Determination of the saturated hydraulic conductivity, Ks, and the water retention curve, θ(h), is of paramount importance to characterize the hydraulic behavior of the vadose zone. Given the van Genuchten hydraulic model, defined by the residual, θr, and saturated, θs, volumetric water content and the α and n parameters, this work presents a new laboratory procedure to estimate Ks, θs, n and α for a drainage process, αdr, from the inverse analysis of successive drainage steady-states curves generated by a tension-gradient between the surface and the base of a soil column. To this end, a double disc system, one connected a bubbling tower and placed at the soil surface and the second one placed under the soil core, was employed. The second disc was connected to an air-vacuum system. The experiment presented two parts: a first 1D downward infiltration at saturation on a dry soil column, followed by successive drainage steps. During the drainage process, the tension of the upper and lower discs varied between 0 and −5 cm, and from −5 to −100 cm, respectively. The soil sorptivity, S, and θs were calculated from the 1D transient infiltration measure, Ks was calculated by Darcy’s law, αdr and n were optimized from the inverse analysis of the steady-state curves under tension-gradient and α for a wetting process, αw, was calculated from previously obtained S, θs, Ks and n. Once Ks estimated, αdr and n were optimized by minimizing the Q=hb-hn objective function, where hb and hn are the experimental and calculated tensions at the base of the soil core. Given a αdr value, the optimimum n was computed as the value that provides a minimum Q. By repeating this process for a sequence of αdr, different Q-isolines were obtained, one for each hb value, which crossing-point corresponded to the actual αdr and n values. The method was tested on 2.5 cm high columns of four different synthetic soils. Next, it was applied on an experimental sand column of 5 cm height and on 2.5 cm high columns filled with sieved loam, clay loam and clay soil. The estimated αdr and n were compared with corresponding values measured in the same soils with the pressure plate technique and αw was contrasted with the corresponding value calculated with an empirical hysteresis model. The method, which was fast (from 1 to 2 h) and easy to implement for small-scale experiments, was successfully applied to soil samples 2.5 cm high and allowed to explore a range of soil tensions from 0 to −100 cm. Overall, accurate estimates of θs, Ks, αdr and n were obtained in both synthetic and experimental soils. A significant relationship was also obtained between αw estimated from S and the corresponding value calculated from the hysteresis model.
Read full abstract