How Dynamic Non-Equilibrium Water Flow Affects Estimates of the Water Retention Curve - a Modeling Approach.

The water retention curve is the most fundamental curve in soil physics. It relates the water content of a porous medium to its energy status by assuming equilibrium within a representative elementary volume. However, since the 1967 and the pioneering work of Topp et al. (1967), experimental observations have shown that the water retention curve estimated under static and dynamic flow conditions can differ considerably. This phenomenon is often named “dynamic non-equilibrium effect” and it is often observed in transient laboratory experiments designed for the estimation of the soil hydraulic properties (SHPs). In the case of experiments with controlled pressure head boundary conditions such as the Multistep Outflow (MSO) method, non-equilibrium effects appear as a relaxation in the cumulative outflow (system-averaged water content) while the pressure head in the soil column indicates hydrostatic equilibrium. For experiments with flux boundary conditions like the Multistep Flux (MSF) method, non-equilibrium effects appear as a relaxation of the pressure head while the flux density and macroscopic water content distribution appear static. These effects influence strongly the estimated SHPs and the use of them for modelling purposes is questionable. We propose a numerical model to quantitatively describe dynamic non-equilibrium effects during unsaturated water flow. The model considers two continua at the macroscopic scale: one continuum is described by the Richards equation and the second, associated with non-equilibrium water flow, assumes a time dependent equilibration of the water content as described in the approach of Ross and Smettem (2000). The new model was fitted to experimental data for the same soil column but for two different experimental types: MSO and MSF. Results confirm that the model describes the observed dynamic non-equilibrium effects very well for both MSO and MSF experiments. More importantly, the values of the estimated non-equilibrium parameters for both experimental types were almost identical.