Multitracer irrigation experiments for assessing the relevance of preferential flow for non-sorbing solute transport in agricultural soil

•Preferential transport paths were visualized below the plough pan in a sandy soil.•Preferential flow accelerated cumulative tracer leaching during winter by 13–30%•The relevance of preferential flow for our sandy poorly-structured arable soil was highly scale-dependent. This study addresses the relevance of preferential flow for the transport of a non-sorbing solute in a sandy agricultural field with nitrate-contaminated groundwater. Two multitracer irrigation experiments were conducted 1) in October 2017, and 2) in March 2018 on a 2 m × 2 m plot on an agricultural field in Northern Germany (Cloppenburg region). For irrigation 1, Cl-, H218O, and 15NO3- were used as tracers, for irrigation 2, Br-, 2H2O, and 15NO3- were used as tracers. Between both irrigation events the study plot was subject to natural precipitation. The study plot was instrumented with soil moisture sensors and suction plates. The occurrence of preferential flow was studied based on the following data: (1) tracer breakthrough curves collected by five suction plates; (2) the spatial distribution of resident tracer concentrations in soil at the end of the experiment; and (3) 3D-time lapse electrical resistivity tomography (ERT) during both irrigation events. The spatial distribution of water residence times at the end of the experiment was calculated based on measured tracer concentrations in the soil and the known fingerprints of three sources of water (irrigation 1, rain and irrigation 2) applied to the plot. The relevance of preferential flow for the leaching of a non-sorbing solute was assessed by comparing the observed and Hydrus-1D-modeled cumulative leaching of Cl-. In comparison to the conservative tracers 1–11% of the applied 15NO3- was removed from soil solution most probably due to microbial processes. Continuous preferential flow paths were observed below the plow pan. The same preferential flow paths seem to be used repeatedly by the successive rain/irrigation events. However, the mobile-immobile regions appear to exist only for a short period of time, because fast matrix flow balances the spatial heterogeneity in tracer distribution. The residence time of soil water varied from 1 to 3 months at 40–50 cm depth and was not older than 4 months at 190–200 cm depth. Hydrus-1D transient model calculations predicted 13%-30% less leaching of a conservative tracer during wintertime compared to the experimentally estimated averaged leaching from our study plot. Therefore, our main resu