Introduction:

Tracing the reactive material through soil is crucial for the health of the environment, water resources and agricultural products. One of the widespread elements in agricultural lands is sodium, which affects on soil physical structure and water movement and solute transport in soil. Therefore, it is important to trace and remediate the sodium transport through the soil and know the fate of this element in the root zone. Many factors can affect the transport of an ion through porous media, in which soil moisture condition and application of amendment such as vermicompost are very effective due to altering the hydraulic and physical conditions of the soil. Therefore, the aim of the current study is to model the sodium transport in natural soil and soil amended with vermicompost under saturated and near-saturation conditions with three models of equilibrium, one-site sorption and two-site sorption. Moreover, sensitivity analysis was employed to explore the effective parameters of the breakthrough curve of sodium.



Methods:

Two media including natural soil with loamy sand texture and soil amended with vermicompost were prepared, in which amended soil was achieved by mixing 1.45 gr vermicompost with 100 gr natural soil. In order to conduct the experiment, 350 gr of each soil sample was placed into the columns with a length and a diameter of 10 and 5.95 cm, and then all soil columns were saturated for 24 hrs and leached with distilled water for another 24hrs. After leaching, a solution containing 1mM KNO3 and 0.435mM NaCl was applied with flow rates of 3.17-4.28 and 2.5 cm3/min for 270 minutes under saturated and near-saturated conditions (water saturation (Sw)=1 and 0.98), and soils were leached with distilled water again under aforementioned conditions during 300 minutes. The effluent solutions were collected and their Na contents were measured using flame photometry. Next, Hydrus-1D program was employed to simulate the experimental breakthrough curves of sodium using Advective-Dispersive equation coupled with equilibrium, one-site sorption and two-site sorption models. The unknown coefficients were KD for equilibrium model, KD and α for one-site sorption model and KD, F and α for two-site sorption model. The sensitivity analysis was performed for all unknown coefficients of the models to clarify the role of each coefficient on the simulated curve.



Results:

The values of RMSE for equilibrium, one-site sorption and two-site sorption models were respectively 0.02-0.19, 0.003-0.03 and 0.003-0.03, which indicates equilibrium model has an unacceptable accuracy compared to the one-site and two-site sorption models, and two-site model has the highest accuracy. Equilibrium model had poor performance at all parts of all curves, while the poor performance of the other models occurred at the breakthrough point and slop of some curves. The relatively similar accuracy of one-site model with two-site model and the fraction of sorption sites (F)<0.5 in two-site model indicated that the transport of sodium in the both soils was mainly controlled by the sorption/desorption kinetic process. The application of vermicompost and desaturation from Sw=1 to 0.98 decreased the values of F to 66.5 and 85.5% respectively, indicating that the kinetic process was more predominant in amended soil and unsaturated conditions. The values of distribution coefficient (KD) were altered by changing the conditions in all models. The values of KD increased by the application of vermicompost and soil desaturation to 3.4 and 12.6%, respectively. It shows that the sorption capacity of amended soil is higher than that of un-amended soil, and it is increased under unsaturation conditions compared to saturated conditions due to the presence of air bubbles in soil and a delay in water and solute transport. Employing the fraction of sorption sites in two-site model compared to on-site model changed the values of first-order rate coefficient (α) from 0.01-0.032 to 0.005-0.016 min-1 leading to a reduction by 47%. The values of α were decreased by a reduction in flow rate due to the lower sodium diffusion. The results of sensitivity analysis showed that KD had the most effect among the coefficients on fitting the breakthrough curve, and altered all parts of the curves including the breakthrough and peak points and the slope. The one-site model had low sensitivity to α, which its variation led to a small change around the peak point. The sensitivity of two-site model to the both coefficients of F and α was moderate, by which the curve had a variation on the peak point and slope. Overall, the sensitivity of coefficients of two-site model showed the trend of KD>F>α.