数据驱动的极端风暴潮越堤洪水实时仿真与耦合风险定量模型研究

Global climate change and rapid urbanization had made extreme disasters occur in coastal areas of China frequently, especially storm surge. Quantifying the risk of composite disasters and studying the distribution of losses that may be caused would be essential to improve the ability of the region to withstand extreme storm surge disasters. The semi-coupling and semi-quantitative analysis method was mainly used to study the quantitative and expected loss distribution of marine storm surge combined with extreme disaster risk. The risk of storm surge disasters, the vulnerability of dike projects, the risk of flooding space and the overall coupling between them had not been considered comprehensively. In this topic, the coupling and boundary treatment technology between subsystems of storm surge complex disaster would be deeply studied, taking the coupled numerical model of land-sea disaster risk as the research focus and developing a hydrodynamic model of storm surge flood across dike by interdisciplinary approach. With the sensitivity analysis of cell scale, the optimization of cell spatial data structure would be realized. The parameters and states of the models would be modified and optimized by the interaction between the observed data and the dynamic model, the performance of the simulation model under complex terrain would also be further studied. Finally, the interaction mechanism between the dynamic process of storm surge disasters and typical disaster-bearing bodies along the coast would be analyzed and coupling calculation and auxiliary decision making of comprehensive risk would be realized through system integration. The research results would provide methodological and technical support for the establishment of China's comprehensive disaster prevention and mitigation system.

全球气候变化及过速城市化使我国沿海地区极端灾害频发，尤其是风暴潮灾害，量化复合灾害风险并研究可能造成的损失分布，对提高区域抵御极端风暴潮灾害的能力至关重要。对海洋风暴潮复合极端灾害风险定量与期望损失分布的研究，目前主要采用半耦合半定量分析方法，未综合考虑风暴潮致灾危险性、堤防工程脆弱性、承灾空间淹没风险及其之间的整体耦合作用。本课题将深入研究风暴潮复合灾害各子系统之间的耦合与边界处理技术，将海陆区域灾害风险耦合数值模型作为研究重点，跨学科实现风暴潮越堤洪水水动力学模型的开发，伴随对元胞尺度的敏感性分析，实现元胞空间数据结构的优化。通过观测数据与动力学模型的交互耦合进行模型参数与状态修正，进一步研究复杂地形下仿真模型的性能。精细化解析风暴潮灾害动力过程与沿海典型承灾体的相互作用机制，并通过系统集成实现综合风险度的耦合计算与辅助决策，研究成果将为我国全面防灾减灾体系的建立提供方法论与参考依据。

风暴潮灾害在我国是最常见和破坏性最大的自然灾害之一，准确高效地进行风暴潮洪水模拟及风险评估可以最大限度地减轻风暴潮灾害损失。在总结国内外风暴潮灾害损失预测研究的基础上，课题通过风暴潮致灾机理、灾害系统理论以及风险管理理论，基于致灾因子危险性、孕灾环境敏感性、承灾体脆弱体、防灾减灾能力四个方面构建了风暴潮灾害风险等级评估指标体系，提出了最优的组合回归模型，并结合风暴潮灾害评估指标体系对风暴潮区域损失区间进行总体预测。开发的虚拟样本生成技术和集成学习模型，可以有效地抑制模型过拟合，降低过拟合带来的泛化误差。针对特殊功能需求的海岸防护区域，实现了对突变地形描述失真问题的修正，进一步构建了基于栅格水动力学的元胞自动机模型，将二维浅水动力学方程离散化并作为转化规则引入到元胞自动机模型，利用动量和能量守恒机制模拟风暴潮洪水演进过程，通过简化次要信息实现及时高效地模拟风暴潮洪水致灾过程。基于多源遥感数据和深度U-Net网络搭建了潮水边界提取技术，通过基于水平集函数理论构造的潮水边界跟踪模型进行淹没过程的动态修正。对历史风暴潮灾害反演其过程，进行潮灾风险评估。引入多准则风险综合分析方法，综合考虑人口、经济、社会环境因素与评价指标的随机性和模糊性，结合模拟得出的淹没水深、淹没范围等风暴潮洪水灾害信息进行综合分析。项目深化了对风暴潮灾害动力学与风险评估过程的研究，研究成果将为我国沿海防潮减灾提供理论依据。