Author: Cheng-Wei Yu

Publisher:

Published: 2020

Total Pages: 380

ISBN-13:

Simulating dynamic river networks at continental scales has been a long-term goal of the hydraulic and hydrologic communities. However, difficulties in computational costs, data availability, and numerical stability have been major obstacles in applying the full, dynamic Saint-Venant equations over such scales - i.e., creating simulations of "Continental River Dynamics." Recent research has provided progress in addressing both computational costs and data availability, but numerical instability for large-scale models has been problematic (until addressed by the study herein). Prior methods for ensuring numerical stability for short river reaches are inadequate for large-scale Saint-Venant simulations due to their expensive computational cost or incompatibility with numerical schemes designed for large-scale systems. The present research addresses and provides solutions for three major sources of numerical instability in river network modeling: (i) poorly defined initial conditions, (ii) non-smooth source terms in the governing equations, and (iii) channel geometry that is insufficiently smooth or perhaps improperly defined. In the first part of this study, a robust steady Saint-Venant model (SSM) is developed. This new model applies a concept from graph-theory to provide consistent initial conditions for every computational element in a large-scale river network simulation - which is a necessary step for efficiently starting a large-scale unsteady simulation. The present SSM method can effectively generate the initial condition for different scale of river network up to 5000 x faster than the traditional method, and save up to 99% of unnecessary numerical iterations. In the second part of this work, the problem of non-smooth source terms in the governing equations is solved. A new “reference slope” concept is proposed to ensure smooth source terms and eliminate potential numerical oscillations. It is shown that a simple algebraic transformation of channel geometry can provide a smooth reference slope while preserving the correct cross-sectional flow area and the piezometric pressure gradient that drives the flow. The reference slope method ensures a Lipschitz-continuous source term while maintaining all the underlying complexity of the real-world geometry. The method systematically alleviates the oscillatory numerical behaviors provoked by non-smooth bed slope (although it cannot address problems associated with non-smooth cross-sectional area). In the third section of this work, a new sweep-search algorithm (SSA) is developed to identify locations in a network where changes in the channel cross-sectional geometry are responsible for numerical convergence failures that prevent large-scale simulation of a large-scale river network. By taking advantage of computational speed in SSM (developed in the first part of this research), the SSA can recursively examine the numerical stability of each computational element in the network. The new SSA approach provides for automated identification of problem locations in large river network data sets, which is necessary to effectively apply the full dynamic Saint-Venant equations to large-scale river networks. In summary, the research in this study provides further building blocks toward making Continental River Dynamics simulations a reality