Attribution of streamflow changes across the globe based on the Budyko framework

Differentiating and clarifying the driving factors behind streamflow changes are critical for highlighting hydrological responses to changing environments. However, due to the limited number of hydrological stations, the dominant factor controlling global observed streamflow change remains unclear and intensely debated. Here, we revisit this scientific issue by using the most comprehensive dataset to attribute the observed global streamflow changes during 1960-2014. The results suggest that other factors than precipitation (P) and potential evaporation (E-0) are the most important contributors to global observed streamflow changes, which dominate streamflow change for 48.9-50.9% of the stations. In contrast, the dominant factor translated into P in 72.3-72.9% of stations when using reconstructed streamflow datasets, in agreement with most previous global assessments. These differences indicate that streamflow attributions using reconstructed streamflow might overestimate the effects of P while underestimating the roles of other factors, such as the vegetation and human impact. At the global scale, the other factors affected by many catchment characteristics and their impacts on streamflow change have remarkable regional differences. This study highlights the necessity to apply the observed data in streamflow attribution to avoid biased conclusions re-garding the dominant factor of streamflow changes. (C) 2021 Elsevier B.V. All rights reserved.

Automated summary

This study suggests that factors other than precipitation and potential evaporation are important contributors to observed global streamflow changes, dominating at 48.9-50.9% of stations. Using reconstructed streamflow datasets, precipitation becomes the dominant factor in 72.3-72.9% of stations, indicating potential overestimation of precipitation effects in streamflow attributions. Global streamflow changes are influenced by various factors with regional differences, emphasizing the importance of using observed data to avoid biased conclusions.