Author(s): Hannah Spero; Donna Calhoun
Linked Author(s): Hannah Spero
Keywords: GeoClaw, HEC-RAS, dam-break modeling, Teton Dam failure, tsunami modeling
Abstract: I. Model Overview: Worldwide dams fail and endanger downstream populations, and thus improving our understanding of numerical models is critical. Moreover, although modeling studies of dam-break problems abound, studies using historical events are rare as they require robust datasets, lending to remaining questions for model parameterization. This study compares HEC-RAS (Hydraulic Engineer Center’s- River Analysis System) dam failure software to the GeoClaw geophysical modeling software by modeling the 1976 Teton Dam failure for GeoClaw validation. GeoClaw is a Fortran-based open-source software, which uses Python for the user interface and downstream flow visualization. It has been benchmarked for storm surge and tsunami modeling, demonstrating efficacy to simulate dam-break flow regimes. Geoclaw uses high-resolution shock-capturing finite volume methods on logically rectangular grids, including latitude-longitude grids on the sphere. Dry states are handled automatically to model flood inundation, and adaptive mesh refinement is used to solve large-scale dam failure problems efficiently. As the Teton Dam failure domain contains complex terrain (constrained flow in river canyon to floodplain), our GeoClaw model discretizes friction using the empirical Manning formulation with a uniform ‘n’ value. Additionally, GeoClaw can be run in parallel using shared memory via OpenM lending to low computational costs. II. Model Comparison: This study addresses existing knowledge gaps by quantifying the influence of breach parameterization, characteristic mesh cell size, adaptive mesh refinement, and reservoir volume on downstream flow. The comparison between industry-standard HEC-RAS and GeoClaw indicates the dam-break model validated by its stability, low computational cost, and conservation properties. GeoClaw results are commensurate with historical gauge records, field observational data, and HEC-RAS model results. A sensitivity analysis of the Teton Dam model demonstrates the Teton Dam breach time of one hour has little impact on the downstream flow arrival times or inundation extent. We find the reservoir volume impacts flow arrival times and maximum depths, whereas the characteristic mesh size affects the computational cost. III. Additional Applications: Recent work focuses on coupling granular-fluid flow models for landslides to tsunami depth-averaged flow models for landslide-generated tsunami formation and propagation. Climate change threatens coastal communities, and dangerous landslides are expected to increase. Understanding the hydro-environment where significant landslides can trigger tsunamis with little to no warning time, generate waves that reach shore more quickly than they would from a tectonic tsunami, and run up farther inland than tectonic tsunamis is critical. To better predict and understand these hazards, we must improve depth-averaged flow numerical modeling that couples granular-fluid flows for landslides and the influence of landslide dynamics on tsunami formation and propagation. Recent advances in high-performance computing now allow for simulating the seamless complex landslide-triggered tsunamis. This research combines the 2D GeoClaw shallow-water equations and a 2D debris model, D-Claw for modeling efforts.