Abstract
This paper investigates the treatment of the soil water budget of two parametric models currently used in atmospheric models for climate studies. Because the parametric models are intended to represent areally averaged behavior, results of the water budget for both models are compared to output of a well-tested physically based, one-dimensional unsaturated flow model with spatially heterogeneous soil hydraulic properties. Computations are performed for the three models using two datasets of soil hydraulic properties and three separate years of daily average meteorological conditions. Neglecting the percolation process in land-surface parameterizations can lead to very unrealistic results in evapotranspiration estimates. Evapotranspiration efficiencies of the parametric models show more rapid time fluctuations compared to the physically based model. Furthermore, it appears that a selected reference set of soil hydraulic properties behaves similar to the areally distributed properties for soil water balance computation in the absence of surface runoff.
