城市洪涝灾害预警与全景式决策云平台

Influenced by both global climate change and human activities, especially with the increasing risk of flood due to rapid urbanization in China, there appears a severe challenge in urban flood early warning and control issue. The idea of big-data-driven has not only fundamentally changed the traditional thinking and paradigm of disaster management, but also provide new research paradigm and method for urban flood early warning and control. This project initiates at the management requirements of urban flood risk decision in China in the context of changing environment, according to the features of highly urbanized rainstorm flood hazard inducing and bearing environment variations. It is based on big-data-driven decision-making paradigm, with all methods such as field investigation, big data analysis, cloud platform development, etc. The research has following emphases: establish a unified spatial expression of urban flood big data resource pool; identify precursor information of disaster early warning with big-data-driven‘correlation＋causality’ method; build an ‘observation-perception-discrimination’ disaster early warning mode based on risk assessment; develop a ‘data + model’ driven panoramic decision paradigm and cloud service platform. Besides, with three case studies in Nanjing, Shenzhen and Jingdezhen, a universal and operable solution as well as supporting platform of urban flood management system is developed, providing theoretical and practical guidance for urban flood public security management.

受全球气候变化和人类活动的双重影响，特别是我国城市化快速发展导致洪涝灾害风险加剧，城市洪涝灾害预警和防控面临严峻挑战。大数据驱动使传统的灾害管理思维和范式发生根本性变革，也为城市洪涝灾害预警防控提供了全新的研究范式和研究方法。本研究立足变化环境下我国城市洪涝风险决策的管理需求，针对高度城市化暴雨洪涝孕灾承灾环境变异的显著特征，以大数据驱动的决策范式为基础，结合实地调查、大数据分析、云平台开发等手段，重点研究：统一空间表达的城市洪涝灾害大数据资源池构建，大数据驱动的“关联+因果”灾害预警前兆信息辨识，基于风险研判的“观测-感知-辨析”灾害预警方法，“数据+模型”驱动的灾害全景式决策范式及云服务平台研发，并在南京、深圳等典型城市开展实证和示范研究，形成普适性、具有可操作性的城市洪涝灾害管理系统解决方案及支撑平台，为城市洪涝灾害公共安全管理提供理论、方法及实践指导。

城市地区人口、建筑、道路、财产等高度集聚，现有基于气象、水文等单要素的预报预警方法与技术，难以满足多要素复杂条件下城市洪涝灾害预警与防控的需求。大数据时代传统的灾害管理思维和范式正发生根本性变革。本项目借助大数据与人工智能技术，创建了大数据驱动的城市洪涝灾害风险智能预警理论、方法与技术，实现了洪涝灾害复杂过程实时感知和精准风险预警。取得如下创新成果：①发明了洪涝灾害事件的知识表达与态势智能感知技术，解决了洪涝灾害情景重构的难题。构建了“采样观测-要素提取-风险表达”的城市洪涝灾害态势感知框架；发明了基于文本挖掘的洪涝灾害事件知识图谱构建方法；提出了基于Mask R-CNN与自然语言处理的洪涝灾害的全要素识别技术。②提出了大数据驱动的城市洪涝灾害模拟与风险预警方法体系，刻画出“暴雨-承灾体-淹没”的灾害风险形成机制，实现了洪涝灾害风险分区、分级、分类实时智能预警。提出了数据与物理模型驱动的城市洪涝灾害淹没模型；构建了基于Source-Pathway-Receptor的城市洪涝灾害风险测度框架；构建“观测-感知-辨析”的城市洪涝灾害动态预警模式，提出了面向出行的城市洪涝灾害风险预警方法。③提出了多重赋能下城市洪涝灾害应急治理与韧性管控体系，实现了跨系统、跨部门、跨层级的城市洪涝灾害风险管控。构建了即时信息交互下的社区空间网格洪涝灾害应急疏散体系；构建了城市灾害系统性风险的“元治+数治”治理框架与现实路径；构建了城市生命线子系统的洪涝韧性动态模拟模型，提出了城市洪涝灾害应急响应决策方法与韧性规划策略。开发了城市洪涝灾害全景式决策及云服务平台，实现了灾害智能感知、精细化动态模拟、实时预报预警、应急决策等功能，在长三角城市群、珠三角城市群、深圳市、景德镇市示范应用，形成了普适性、具有可操作性的城市洪涝灾害管理系统解决方案及支撑平台，为城市洪涝灾害公共安全管理提供理论方法及实践指导。