Stochastic analysis of a field-scale unsaturated transport experiment

Modelling of field-scale transport of chemicals is of deep interest to public as well as private sectors, and it represents an area of active theoretical research in many environmentally-based disciplines. However, the experimental data needed to validate field-scale transport models are very limited due to the numerous logistic difficulties that one faces out. In the present paper, the migration of a tracer (Cl−) was monitored during its movement in the unsaturated zone beneath the surface of 8 m×50 m sandy soil. Under flux-controlled, steady-state water flow (Jw=10 mm/day) was achieved by bidaily sprinkler irrigation. A pulse of 105 g/m2 KCl was applied uniformly to the surface, and subsequently leached downward by the same (chloride-free) flux Jw over the successive two months. Chloride concentration monitoring was carried out in seven measurement campaigns (each one corresponding to a given time) along seven (parallel) transects. The mass recovery was near 100%, therefore underlining the very good-quality of the concentration data-set. The chloride concentrations are used to test two field-scale models of unsaturated transport: (i) the Advection-Dispersion Equation (ADE), which models transport far from the zone of solute entry, and (ii) the Stochastic-Convective Log-normal (CLT) transfer function model, which instead accounts for transport near the release zone. Both the models provided an excellent representation of the solute spreading at z>0.45 m (being z=0.45 m the calibration depth). As a consequence, by the depth z≈50 cm one can regard transport as Fickian. The ADE model dramatically underestimates solute spreading at shallow depths. This is due to the boundary effects which are not captured by the ADE. The CLT model appears to be a more robust tool to mimic transport at every depth