Calibration of a 3D Hydrodynamic Meandering River Model Using Fully Spatially Distributed 3D ADCP Velocity Data

摘要

This study used fully spatially distributed three-dimensional (3D) velocity data for the improved calibration of a 3D hydrodynamic model. The conventional calibration methods do not account for the spatial distribution of the fully 3D flow field. This study compared the results of the proposed calibration approach with those in the literature to show how choosing different calibration methods can change the model predictive capability. The proposed calibration approach can advance the model prediction of 3D flow complexity; consequently, it enhances the potential for the estimation of river processes such as channel morphodynamics and contaminant mixing. To test the proposed methodology, a tortuously natural meandering river was simulated using the Delft3D hydrodynamic model. The spatially intensive acoustic Doppler current profiler (ADCP) velocity data collected throughout the river model domain provided the fully 3D distributed velocities for the model calibration. For the accurate and realistic comparison of the fully 3D predicted velocities and the measured velocities, an algorithm was developed to match the location of each ADCP bin with the 3D model grid points. The results of this study demonstrated that the fully 3D calibration method led to a calibrated model parameterization that was different from that obtained from conventional calibration approaches. The t-test analysis illustrated that with or without implementation of the horizontal large eddy simulation (HLES) model option, the model results were sensitive to the user input background horizontal eddy viscosity, confirming the importance of an appropriate calibration of the developed 3D model. The results of various calibration approaches were validated using the 3D velocity data collected at a bend section in the river using an acoustic Doppler velocimeter (ADV). It was shown that the proposed 3D calibration approach resulted in the best model performance. The results of this study suggest that the proposed methodology can provide a better calibration potential for the simulation of natural meandering rivers with complex geometry.