Dam failures occur worldwide, often from factors including aging structures, extreme hydrologic loading, and design oversights related to the changing climate. Understanding and mitigating risk to downstream inhabited areas require developing and improving low-cost high-fidelity tools, such as numerical models, which allow emergency managers to predict the consequences of dam failures better. Two-dimensional (2D) depth- averaged hydraulic models can provide valuable insights into the importance of breach parameters or downstream flow characteristics, but historical studies considering historic failures using real topographies are less common in literature. This study compares GeoClaw, a 2D hydraulic model with novel adaptive mesh refinement capabilities, to an industry standard software HEC-RAS (Hydrologic Engineering Center - River Analysis System) using the 1976 Teton Dam failure as a case study. The suitability of GeoClaw for dam failure modeling is determined based on its capability to resolve inundation extent and flood wave arrival times. This study performs sensitivity analyses of the HEC-RAS model to compare an instantaneous dam breach assumption with a time-dependent breach formation for quantifying the model uncertainty. We find the 2D GeoClaw dam-break model results compare reasonably with historical gauge records and field observational data and HEC-RAS results. The model demonstrates stability and relatively low computational costs. Our findings highlight opportunities for future work, with the GeoClaw software performance supporting continued studies to evaluate performance. Outcomes of this study will assist dam owners, floodplain managers, and emergency managers by providing an additional tool for estimating the impacts of dam failures to protect lives and infrastruc- ture downstream.
Download : sp-ca-sh-2022.pdf