Morphodynamic river processes and techniques for assessment of channel evolution in Alpine gravel bed rivers

Over the past 10 years many restoration projects have been undertaken in Austria, and river engineering measures such as spur dykes and longitudinal bank protection, which imposed fixed lateral boundaries on rivers, have been removed. The EU-Life Project “Auenverbund Obere Drau” has resulted in extensive restoration on the River Drau, aimed to improve the ecological integrity of the river ecosystem, to arrest riverbed degradation, and to ensure flood protection. An essential part of the restoration design involved the consideration of self-forming river processes, which led to new demands being imposed on river management. aper illustrates how model complexity is adapted to the solution and evaluation of different aspects of river restoration problems in a specific case. Point-scale monitoring data were up-scaled to the whole investigation area by means of digital elevation models, and a scaling approach to the choice of model complexity was applied. Simple regime analysis methods and 1-D models are applicable to the evaluation of long-term and reach-scale restoration aims, and to the prediction of kilometre-scale processes (e.g. mean river bed aggradation or degradation, flood protection). 2-D models gave good results for the evaluation of hydraulic changes (e.g. transverse flow velocities, shear stresses, discharges at diffluences) for different morphological units at the local scale (100 m–10 m), and imposed an intermediate demand on calibration data and topographic survey. The study shows that complex 3-D numerical models combined with high resolution digital elevation models are necessary for detailed analysis of processes (1 m–0.01 m), but not for the evaluation of the restoration aims on the River Drau. In conclusion, model choice (complexity) will depend on both lower limits (determined by the complexity of processes to be analysed) and upper limits (field data quality and process understanding for numerical models).