Synergistic impacts of climate and land cover change, and water management on terrestrial hydrology over headwaters with an integrated attribution framework

Located over eastern Tibetan Plateau, the upper Yellow River basin (UYRB) provides about half of the total annual discharge of the entire Yellow River basin in northern China, and influences more than one hundred million people over downstream regions. In the Anthropocene, human activities such as greenhouse gases emission, human-induced land cover change and water management are changing the terrestrial hydrological process and streamflow extremes over UYRB. However, quantifying their separate influence is a great challenge due to limited observations and difficulty in modeling hydro-thermal processes over alpine regions.

Here we find significant fingerprints of anthropogenic climate change and land cover change in decreasing total water storage and increasing extremely low streamflow over UYRB headwater. While anthropogenic climate change, reservoir operation and land cover change significantly decreasing the probability of extreme flooding event over the UYRB by 31%, 45% and 10% respectively. The newly-developed Conjunctive Surface-Subsurface Process version 2 (CSSPv2) land surface model was first implemented at a high resolution (3km) over the UYRB. Comprehensive evaluations show the model well captures the variation and variability of hydrological variables. Simulations with and without land cover change were then compared to assess the impact of land cover change, while reservoir influence was calculated by comparing the modeled naturalized streamflow with observed streamflow. CSSPv2 was also driven by CMIP5 outputs with natural or anthropogenic forcings to assess influence of anthropogenic climate change. An integrated hydro-climate attribution framework was finally used to unify the contributions of different factors. Our results highlight the local-scale human influences (including land cover change and water management) on the streamflow extremes, which are still not well incorporated in current global climate models for detection and attribution studies.