Assessing Isotopic Mixing Assumptions in Tracer‐Aided Hydrological Models: Implications for Hydrological Partitioning and Transit Time Distribution

ABSTRACT Tracer‐aided hydrological models provide valuable insights into water source contributions and transit time dynamics, yet their interpretations are strongly conditioned by mixing assumptions. In this study, we advanced the Two‐Reservoir StorAge Selection (TRSAS) model by incorporating three alternative isotopic mixing scenarios—complete mixing (TRSAS‐CM), partial mixing (TRSAS‐PM) and no mixing (TRSAS‐NM)—and applied it to a humid hilly watershed in eastern China. Mixing assumptions were applied only to the upper reservoir, while the lower reservoir was kept fully mixed, and all models were calibrated solely to discharge and streamflow δ 18 O. All models reproduced discharge and streamflow δ 18 O with satisfactory accuracy; however, their skill varied across observed variables. TRSAS‐NM achieved the best performance for calibration targets, while TRSAS‐PM better captured non‐calibration signals such as soil and groundwater isotopes dynamics. Mixing assumptions exerted strong impacts on hydrological partitioning; specifically, the absolute difference in simulated fast flow contributions among the three mixing scenarios reached up to 0.18 (ranging from 0.33 to 0.51). Seasonal contrasts were most pronounced under TRSAS‐CM and most stable under TRSAS‐NM. The models also diverged in transit time of fast flow in wet condition: TRSAS‐CM simulated the lowest mean event water fraction in fast flow component (0.19), TRSAS‐PM intermediate (0.23) and TRSAS‐NM the highest (0.31). Notably, the TRSAS‐PM scenario, which most closely reflects the watershed's hydrological behaviour, simulated event water fraction in fast flow reaching up to 0.6 during some wet‐period episodes. These findings demonstrate that isotopic mixing exerts a first‐order control on tracer‐based estimates of flow partitioning and transit time distributions, underscoring the importance of carefully selecting mixing assumptions to avoid biased or misleading interpretations.