Abstract
Abstract. Baseflow is an important component in hydrological modeling. This process is usually modeled by using the linear aquifer storage–discharge relation approach, although the outflow from groundwater aquifers is nonlinear. To identify the accuracy of baseflow estimates in rivers dominated by snow and/or glacier melt in arid and cold northwestern China, a nonlinear storage–discharge relationship for use in SWAT (Soil Water Assessment Tools) modeling was developed and applied to the Manas River basin in the Tianshan Mountains. Linear reservoir models and a digital filter program were used for comparisons. Meanwhile, numerical analysis of flow recession curves from 78 river gauge stations revealed variation in the coefficients of the nonlinear relationship. It was found that the nonlinear reservoir model can improve the streamflow simulation, especially for low-flows. The highest Nash–Sutcliff efficiency and lowest Percent Bias were obtained when compared to the one- or two-linear reservoir approach. The exponent b of the aquifer storage–discharge function varied mostly between 0.0 and 0.1, which is much smaller than the suggested value of 0.5. The coefficient a of the function is related to catchment properties, primarily the basin and glacier areas.