A process proof test for model concepts: Modelling the meso-scale

Abstract In hydrological modelling the use of detailed soil data is sometimes troublesome, since often these data are hard to obtain and, if available at all, difficult to interpret and process in a way that makes them meaningful for the model at hand. Intuitively the understanding and mapping of dominant runoff processes in the soil show high potential for improving hydrological models. In this study a labour-intensive methodology to assess dominant runoff processes is simplified in such a way that detailed soil maps are no longer needed. Nonetheless, there is an ongoing debate on how to integrate this type of information in hydrological models. In this study, dominant runoff processes (DRP) are mapped for meso-scale basins using the permeability of the substratum, land use information and the slope in a GIS. During a field campaign the processes are validated and for each DRP assumptions are made concerning their water storage capacity. The latter is done by means of combining soil data obtained during the field campaign with soil data obtained from the literature. Second, several parsimoniously parameterized conceptual hydrological models are used that incorporate certain aspects of the DRP. The result of these models are compared with a benchmark model in which the soil is represented as only one lumped parameter to test the contribution of the DRP in hydrological models. The proposed methodology is tested for 15 meso-scale river basins located in Luxembourg. The main goal of this study is to investigate if integrating dominant runoff processes, which have high information content concerning soil characteristics, with hydrological models allows the improvement of simulation results models with a view to regionalization and predictions in ungauged basins. The regionalization procedure gave no clear results. The calibration procedure and the well-mixed discharge signal of the calibration basins are considered major causes for this and it made the deconvolution of discharge signals of meso-scale basins problematic. From the results it is also suggested that DRP could very well display some sort of uniqueness of place, which was not foreseen in the methods from which they were derived. Furthermore, a strong seasonal influence on model performance was observed, implying a seasonal dependence of the DRP. When comparing the performance between the DRP models and the benchmark model no real distinction was found. To improve the performance of the DRP models, which are used in this study and also for then use of conceptual models in general, there is a need for an improved identification of the mechanisms that cause the different dominant runoff processes at the meso-scale. To achieve this, more orthogonal data could be of use for a better conceptualization of the DRPs. Then, models concepts should be adapted accordingly.