| Prediction of the climate change impact on regional hydrological system has uncertainty. For hydrological models, the uncertainty related to compensating interactions among parameters could cause equifinality problem and misrepresent the hydrologic process in a basin. The SWAT (Soil and Water Assessment Tool) contains a wide range of parameters and is especially susceptible to this problem. The parameter uncertainty of SWAT is reduced using the global optimization algorithm SUFI2 (sequential uncertainty fitting) with a multi-objective function consisting of multi-site streamflow and basin-wide water storage anomalies (from GRACE-Gravity Recovery and Climate Experiment) as well as groundwater table measurements. The multi-stage and multi-variate calibration gradually reduces the number of controlling parameters and related value ranges toward reliable and realistic model performance for different goal variables.;The newly available North American Regional Climate Change Assessment Program (NARCCAP) dynamically downscaled climate variables were used as atmospheric forcing for the highly-calibrated SWAT model which runs with different potential evapotranspiration (PET) methods to assess the hydrological change and uncertainty. The NARCCAP temperature and precipitation predictions were refined using a bias correction method (PDF/CDF equalization). The results showed that, following the seasonal variability of precipitation, various water fluxes would increase in most seasons except the summer. Expected precipitation tends to increase in intensity with little variation in frequency, triggering faster water accumulation to form floods. The greatest streamflow increase would occur from November to February, increasing by around 10% on average. Slight increases of 3% emerge in the other months except for July and August in which river discharge decreases by around 2%. Uncertainty in and differences between the climate models affect the degree of hydrological changes with annual variability related to the climate model and the amount of precipitation during wet seasons. This study suggests an even wetter environment in 2040-2069 compared to the historically wet period of 1968-1997, with the possibility of more flooding. |
