Archetypal flow regime change classes as signatures of anthropogenic drivers of global streamflow alterations

Streamflow – a key component of the water cycle – is experiencing drastic alteration due to human actions. While existing studies have widely assessed the global extent and degree of this change, understanding of its drivers has been limited because previous global-scale approaches have largely relied on modelled hypothetical scenarios. Here, we overcome these limitations by providing a systematic association analysis of streamflow change and its drivers. We use observed streamflow data in 5,163 catchments globally and combine them with data on precipitation, evapotranspiration, water use, and damming. Building on a robust annual trend...  more

Streamflow – a key component of the water cycle – is experiencing drastic alteration due to human actions. While existing studies have widely assessed the global extent and degree of this change, understanding of its drivers has been limited because previous global-scale approaches have largely relied on modelled hypothetical scenarios. Here, we overcome these limitations by providing a systematic association analysis of streamflow change and its drivers. We use observed streamflow data in 5,163 catchments globally and combine them with data on precipitation, evapotranspiration, water use, and damming. Building on a robust annual trend analysis covering years 1971–2010, we first determine archetypal flow regime change (FRC) classes, and then use them to investigate associations between streamflow change and its drivers. We find that 89% of all catchments are assigned to four main FRCs, which indicates globally consistent flow regime changes. By associating driver trends with the FRCs, we further characterise them by trends and changes in the four investigated drivers. We find that FRCs depicting decreasing streamflow quantity and variability are strongly associated with direct human drivers, either from water use or damming. In contrast, indirect drivers (precipitation and evapotranspiration) are more dominant in FRCs that depict increasing streamflow quantity and variability. Our observation-based association analysis substantiates the findings of existing model-based studies and can thus add detail and validation to their interpretations. This may support developing and adopting efficient measures to mitigate streamflow change and its subsequent impacts across scales.