Behaviour of nourishments in quasi 3-dimensional graded sediment models

Abstract

Bed degradation in a number of Dutch river branches, like the Bovenrijn, can cause various problems in the near future. At low waters the navigation depth at the non-erodible layer near Emmerich can become too small. Other problems could be lower ground water levels and stability of structures in and near the river, like groynes. To diminish negative effects of bed degradation, nourishing material can be an effective solution. For a better understanding of nourishment behaviour and a better prediction of nourishment propagation, a tracer nourishment released in Germany in 1996 has been modelled and compared with field data. This tracer nourishment was released in the river Rhine at Iffezheim, Southern Germany, chainage kilometre 336. Propagation of this tracer has been recorded to approximately 60 kilometres downstream of the dumpsite. The model used in this research, is a quasi-3D model with a graded sediment module. Simulating with a graded sediment module is important, since the mixture of the tracer nourishment modelled is different from the original bed material and there is an interest in the difference in behaviour between the finer and coarser tracer fractions. A quasi-3D format is used, because spatial scales less than the river width and transverse sorting effects might be important, as well as the parameterization of important 3D effects, like spiral flow. For the description of sediment transport a modified Meyer-Peter-Müller formula is used, the sediment balance of the river bed is described by the model of Hirano. Hiding and exposure effects are implemented by the formulation of Egiazar off, modified by Ashida & Michiue. The bed load transport vector is adjusted by formulations for the effects of spiral flow and transverse bed slope. The roughness is calibrated against the water level for several relevant discharges. The sediment transport formula is calibrated against the yearly sediment transport. The discharge is schematized in two ways: either a constant representative discharge that yields the same yearly transport, or a hydrograph with five different discharge levels is used. Results of case studies show that the thickness of the active layer, hiding and exposure effects and the discharge schematization are important parameters for propagation of the tracer nourishment. Hiding and exposure effects appear to be quite different for a number of existing formulations. Simulation with a hydrograph instead of a constant representative discharge shows important differences: the propagation speed of specific sediment fractions is different. The coarsest fractions move just a little bit, but are still hardly mobile, though not completely immobile as in the computations with the constant representative discharge. Compared to the field data, the finer fractions propagate too slowly, but the coarser fractions hardly move at all. To change this, relevant parameters that can be changed within the model concept used are the active layer thickness, critical Shields value and hiding and exposure relation. It is uncertain however, if the model concept used can represent a satisfactory approximation of the behaviour of all nourishment fractions simultaneously. A number of physical processes that occur in the river reach just downstream Iffezheim, are not included in the model concept. Significant dunes appear to be present in this river reach, introducing vertical and horizontal sorting processes and different hydrodynamic conditions. Furthermore, it is questionable if the critical Shields value should not be variable with the sediment diameter. Finally, navigation appears to be possibly important. Development of a model which takes into account these physical considerations could be beneficial for prediction of sediment nourishments. On locations where nourishment could be an effective solution, field data of the propagation of specific tracer fractions could give important information for rough estimations of nourishment behaviour.