IMIS

Both, the MIKE SHE and WetSpa models, simulate small decreases of the baseflow and increases of the overland flow and peak flows, particularly during summer. However, the WetSpa model estimated overland flow volumes and peak flows up to four times higher than the MIKE SHE model. Increases of the peak flows between 13.3% and 24.2% were estimated by the MIKE SHE model against increases between 42.5% and 71% by the WetSpa model. During summer, when heavy rainfall are more likely to occur, the predicted impact and model-based impact uncertainty is even larger. The higher peak flow estimates by the WetSpa model were attributed to the high increase of the overland flow volume compared to the MIKE SHE model, which was mainly induced by the implementation of the imperviousness factor of the urban area in the WetSpa model structure. However, also the different routing procedures, in which infiltration and evaporation losses might occur in MIKE SHE, and the high interception differences between the models contribute – in lesser extent – to the hydrological impact uncertainty in response to urban growth. The effects of the urban development are manifested on the different hydrological processes too. Urban land prevents infiltration and further recharge. However, in both models only small decreases to even small increases in these processes were quantified. Although both models performed equally well under the reference conditions, their impact predictions in reponse to urban growth indicate still quite some uncertainty is present, not only in prediction of peak flow changes, but also in the underlying hydrological processes. There is a also large spatial variability in the urban impact over the Grote Nete basin. Along the Grote Laak stronger increases of the peak flow were simulated than elsewhere in the basin. These peak flow changes are strongly correlated to the changes of urban area in the watershed and but also to the changes of in other land covers (grassland, cropland). However, each watershed reacts in a different way to the urban land use changes, depending on its initial urban land fraction and coverage. Given that it is difficult to make definite statements on the accurateness of the extreme flow and flow change results of the individual models, a multi-model ensemble approach where different plausible model structures are applied and demonstrated by this study, is extremely useful, and allows decision making to be based on uncertainty assessment that includes model structure related uncertainties. However, it needs to be stated that urbanization is much more complex than what was handled here. For more precise estimates of the urban growth impact one needs to account also for the changes in sewer system infrastructure, the effects of infiltration to or leakage from pipes and sewer systems, the effect of external discharges from wastewater treatments plant or non-runoff related activities in the catchment (e.g. industry), the water transfers between catchment boundaries, etc.