shows strong flow correlations with) vary spatially and are mostly positive for high flows. We conclude that spatial flow correlations are considerable for both high and particularly low flows as a result of a combination of spatially related hydro‐meteorological drivers whose importance varies by extreme type and season. N2 - Plain Language Summary: Droughts and floods can happen in multiple locations at once with important implications for flood and drought risk. Still, the spatial relationships between events and the reasons for them are not well studied. Here, we therefore ask how spatial relationships of both types of extremes and their meteorological drivers differ and vary by season. We compare networks of overall, low and high flows, determine hydro‐meteorological drivers of these networks, and map past changes in flow dependence using a large‐sample data set in Central Europe. Our network comparison shows that low flows are correlated more strongly and over longer distances than high flows and both high‐ and low‐flow networks are strongest in spring and weakest in summer. Our driver analysis shows that high‐flow dependence is governed by precipitation dependence in winter and evapotranspiration dependence in summer and fall while low flow dependence is most strongly governed by snowmelt in winter and evapotranspiration and snowmelt in fall. Finally, our change analysis shows that changes in connectedness (i.e., the number of catchments a catchments shows strong flow correlations with) vary spatially and are mostly positive for high flows. We conclude that spatial flow correlation is considerable for both high and particularly low flows highlighting the need to consider it in risk assessments. N2 - Key Points: We propose and use a tail dependence measure to map and compare complex networks of high and low flows in Central Europe at a seasonal scale. Low flows are related more strongly and over longer distances than high flows and relationships are strong in spring and weak in summer. Seasonal flow correlation is shaped by spatial dependence in drivers with varying importance of precipitation, evaporation, and snowmelt. UR - http://resolver.sub.uni-goettingen.de/purl?gldocs-11858/9792 ER -