大数据驱动的城市洪涝灾害风险感知与预警决策研究范式

王慧敏 黄晶 刘高峰 佟金萍 曾庆彬

摘 要：城市洪涝灾害防控是国家防洪安全的重大需求，更是城市公共安全管理的重要内容。如何利用大数据实现全信息融合、全过程管控、全社会参与的城市洪涝灾害预警与全景式决策是国内外关注的前沿热点和重点领域。本文从“跨界关联、粒度缩放、全局视图”视角出发，提出了大数据驱动的城市洪涝灾害风险感知与预警决策研究范式，包括：利用天空地网多源异构数据信息，基于“数据驱动”的关联关系挖掘与“模型驱动”的因果解析方法，感知暴雨洪涝情景态势;建立基于暴雨洪涝情景的空间动态网络风险研判模型，构建“观测-感知-辨析”的城市洪涝灾害动态预警模式;创建以数据为中心的扁平化城市洪涝灾害管理框架，提出全景式灾害应急合作响应与风险控制策略。

关键词：大数据;城市洪涝;风险感知;预警;决策

中图分类号：C93 文献标识码：A 文章编号：2097-0145（2022）01-0035-07 doi：10.11847/fj.41.1.35

Abstract：Urban flood disaster prevention and control is a major demand for national flood control and is also an important part of urban public safety management. How to use big data to achieve full information fusion， whole process control and participation of the whole society in urban flood disaster warning and panoramic decision-making is the frontier hotspot and key area of concern at home and abroad. From the perspective of “cross-border association， granularity scaling and global view”， this study proposes a research paradigm of big data driven urban flood disaster risk perception， early warning， and decision. Based on “data-driven” association mining and “model-driven” causality analysis method， multi-source heterogeneous data information of sky and earth network is used to perceive rainstorm and flood scenarios. The spatial dynamic network risk analysis model based on rainstorm and flood scenarios is established， and the “observation-perception-discrimination” urban flood disaster dynamic warning model is constructed. A data-centered flattened urban flood disaster management framework is established， and a panoramic disaster emergency cooperative response and risk control strategy is proposed.

Key words：big data; urban flood; risk perception; early warning; decision

1 引言

城市化、工業化的快速发展及人类生产生活方式的转变，加剧了水资源压力，显著改变了降雨径流、汇流路径和方式等水文过程，水循环多过程的非一致性特征凸现[1，2]，表现为城市“雨岛效应”凸显、极端暴雨频发，城市内涝升级，“逢雨必涝”、“城市看海”已成常态。北京“7.21”暴雨、武汉、深圳以及郑州“720”特大暴雨，严重威胁了人民生命财产安全。据《中国水旱灾害公报》数据显示，近20年全国洪涝灾害直接经济损失年均1500多亿元，占GDP的1.5%～2%;近10年年均损失高达2300多亿元，灾害损失呈显著增长趋势。……