Keywords
Numerical oscillations, Continental hydrodynamic simulation, Boundary condition configurations
Location
Colorado State University
Start Date
26-6-2018 5:00 PM
End Date
26-6-2018 7:00 PM
Abstract
Recent developments in computational techniques and data availability create the opportunity to run full hydrodynamic simulations (one-dimensional Saint-Venant equations) of river networks at regional-to-continental scales. However, the literature provides few guidelines on boundary condition configurations for such simulations. Unfortunately, any model that is sufficiently accurate in dynamics will typically be prone to numerical oscillations and instabilities associated with sharp gradients in boundary conditions. Damping numerical oscillations and/or manual ad hoc adjustment of boundary conditions is a relatively trivial problem when simulating steady or slowly-varying hydraulic behaviors in a short river reach. However, given the massive data required for a continental-scale hydrodynamic simulation, such ad hoc approaches cannot be applied or readily automated. As a result, numerical oscillations from the inappropriate configuration of boundary conditions can easily cause catastrophic numerical results.
Our work examines the effects of boundary condition configurations in large-scale river network simulations. The study includes: (1) effects of channel bathymetry spatial resolution on representing the underlying physics; (2) the impacts of bottom slope discontinuities that are typically in real-world channel geometry data; and (3) the influence of lateral inflow boundary conditions on numerical stability and oscillations. We focus on evaluating the needed level of detail in channel geometry and introduces the concept of a “reference bed-slope line” to reduce effects of inhomogeneous source terms in the dynamic equations. We analyze the relationship between the magnitude of inflows and the channel geometry, and propose a lateral flow-limiter to reduce the numerical oscillations associated with sharp changes in inflow conditions.
A Thorough Examination for Boundary Condition Configuration in Continental-Scale Hydrodynamic Simulation
Colorado State University
Recent developments in computational techniques and data availability create the opportunity to run full hydrodynamic simulations (one-dimensional Saint-Venant equations) of river networks at regional-to-continental scales. However, the literature provides few guidelines on boundary condition configurations for such simulations. Unfortunately, any model that is sufficiently accurate in dynamics will typically be prone to numerical oscillations and instabilities associated with sharp gradients in boundary conditions. Damping numerical oscillations and/or manual ad hoc adjustment of boundary conditions is a relatively trivial problem when simulating steady or slowly-varying hydraulic behaviors in a short river reach. However, given the massive data required for a continental-scale hydrodynamic simulation, such ad hoc approaches cannot be applied or readily automated. As a result, numerical oscillations from the inappropriate configuration of boundary conditions can easily cause catastrophic numerical results.
Our work examines the effects of boundary condition configurations in large-scale river network simulations. The study includes: (1) effects of channel bathymetry spatial resolution on representing the underlying physics; (2) the impacts of bottom slope discontinuities that are typically in real-world channel geometry data; and (3) the influence of lateral inflow boundary conditions on numerical stability and oscillations. We focus on evaluating the needed level of detail in channel geometry and introduces the concept of a “reference bed-slope line” to reduce effects of inhomogeneous source terms in the dynamic equations. We analyze the relationship between the magnitude of inflows and the channel geometry, and propose a lateral flow-limiter to reduce the numerical oscillations associated with sharp changes in inflow conditions.
Stream and Session
A6: Innovation in Continental Scale Modelling for Decision-making, Research, and Education