Evaluating the potential for simulating floodplain processes using two-dimensional distributed models
Mitchell, Christabel Anne
PublisherUniversity of Exeter
MetadataShow full item record
Previous studies of overbank hydraulics and sedimentation have adopted a variety of methods to investigate overbank processes, using either field based, physical or numerical modelling approaches. In this study a two-dimensional depth-averaged model of overbank hydraulics (Hydro2de) is tested and then used to develop a two-dimensional model of suspended sediment transport and deposition. A model grid was used with a higher resolution than those employed in most equivalent model applications in order to represent the topographic complexity of the natural floodplain in detail. The models were calibrated and rigorously tested using a varied and detailed set of spatially and temporally distributed field data. Models were applied to a short reach of the River Culm, Devon, UK. A wide range of field and laboratory data were used to enable model implementation, calibration and validation. Continuous monitoring of stage and sediment concentration were carried out at the site for the period of study and existing data were used from gauging stations located up- and downstream of the site. A large data set of spatially distributed sediment concentration, velocity and flow depth data were collected across the floodplain over a series of flood events. Inundation boundaries were identified using GPS surveys supplemented by field observations and both ground and aerial photography. The amount of overbank deposition was measured for individual floods using astroturf sedimentation traps. Medium-term sedimentation rates were estimated at selected locations using the Caesium-137 whole core technique. Both models were shown to yield good results which, when compared with field measurements were within the limits of uncertainty associated with these data. Simulation results allow a number of conclusions to be drawn about floodplain processes and their representation by numerical models. First, representation of complex, small-scale floodplain topography (features with dimensions <10 m) is of considerable importance when modelling flow and sedimentation on natural floodplains. Second, sediment transfer from the channel to the floodplain is dominated by advection rather than diffusion. Third, low magnitude events are more important than infrequent high magnitude events in terms of total conveyance losses, despite the lower sediment load of the former. Fourth, high resolution distributed models may be used to develop simple conceptual statistical models that capture the complexity of hydraulic conditions on natural floodplains. Finally, further work on floodplain modelling should concentrate on the development of a physically-based understanding of deposition processes in the presence of vegetation.