Assessing the future hydrological cycle in the Xinjiang Basin, China, using a multi-model ensemble and SWAT model

To account for CO2 impacts on the vegetation leaf area index (LAI) and the different responses of stomatal conductance in various vegetation species, the modified Soil and Water Assessment Tool (SWAT) model is calibrated in the Xinjiang Basin with the measured daily streamflow from 1961 to 1970, showing that SWAT accurately reproduces monthly and annual streamflow from 1971 to 2005 and actual evapotranspiration (AET) from 1983 to 2005. AET and streamflow in the basin for the 21st century are assessed using a multi-model ensemble approach in which downscaled and bias corrected outputs are compared to the baseline (1971-1999) and used to drive the SWAT model. The ensemble average annual precipitation decreases in the three scenarios, with maximum decreases of 7.61% during 2011-2050 and 4.80% during 2051-2090 in scenario A2. In contrast, the ensemble average annual and monthly temperature (minimum and maximum) increase by different amounts during the two periods in each scenario. For each emissions scenario, the ensemble average annual AET will increase, while the average annual streamflow will decrease. Under the A2 scenario, the maximum changes of average annual AET (streamflow) are +18.31% (-20.93%) and +20.93% (-23.43%) during 2011-2050 and 2051-2090, respectively. Compared to the baseline, the ensemble average annual minimum streamflow during 2011-2050 will decrease by approximately 20% under the A1B and A2 scenarios and by 7% in the B1 scenario. Thus, the ensemble average annual maximum streamflow will decrease under each scenario during the two periods, except for a 6.42% increase in the A1B scenario during 2051-2090. A Generalized Extreme Value (GEV) analysis, simultaneously, suggests that the maximum annual streamflow will decrease during the two periods under each scenario, with exception during 2051-2090 in the scenario A1B, and the magnitudes will decrease for 2-, 5-, 10-, 20- and 50-year return periods.