Abstract
Novel soil datasets and the application of pedotransfer functions provide soil hydraulic input data for modelling hydrological processes at different scales. We aimed to evaluate the reliability of soil hydraulic parameters derived by indirect methods in simulation of soil moisture time series and water budgets at profile level of three sites (Forest, Orchard and Grassland) from a Central European catchment (Lake Balaton, Hungary). Five soil-vegetation-atmosphere model variants were set up with the Hydrus-1D model for each site, differing only in the parametrization of input soil data: i) a calibrated reference, ii) measured values, iii) values predicted from measured basic soil properties, iv) values predicted from national soil map information, v) values derived from the 3D soil hydraulic dataset of Europe. Calibrated soil parameters led to Nash-Sutcliffe efficiency 0.50, 0.54 and 0.71 for the Forest, Orchard and Grassland Site respectively. The outcomes for model efficiency of soil moisture underline the superiority of local databases over regional ones and the need for more detailed vertical discretization during modelling. The model performance according to soil moisture and water budget accuracy led to different rank order of model variants. Water budget comparisons indicated moderate differences between the hydrologic fluxes simulated by the different model variants, emphasizing the uncertainties associated with soil hydraulic parametrization either at local or at watershed scale.