Inverse simulation of non-steady-state evaporation using nonequilibrium water retention data: a case study

The aim of this investigation was to check the capability of the Mualem/van Genuchten model to predict unsaturated soil hydraulic conductivity values from water retention data in the absence of spatial variability and effects of macropores. Furthermore, different techniques of parameter estimation were tested. Soil cores of 250 cm3 were filled with a repacked silty loam soil. After measuring the saturated hydraulic conductivity and equilibrium soil water retention, the soil cores were subject to non-steady-state evaporation (bottom sealed). Pressure head and soil mass were recorded every 60 min. Based upon the van Genuchten/Mualem model and the Darcy–Buckingham equation, an inverse simulation was carried out. As a first option, the parameters θr, θs, α, and n were fitted to water retention data. The remaining parameters L (coefficient of tortuosity) and Ks were fitted to pressure head versus time recordings during evaporation. Equilibrium soil water retention data yielded a poor fit. The observed pressure head values during evaporation, h(t), could not be described sufficiently by using hydraulic functions determined from equilibrium retention values. A good fit was obtained in case nonequilibrium water retention measurements from the same experiment were used. Obviously, no equilibrium between pressure head and water content was established during evaporation under atmospheric conditions leading to an evaporation rate of about 3 mm day−1. As a second option, all the parameters were fitted simultaneously to both nonequilibrium soil water content data and tensiometer readings during evaporation. Simultaneous fitting yielded slightly decreased errors of simulated h(t) values and slightly increased errors of calculated soil water content. Another simulation model was used to calculate vapor flow in soil during evaporation. Results indicate that vapor flow may be significant in the range h<−500 cm. Before this range was reached, vapor flow could be neglected. Based on the results, we would like to recommend measuring transient water retention under conditions similar to those which are to be predicted in the field. Not only because of hysteresis but also because of transient effects, water retention characteristics of infiltration and evaporation should be distinguished.