Validating coupled flow theory for bare-soil evaporation under different boundary conditions

Evaporation from bare soil is an important hydrological process and part of the water and energy balance of terrestrial systems. Modeling bare-soil evaporation is challenging, mainly due to nonlinear couplings among liquid water, water vapor, and heat fluxes. Model concepts of varying complexity have been proposed for predicting evaporative water and energy fluxes. Our aim was to test a standard model of coupled water, vapor, and heat flow in the soil using data from laboratory evaporation experiments under different boundary conditions. We conducted evaporation experiments with a sand and a silt loam soil and with three different atmospheric boundary conditions: (i) wind, (ii) wind and short-wave radiation, and (iii) wind and intermittent short-wave radiation. The packed soil columns were closed at the bottom (no water flux) and instrumented with temperature sensors, tensiometers, and relative humidity probes. We simulated the evaporation experiments with a coupled water, vapor, and heat flow model, which solves the surface energy balance and predicts the evaporation rate. The evaporation dynamics were predicted very well, in particular the onset of stage-two evaporation and the evaporation rates during the stage. A continuous slow decrease of the measured evaporation rate during stage-one could not be described with a constant aerodynamic resistance. Adding established soil resistance parametrizations to the model significantly degraded model performance. The use of a boundary-layer resistance, which takes into account the effect of point sources of moisture, improved the prediction of evaporation rates for the sandy soil, but not for the silt loam. Evaporation experiments under different atmospheric conditions were modeled with coupled water-vapor-heat flow.Model predicted evaporation dynamics well, but slightly falling rates during stage-one could not be modeled.Parameterization of soil hydraulic properties accounts for sorbed water and film flow.Unjustified empirical soil resistance parameterizations degraded model performance.Use of a boundary-layer resistance described the decrease during stage-one for sand, but not for loam.