Evaluation of runoff, peak flow and sediment yield for events simulated by the WEPP model in a Belgian agricultural watershed

The physically based WEPP (Water Erosion Prediction Project) model was implemented in a small agricultural watershed located in central Belgium, called Ganspoel. The watershed, mainly agricultural and resulting in a smooth topography, covers about 115 hain a landscape typical of large parts of central Europe. Seventeen runoff, peak flow and sediment yield events, collected during a 2-year monitoring period, were simulated by the model. Even though the runoff volume predictions were well correlated to the corresponding observations, WEPP prediction capability was generally unsatisfactory also when different set-up methods of the soil effective hydraulic conductivity were used. The poor performance achieved for runoff volume and peak flow simulations affected sediment yield predictions. The differences between observed and simulated values for runoff, peak flow and sediment yield events may depend on: i) the great number of small runoff and sediment yield events within the available database with which is associated large natural variation and which in many cases are not well reproduced by WEPP; ii) the lack of model calibration processes; iii) the scarceness of information about some important soil physical and hydrological parameters; iv) the land use heterogeneity and crop schedule complexity of the Ganspoel watershed.