Investigation into preferential flow in natural unsaturated soils with field multiple-tracer infiltration experiments and the active region model

• Field multiple-tracer infiltration experiments were conducted in natural soils.
• Effects of macropores and wetting front instability on preferential flow were studied.
• Position of macropores was studied according to the observations and predictions.
• Macropores transported the sequentially applied mixture solutions differently.
• Efficiency of active region model in predicting preferential flow was evaluated.

Preferential flow in natural unsaturated soils is common, but difficult to characterize and predict. The major objective of this research is to investigate the preferential flow patterns with field-scale multiple-tracer infiltration experiments and to evaluate the capability of the active region model (ARM) in predicting the field-scale preferential flow and transport processes. For this purpose, the mixture solutions of iodine and bromide, iodine and nitrate, and again iodine and bromide, as the tracing solutes, were applied sequentially in two plots in natural unsaturated loam soil to illustrate the flow and transport processes. The distributions of soil water content and concentrations of applied tracing solutes (NO3- and Br−) were measured after experiments and predicted using ARM and the mobile–immobile region model (MIM). The relative root mean square errors (RRMSE) between those predictions (from ARM and MIM) and measured results were calculated for quantitatively evaluating the prediction accuracy and comparing the modeling efficiency of the two models. Both field observations and the ARM predictions indicated that there were macropores in Plot 1 but not in Plot 2, and the macropores in Plot 1 were mainly in the top 20 cm soil layer. The mixture solutions transported in the top 20 cm soil layer in Plot 1 were mainly from the soil surface directly and less affected by the macropore flow, while the preferential flow in the soil layer below 20 cm was considerably affected by the macropores and more applied mixture solutions were delivered into the deep soil layer quickly. Compared to the mixture solutions applied in the first and third steps, more mixture solution applied in the second step was transported to the deep soil layer by macropores, corresponding to obvious peaks of soil water content and NO3- concentration distributions observed in the deep soil layer in Plot 1. On the other hand, unstable flow was the major preferential flow behavior in Plot 2, inducing no obvious peaks of soil water content and solutes (NO3- and Br−) concentrations observed in the infiltrated soil profile. The comparisons between predicted and observed results in Plot 2 indicated that the ARM captured the overall behavior of unstable flow and associated tracer transport better than the MIM; however, to well characterize the macropore flow process, the ARM needs to be improved to include the effects of macropores.