Prediction of soil evaporation measured with weighable lysimeters using the FAO Penman-Monteith method in combination with Richards' equation

Multiannual data (2016-2018) from 12 weighed lysimeters (four soil types with textures ranging from sandy loam to silt loam, three replicates) of the TERENO SOIL-Can network were used to evaluate if evaporation (E) rates could be predicted from weather data using the FAO Penman-Monteith (PM) method combined with soil water flow simulations using the Richards equation. Soil hydraulic properties (SHPs) were estimated either from soil texture using the ROSETTA pedotransfer functions, from in situ measured water retention curves, or from soil surface water contents using inverse modeling. In all years, E was water limited and the measured evaporation rates (E-m) surprisingly did not vary significantly among the four different soil types. When SHPs derived from pedotransfer functions were used, simulated evaporation rates of the finer textured soils overestimated the measured ones considerably. Better agreement was obtained when simulations were based on in situ measured or inversely estimated SHPs. The SHPs estimated from pedotransfer functions represented unrealistically large characteristic lengths of evaporation (L-C), and L-C was found to be a useful characteristic to constrain estimates of SHPs. Also, when soil evaporation was water limited and E-m rates were below E-pot (PM evaporation scaled by an empirical coefficient), the diurnal dynamics of E-m followed those of E-pot. The Richards equation that considers only isothermal liquid water flow did not reproduce these dynamics caused by temperature dependent vapor transport in the soil.