Stable isotope tracing is widely used to track water movement in the Groundwater-Soil-Plant-Atmosphere Continuum system. Physics-based modeling of soil water flow and stable isotope transport has the potential for providing continuous and real-time isotopic information. However, the evaporation fractionation effect is often ignored in current models, or they still implement only a simple treatment of evaporation fractionation. The lack of these considerations will result in simulation errors that may be propagated into practical applications.To solve these problems, we first adapted and tested the HYDRUS-1D model, a numerical model widely used to simulate variably-saturated water flow and solute transport in porous media, by including an option to simulate isotope fate and transport while accounting for evaporation fractionation. The numerical results obtained by the adapted model are in excellent agreement with existing analytical solutions. Additional plausibility tests and field evaluation further demonstrate the adapted model’s accuracy.
We then investigated the impact of considering evaporation fractionation on model performance and practical applications (travel times and evaporation estimation). The global sensitivity analysis using the Morris and Sobol' methods and the parameter estimation using the Particle Swarm Optimization algorithm show that the Kling-Gupta efficiency (KGE) index for isotope data can increase by 0.09 and 1.49 for the humid and arid datasets, respectively, when selecting suitable fractionation scenarios. Considering evaporation fractionation using the Craig-Gordon (CG) and Gonfiantini models is likely to result in older water ages than the no-fractionation scenario estimates for the humid dataset. The direct use of simulated isotopic compositions in the no-fractionation scenario may result in large biases in practical applications in the arid zone.
We further explored the impact of considering soil tension control on model performance and practical applications (spatial-temporal origin of RWU). The results show that considering soil tension control (the TC_Frac scenario) leads to a depleted surface isotopic composition compared with only considering temperature control (CG_Frac). The contribution ratios of all soil layers in the TC_Frac scenario are always between the no fractionation (Non_Frac) and CG_Frac scenarios. The order of both drainage and RWU travel times is: Non_Frac>TC_Frac>CG_Frac. All methods can reflect the overall vertical trends of contribution ratios of different soil layers to RWU, and temporal trends of drainage and RWU travel times, although absolute differences between different methods always exist. Overall, the impact of the soil temperature fractionation effect is much more important for model performance and practical applications than the impact of the soil tension fractionation effect.