A Novel Dynamically Coupled Storm Surge Hazard-Infrastructure Model for Effective Real-Time Risk-Informed Decision Making

美国沿海地区面临风暴潮的巨大风险。来自建模工作的数据可以为决策者提供关键信息，以应对这些风险，但可用的模型范围有限。目前基于脆弱性的防洪系统模型主要集中在一个单一的故障模式，没有考虑各种故障模式之间的因果关系和时间相关性。失效评估被视为时间上的快照，忽略了失效过程的时间演化。目前的方法中采用的涌浪过程线，独立地确定故障概率的可靠性模型忽略了时空涌浪响应的地质结构的性能的影响。此外，我们缺乏一个准确的概念，这种信息短缺如何损害决策者在风暴潮风险和基础设施投资的关键判断。这项工作将为下一代风暴潮和脆弱性模型的开发提供条件，并将为模型对决策的影响提供实验验证和评估，具体如下：1）开发一个适应性分辨率的风暴潮模型，以响应防洪系统不断变化的状态; 2）推导与风暴潮模型完全结合的新的地质结构的时间相关的多维脆弱性模型; 3）开发和利用人在回路实验来验证和测试这些模型对实时决策的影响，（4）为学生和教师创造更多的教育和研究机会，沿着向关键利益攸关方传播研究知识。在该项目下进行的研究将对下一代风暴潮模型的开发产生重大影响，这些模型将与时间依赖脆弱性模型充分结合，以提高洪水情景的预测能力。该研究还将提高决策者如何利用风暴风险评估信息做出关键决策的理解。最终，这项研究将导致有关灾难性风险和基础设施故障的更明智的决策（例如，疏散决策、搜索和救援行动、基础设施投资以及事件前、事件中和事件后规划）。该教育计划将提供基础设施建模、复原力和风险分析方面的综合新课程。此外，该教育计划将提高K8教师在课堂上教授工程学的自我效能，并帮助工程学学生发展教学技能。结果将分发给关键的联邦和区域利益攸关方（例如，部国土安全部、联邦应急管理局、美国海岸警卫队）以及行业合作伙伴。 该项目的结果将通过开发一个由水动力和脆弱性模型组件组成的新型动态耦合建模系统，显著改善风暴潮建模。随机有限差分模型与机器学习技术相结合，将能够生成一类新的多维脆弱性表面，这将增强我们对各种故障过程的理解，并表征故障概率的时间演化。耦合浪涌/脆弱性模型将适应网格分辨率，以应对不断变化的条件下的防洪系统，从而提高预测能力。实验分析将提供这些建模能力如何提高实时决策的评估。

Coastal areas in the US face substantial risk from storm surge. Data from modeling efforts can provide crucial information to decision makers to act against these risks, but available models are limited in scope. Current fragility-based models for flood defense systems are primarily focused on a single mode of failure without consideration of causal relationships and temporal correlations among various failure modes. Failure assessment is treated as a snapshot in time neglecting the time evolution of failure processes. The adoption of surge hydrographs in current methods to independently determine failure probabilities from reliability models neglects the impact of the performance of geo-structures on spatio-temporal surge response. Moreover, we lack an accurate conceptualization of how this informational shortfall impairs decision makers in making critical judgments about storm surge risk and infrastructure investment. This work will provide for the development of the next generation in storm surge and fragility models and will provide for an experimental validation and assessment of the models' effects on decision making, as follows: 1) Development of an adaptive-resolution storm surge model that responds to the changing state of flood protection systems; 2) Derivation of novel time-dependent, multi-dimensional fragility models of geo-structures, fully integrated with the storm surge model; 3) Development and utilization of human-in-the-loop experimentation to validate and test the effects of these models on real-time decision making, and 4) Creation of enhanced educational and research opportunities for students and teachers, along with dissemination of research knowledge to critical stakeholders. The research performed under this project will have a significant impact on the development of the next generation of storm surge models that are fully integrated with time-dependent fragility models to improve forecasting capabilities of flooding scenarios. The research will also improve understanding of how decision makers utilize storm risk assessment information to make critical decisions. Ultimately, this research will lead to more informed decisions about catastrophic risk and infrastructure failure (e.g., evacuation decisions, search and rescue operations, infrastructure investment, and pre-, during, and post-event planning). The educational plan will provide for integrated new curriculum in infrastructure modeling, resilience and risk analysis. Moreover, the educational plan will enhance the self-efficacy of K8 teachers to teach engineering in classrooms and help engineering students to develop pedagogical skills. Results will be disseminated to (and validated with) key federal and regional stakeholders (e.g., Dept. of Homeland Security, FEMA, US Coast Guard) as well as industry partners. The results from this project will provide a significant improvement in storm surge modeling through the development of a novel, dynamically coupled modeling system consisting of hydrodynamic and fragility model components. The stochastic finite difference models combined with machine learning techniques will enable generation of a novel class of multi-dimensional fragility surfaces that will enhance our understanding of various failure processes and characterize time evolution of failure probabilities. The coupled surge/fragility model will adapt the mesh resolution in response to changing conditions of the flood protection systems, resulting in improved forecasting capabilities. The experimental analysis will provide for an assessment of how these modeling capabilities improve real-time decision making.