Full-Range Soil Hydraulic Properties from Numerical Inversion of Transient Evaporation Experiments.

The assessment of the soil water balance in semi-arid regions requires an accurate quantification of bare soil evaporation. Actual evaporation from dry soil cannot be predicted without detailed knowledge of the complex interplay between liquid, vapor and heat flow. Soil hydraulic properties exert a strong influence on evaporation rates during stage-two evaporation. Recently, significant progress has been made in the experimental characterization of soil water retention in dry soils, but the determination of unsaturated hydraulic conductivity in medium to dry soils remains a challenge. We conducted laboratory evaporation experiments on large packed soil columns using different soil materials. Water potential was monitored with tensiometers and relative humidity sensors, and evaporation rates were determined gravimetrically. Multiple experiments were performed on the same system with different atmospheric forcings. Data from the most isothermal experiments were used to determine soil hydraulic properties by inverse modeling with the Richards equation. Results clearly demonstrate that hydraulic conductivity cannot be properly parameterized if water flow is conceptualized exclusively as movement of water through water-filled capillaries. The additional consideration of isothermal vapor flow and liquid flow in thin films and pore corners by an effective approach lead to an excellent match with the experimental data. To validate the adequacy of the identified properties, we used them for flow predictions under atmospheric boundary conditions which were different from those used for identifying the soil hydraulic properties. A fully coupled numerical model of water, vapor, and heat flow in soil which considered the surface energy balance and temperature effects on the transport coefficients was applied to predict the system behavior for varying wind speeds and amounts of incoming shortwave radiation. The evaporation rates predicted with the coupled model were in very close agreement with those observed in the experiments thus providing confidence in the identified physical system properties.