Determining the depth of mixing layer in which soil solute releasing from soil to surface runoff on the unsaturated loess slope under artificial rainfall condition

Purpose The mixing zone depth is a critical parameter in many popular agricultural non-point source pollution (NPS) models. Its variation for some special soil conditions and easily erosive region was limitedly studied. This study dealt with the extent of Br − releasing from soil to surface runoff, the actual mixing depth and the calculated one on the unsaturated loess slope, and the influences of soil initial moistures. Materials and methods Br − was placed as a tracer at different soil depths in the loess slope with three soil initial moistures. Br − concentration in surface runoff and soil profiles was monitored under simulated rainfall condition. The actual mixing layer depth was determined by whether Br − in that layer could be detected in surface runoff during the whole rainfall process. The average effective depth of interaction (EDI) was calculated by the confirmation method reported before. Results and discussion The shallower the application depth, the higher the Br − concentration and loss risk in surface runoff. The net loss rate had an exponential decreasing correlation with the solute-applied depth. The depth of Br − peak content in soil profile after the rainfall ended increased with the application depth. Br − concentration in the surface soil was more than that in the runoff at the end of rainfall, which indicated an incomplete mixing between runoff water and soil water in the mixing layer. Under this experimental condition, the actual mixing layer depth for 3.5% soil moisture slope was about 3 cm, that for 10% soil moisture was 6 cm or so, and that for 20% soil moisture was more than 7 cm. The average EDI calculated for 3.5, 10, and 20% soil moisture was 2.31, 3.76 and 4.95 cm, respectively, which were less than the actual mixing layer depth and more than some former studies. Conclusions The traditional 1.0-cm mixing depth in many agricultural NPS models might not be so exact for some special region.