Longer simulation time step of the tracer-aided hydrological model estimates lower contribution of slow runoff components

Abstract

Uncertainty and parameter equifinality are important issues for hydrological models, especially in cold mountainous regions where multiple runoff components enhance the hydrological complexity. Tracer-aided hydrological models integrating water isotope modules are helpful for improving the physical representation of the model structure and parameter identification. Under the common complete-mixing assumption in the distributed tracer-aided model, the simulation time step reflects the water movement velocity at the hydrological unit scale and influences the model performance, which has yet to be addressed in previous studies. Consequently, the question of how the simulation time step influences the performance of tracer-aided hydrological models is poorly understood. In this study, we utilized a tracer-aided model THREW-T for a mountainous basin on the Tibetan Plateau to explore the above questions. The results showed that (1) the simulated variations in streamflow, snow cover area and stream water isotopes were similar among different time scales, but the isotope simulation performance worsened when adopting longer time steps. Simulation time steps ranging from 1 to 24 h led to a 10% difference in the estimated contribution of subsurface runoff, which decreased with increasing time scale. (2) A longer simulation time step yielded a lower simulated variability of stream water isotopes because of the lower water velocity within hydrological units. To offset this effect, the model characterized quicker water movement at the catchment scale by estimating a lower contribution of the slow runoff component. (3) Despite the additional uncertainty resulting from the simulation time step, the involvement of isotopes in the calibration objective still improved the model confidence by increasing the parameter sensitivity to the partitioning among the different runoff components, resulting in more reasonable water apportionment estimates. This study addresses the influence of the time scale on the performance of tracer-aided hydrological models and highlights the importance of adopting a suitable time step when applying these models, rather than the daily scale commonly adopted in most existing modeling practices.