Evaluation of a functional model for simulating boron transport in soil

There has been renewed interest in the application of functional models to the transport of nonpoint source pollutants at polypedon (i.e., farm) and watershed scales because of the ease of their coupling to a geographic information system and to the accepted organizational hierarchy of pedogenetic modeling approaches. However, very little work has been done to evaluate the performance of a functional transient-state model for the transport of a reactive solute over an extensive study period. Subsequently, the functional model TETrans (Trace Element Transport) was evaluated for model performance with boron (B) transport data collected from a meso-scale soil lysimeter column over a 1000-day study period (i.e., 40 irrigations). Because the ability to simulate water now has been evaluated previously for TETrans, the focus of this evaluation centered around the performance of various functional models of B adsorption used as subroutines within the TETrans model, including the (1) Freundlich, (2) kinetic Freundlich, (3) Langmuir, (4) temperature-dependent Langmuir, and (5) pH-dependent Keren adsorption isotherm equations. Model performance was evaluated with statistical functions, specifically the Average Absolute Prediction Error, the Root Mean Square Error, the Reduced Error Estimate and the Coefficient of Residual Mass, and graphic displays of observed and predicted B concentration profiles. Even though no single adsorption isotherm equation, when coupled to TETrans, could be considered poor in its performance, results indicated that the order of model performance was the pH-dependent Keren equation first, followed by the temperature-dependent Langmuir and kinetic Freundlich equations, the Freundlich equation, and, finally, the Langmuir equation. Overall, the TETrans model was able to simulate the transport of B with deviations because no functional adsorption equation incorporated all the influences of pH, ionic strength, temperature, and kinetic effects into a single equation. The inability to correctly predict one of the measured peaks in B concentration near the soil surface suggests that problems with the timing of the sample collection may have occurred for the shallowest sampling depth.