Assessment of water balance simulations for large-scale weighing lysimeters

A challenge of soil water transport modelling is the assessment of various uncertainties resulting from input data, from parameterisation of soil hydraulic characteristics and from estimation of sink terms like plant water uptake and soil evaporation. The objective of this study is to evaluate different modelling approaches for the estimation of soil hydraulic characteristics and evapotranspiration. A dataset from a lysimeter study in South-Germany with rotative crop vegetations over 5 years was used to perform the analysis. The pedotransfer functions that were used to estimate parameters for the representation of soil water retention and hydraulic conductivity have shown to be appropriate for water flow simulations. Although the simulated annual percolation amount is not very sensitive to the used soil hydraulic characteristics for the chosen boundary conditions, the diurnal percolation dynamics are highly sensitive to the soil hydraulic characteristics. The results show a strong dependence of simulated percolation on the used potential evpotranspiration (ET_p) model. The Penman, Penman-Monteith and Haude approach overestimate the measured cumulative actual evapotranspiration in the present study. The measured outflow from the lysimeters can be correctly simulated, if a modified Haude approach is used. However, in the simulation of daily evapotranspiration fluxes the physically based Penman-Monteith approach has higher correlation with measurements than the empirical approach of Haude. For the tested lysimeter data we show that depending on ET_p model choice the simulated percolation amounts vary between 52% and 126% of the measured amounts. Compared to this, the influence of the parameterisation of the soil hydraulic characteristics is small with a variation of up to 5% of the measured outflow.