INTERPRETING NONSTEADY STATE TRACER BREAKTHROUGH EXPERIMENTS IN SAND AND CLAY SOILS USING A DUAL-POROSITY MODEL

The effects of preferential flow on Cl-36 transport in undisturbed san d and clay soil monolith lysimeters were quantified using a dual-poros ity model (MACRO). A double tracer test with H-3 and Cl-36 was perform ed simultaneously to check the possible occurrence of sidewall flow in the lysimeters. In the dual-porosity model MACRO, simulations can be performed in both one and two flow domains. Run in one flow domain, th e model reduces to numerical solutions of Richards' equation and the c onvection-dispersion equation. In the sandy soil, the occurrence of pr eferential flow was tested by simulating in one domain, assuming that a certain pore fraction takes no part in water flow and solute transpo rt. For the clay soil, the one domain case was compared with two domai n simulations accounting for macropore flow. The double-tracer tests s howed that sidewall flow did not occur in either soil type. Simulation s of water flow showed good agreement with observed seepage until late autumn, but were less good during winter because the model does not a ccount for soil freezing and snowpack/snowmelt. Simulated water flows were similar in one and two domain simulations, presumably because wat er contents in the lysimeters were maintained close to held capacity d uring the experiment. The simulations indicated that preferential how occurred in the sandy soil, with the observed Cl-36 breakthrough curve s, assuming an unwetted volumetric pore fraction of 20%, reproduced re asonably well. The rate of Cl-36 leaching was consequently increased b y c. 25% compared with the simulation assuming no preferential flow. M acropore flow was clearly demonstrated in the clay soil. The two domai n simulation matched the breakthough curve and accumulated Cl-36 leach ing accurately when the soil water pressure head defining the boundary between pore domains was set to -50 cm. This implies that preferentia l Cl-36 transport was taking place in a wide range of pore sizes, incl uding smaller mesopores. The one domain simulation failed to predict t he pattern of breakthrough of Cl-36 in the clay soil, in that it serio usly underestimated leaching at early times and overestimated leaching towards the end of the experiment. Accounting for preferential flow w ith the dual porosity model resulted in significantly improved estimat es of solute transport, compared to the classical convective-dispersiv e treatment, for both nonstructured sands and structured clay soils.