7m–p and Table 7). Although we analyzed five hydrological components (e.g., total buy Nintedanib water yield, soil water content, ET, streamflow, and groundwater recharge) simulated in the SWAT model, the model was calibrated and validated using only one component – streamflow. Therefore, predicted estimates of those components that were not calibrated were more uncertain. However,

ET estimates were validated qualitatively with the estimates from the Joint UK Land and Environment Simulator (JULES) model provided by the European Union WATer and Global Change (WATCH) project. Additional uncertainties could also be contributed from (1) uncertainties in the future climate conditions and emission scenarios, Obeticholic Acid nmr (2) errors in GCM predictors, (3) errors in the downscaling of precipitation in SDSM, and (4) errors in the SWAT model. While quantifying many of these uncertainties is often challenging, the interpretation of model results requires consideration of these uncertainties. Analyzing the sources of errors in the projected climate conditions, emission scenarios, and GCM predictor variables was beyond the

scope of this study. The uncertainties in the downscaled precipitation used in this study were generated in our earlier work (Pervez and Henebry, 2014). In brief, the bias in the raw CGCM3.1 precipitation was substantially reduced in the downscaled CGCM3.1 precipitation. There were estimated ±29% and ±28% uncertainties in the downscaled CGCM3.1 precipitation for the A1B and A2 scenarios, respectively (Pervez and Henebry, 2014). It is no surprise that these uncertainties associated

with downscaled precipitation will propagate to the uncertainty Clostridium perfringens alpha toxin of SWAT-simulated hydrological components. Even though uncertainty in the downscaled precipitation was attenuated, the propagated uncertainty in simulated hydrological components because of the uncertainty in the downscaled precipitation is largely unknown. Furthermore, the projected downscaled precipitation may not be accurate at some future time, because the model developed for the downscaling may not adequately capture the changed environmental conditions in a future climate. As a distributed hydrological model, SWAT is subject to large uncertainties (Rostamian et al., 2008). SUFI2 is one of the uncertainty analysis techniques integrated into SWAT that enables users to quantify model errors more systematically while calibrating the model. We used SUFI2 and discussed the model uncertainties in Sections 3.3 and 5.1. The model performance metrics suggested that the SWAT model calibration and validation was satisfactory at the monthly scale, but there were substantial differences between observed and simulated peak streamflow at the daily scale. The high intensity localized precipitation might not have been well represented by the limited number of precipitation stations used in the study.