SENSITIVITY OF PARAMETERS AND APPROXIMATIONS IN MODELS OF TIDAL PROPAGATION AND CIRCULATION

Abstract

In the last ten years a system has been designed for the two-dimensional simulation of the hydrodynamics and water quality in well-mixed estuaries, coastal seas, harbors, and inland waters. The system called SIMSYS2D or WAQUA, can simulate the hydrodynamics in complicated geographical areas, and the model can determine the land/water boundary during simulation [1]. The system accounts for the sources of discharges, tidal flats, islands or dams, and time-varying or time-invarying flow restrictions such as generated by openings in dams, sluices, or storm-surge barriers. In the SIMSYS2D system numerous finite difference approximations of the vertically integrated hydrodynamic equations and their boundary conditions are available to the investigator, A large number of models and model experiments with two-dimensional models have been described in the literature. In many instances, authors have emphasized the advantages of the particular approximations they were using in their model and sometimes made comparisons with other approximations by use of hypothetical geographical areas. Generally, no comparisons are given of different computation methods applied to actual estuaries. Currently, no comprehensive overview of the comparative importance of the approximations of the terms of the hydrodynamic equations based on experiments of typical estuaries, is available. Nor has a practical assessment been made on the comparative importance of timestep size, grid size, depth accuracy, roughness estimates, and approximation of the closure term by viscosity or other expressions. To obtain an insight into the relative importance of computational and systems parameters, a large number of experiments were made with models of the Eastern Scheldt using the SIMSYS2D system in many of its modes. Some of the more important results of this extensive analysis will be presented in this paper.