A comparison of the convection‐dispersion equation and transfer function model for predicting chloride leaching through an undisturbed, structured clay soil

Abstract

SUMMARYThe transfer function mode) (TFM) and convection‐dispersion equation (CDE) were compared for predicting Cl − transport through a calcareous pelosol during steady, nearsaturated water flow. Large, undisturbed soil cores were used at constant irrigation intensities (q0) between 0.3 and 3 cm h−1, with a step‐change in Cl− concentration. The assumption of a lognormal distribution of travel times–characterized by the mean (μ) and variance (σ2)–permitted the flux‐averaged breakthrough curves (BTCs) to be modelled very accurately by the TFM. The BTCs could be modelled equally well by the CDE when both the mean pore water velocity (v) and dispersion coefficient (D) were optimized simultaneously by the method of least squares, but not when v was put equal to q0/v, where V was the mean volumetric water content.The best estimate of v was consistently > q0/v, which suggested that not all the pore water was effective in chloride transport. An operationally defined transport volume (θst) was calculated from the mean () or median (τm) travel times derived from the TFM. Chloride exclusion was not solely responsible for θst() being <V: immobile water also contributed. The positive skewness of the travel time distributions meant that θst(τm) < θst(), indicating the effectiveness of macropore flow in solute transport. Dαv1.42 (from the CDE), and σ2αv (from the TFM), confirmed that Cl− dispersion increased as flow velocity increased.Flux‐averaged concentrations were used to calculate the volume‐averaged resident concentrations. They matched the measured Cl− concentrations most closely when there was a gradual decrease in measured Cl − concentration with depth, but not when Cl − decreased sharply below c. 10 cm. Calculations assuming that all the water was effective in chloride transport gave less accurate results. Comparison of the measured and predicted concentrations of solute demonstrated that this must be a critical part of the evaluation of any model of solute transport.