Time Varying Water Flow Behavior and Spatial Heterogeneity in a Lysimeter - Implications for Parameter Estimation By Model Inversion.

Estimation of effective soil hydraulic parameters for characterization of the vadose zone properties is important for many applications from prediction of solute and pesticide transport to water balance modeling in small catchments. Inverse modeling has become a common approach to infer the parameters of the water retention and hydraulic conductivity functions from dynamic experiments under varying boundary conditions.

To gain further inside into to the water transport behavior of an agricultural field site with a layered, gravelly vadose zone, a lysimeter was taken and equipped with a total of 48 sensors (24 tensiometers and 24 water content probes). The sensors were arranged in 6 vertical arrays consisting of 4 sensor pairs, respectively. Outflow, pressure heads and water contents were measured in four depths in each of the arrays allowing for the estimation of the soil hydraulic properties of the three individual soil layers by inverse modeling.

This experimental design served for the investigation of two important questions: a) do effective soil hydraulic properties at the lysimeter scale exist, more specifically: can a single representative parameter set be found which describes the hydraulic behavior in each of the arrays with acceptable performance ? And b) which degree of freedom is necessary or required for an accurate description of the one dimensional water flow at each of the arrays ?

The results show a high degree of horizontal heterogeneity together with a strong preferential flow behavior within the lysimeter. The estimation of effective soil hydraulic properties for each of the arrays and the whole lysimeter is further complicated by the time varying water flow behavior. Effective soil hydraulic parameters could be obtained for each of the sensor arrays individually for individual time periods, resulting in good agreement between the model predictions and the observations for the individual soil horizons. However, no general parameter set could be identified to describe the integral water flow over all arrays with acceptable performance due to the high degree of horizontal heterogeneity within the soil horizons. The overall results challenge the applicability of one-dimensional models to estimate effective soil hydraulic properties for layered, gravelly vadose zones.