The timing and magnitude of snowmelt are changing due to global warming, posing significant implications for irrigated agricultural areas reliant on snowmelt streamflow. By taking the Shiyang River Basin (SRB) in Northwest China as the study area, this study focuses on accurately quantifying the snowmelt derived streamflow and its variation under the impacts of future climate change. Based on the observation data, we modified the snowmelt factor seasonal variation formula and streamflow module in the source code of the Soil and Water Assessment Tool (SWAT) to improve the simulation capability. Results show that the original SWAT model tends to overestimate streamflow, whereas the enhanced model (SWAT-s) aligns more closely with measured streamflow during both calibration and validation periods. SWAT-s can better capture the monthly streamflow peak and improve the snowmelt streamflow simulation performance. The results indicate the SWAT-s outperforms SWAT and SWAT-C in terms of streamflow simulation, especially in spring. Furthermore, we integrate SWAT with the downscaling method (EDCDFm) and the 6th International Coupled Model Comparison Program (CMIP6) under various climate change scenarios (SSP1-2.6, SSP2-4.5, SSP3-7.0, and SSP5-8.5) to investigate the influence of climate change on future streamflow, evapotranspiration, snowmelt, and soil water content. Compared with the baseline period (1985-2014), our findings project a decrease in streamflow in the upper reaches of the SRB during the future period (2023-2100), while the ensemble average snowmelt streamflow in spring is projected to increase. Evapotranspiration of all four rivers showed an increasing trend with time (P< 0.05), while snowmelt and soil water content showed a decreasing trend (P < 0.05) under the SSP5-8.5 scenario. The study provides the scientific basis for developing streamflow management plans to ensure the sustainable development of basins in arid areas of Northwest China.