基于多智能体的城市洪涝灾害动态脆弱性计算模型构建

随着全球气候变化加剧和城市化进程的发展,暴雨洪涝灾害对城市社会经济活动造成的影响越来越严重。作为量化灾害影响程度、社会经济损失与风险评估的重要方法,城市洪涝灾害脆弱性研究对洪涝灾害风险管理与规划具有重要意义。目前,城市脆弱性的内涵进一步延伸,脆弱性指标体系不断丰富,所取得的研究成果能够反映一定时期内城市洪涝灾害脆弱性的整体特征,却无法同时有效地解决城市洪涝灾害脆弱性的空间非均匀性问题和动态计算问题。本文从脆弱性的动态特征出发,提出了城市洪涝灾害动态脆弱性计算方法,建立了城市洪涝灾害脆弱性量化模型与面向洪涝灾害脆弱性计算的多智能体模型,形成了基于多智能体的城市洪涝灾害脆弱性计算方法。在NetLogo平台上,利用智能体模型对浙江省丽水市城区的动态脆弱性进行了仿真模拟。研究结果表明该模型有助于发现不同时空条件下城市存在的“脆弱区”,为城市洪涝灾害脆弱性的时空分异特征分析、区域脆弱性的精细化评价提供方法支撑,能够为缓解措施的实施和洪涝灾害的精细化管理提供有效指导。

An Agent-based Computing Model of Urban Vulnerability to Flood Hazard

With the intensification of global climate change and the development of urbanization, the impact of heavy rains and floods on urban social and economic activities has become serious. As an important method to quantify the hazard impact and socio-economic loss, and to conduct risk assessment in urban hazard research, study on Urban Vulnerability to Flood Hazard (UVFH) is of great significance to flood risk management and planning. It can also contribute to interpretation of "human-earth" relationship and the effect of human activities on eco-environmental systems through geography and other related disciplines. Recent research has extended the connotation of vulnerability and enriched the indicator system. Their findings can reflect the overall features of urban vulnerability in a period. However, the current studies are still incapable of achieving vulnerability evaluation considering spatial heterogeneity and dynamic characteristic simultaneously. Therefore, in view of the research on flood hazard vulnerability, this paper proposed an agent-based computing model for UVFH calculation by considering its dynamic characteristics. Firstly, after analyzing the dynamic characteristics of UVFH, the computing framework of UVFH is designed, which sets up the quantitative model, and puts forward an agent-based computing method of UVFH. This framework models the flood hazard in four parts: urban environment, flood simulation, agents, and multi-agents' behaviors. Various environmental and socio-economic indicators are selected comprehensively to quantify the urban exposure, susceptibility, and adaptability. Secondly, by coupling the LISFLOOD-FP and SWMM model, the dynamic changes of hazard drivers are simulated. Moreover, the multi-agents' behaviors are designed to simulate hazard bearers' response to flood hazard. By computing the dynamic indicators of exposure and susceptibility, the UVFH can be simulated dynamically. The agent-based computing method highlights the dynamic characteristics of UVFH, thus can be used to simulate different scenarios. Thirdly, taking the Liandu District in Lishui City, Zhejiang Province as an example, two scenarios of 50-year flood hazards with different occurrence times were designed on the NetLogo platform. The results showed that the occurrence time of flood hazard made a difference to overall vulnerability, and people's daily routines had a great impact on the degree and distribution of vulnerability. Moreover, the contributions of exposure, susceptibility, and adaptability to vulnerability varied in places. The agent-based computing model of UVFH proposed by this paper can help discover the “vulnerable zone” of a city, and provide methodological supports for the spatio-temporal variation analysis of UVFH and the assessment of regional dynamic vulnerability, thus providing effective guidance for refined flood hazard management.