Reducing Uncertainty in Hydrologic Model Using Water Isotope and Chloride Data for Glacierized Catchment on the Tibetan Plateau

Abstract

Coupling tracer dynamics with hydrological process simulations in models applied to glacierized catchments can reduce model uncertainty in many applications. However, research quantitatively assessing the impact of different tracer information on model uncertainty in large glacierized catchments remains limited, especially in the Tibetan Plateau (TP). This study developed a tracer-aided model to simulate the stable water isotopes (e.g., ¹⁸O), hydrochemical tracer (chloride ions, Cl⁻) and streamflow generation processes in a typical glacierized catchment on the TP. The model parameters were calibrated using various combinations of streamflow, δ¹⁸O and Cl⁻ concentrations in streamflow. We focused on the impact of δ¹⁸O and Cl⁻ concentrations on model uncertainty, compared to calibration in the absence of such tracer data. Results showed that: (1) The tracer-aided model developed in this study successfully reproduces streamflow, δ¹⁸O, and Cl⁻ concentrations, and can be applied to hydrological process simulations in the TP; (2) Using δ¹⁸O and Cl⁻ concentrations in calibration reduced model uncertainty by limiting parameter and simulation ranges, and constraining the estimates of streamflow components and recharge sources; (3) δ¹⁸O is more effective in reducing uncertainties in recharge sources, such as source-related model parameters (P_r, DDF_gla, DDF_snow) and the estimates of recharge sources, due to the significant variation in isotopic compositions among different recharge sources; (4) Cl⁻ concentrations are more effective in reducing uncertainties in streamflow components, such as streamflow-component-related model parameters (r_s, K_ss, K_g) and the estimates of streamflow components, due to the high-value characteristics exhibited by Cl⁻ concentrations in deep groundwater and the differences in Cl⁻ concentrations in river water during high and low flow periods; (5) Using both δ¹⁸O and Cl⁻ concentrations in our calibration method proved to be the most effective in reducing model uncertainty. Our findings highlight the importance of integrating stable isotopes and hydrochemical properties to reduce uncertainty in hydrological modelling of glacierized catchments on the TP.