城市人因视角下公众灾害感知时空特征及其影响因素研究

Spatio-Temporal Characteristics of Public Disaster Perception and Its Influencing Factors from the Perspective of Urban Ergonomics

近年来全球范围内极端天气事件频发,公众作为城市承灾的主体之一,通过微博等社交媒体表达的对灾害进程的主观感知,深刻反应并影响着应急响应和灾情发展,因此,通过城市人因的视角和方法进行公众灾害感知研究,对城市韧性提升有重大意义。本文以深圳2018年9月台风“山竹”期间的微博数据为例,分别利用LDA模型文本聚类和SnowNLP模型情感训练得到公众灾害主题感知和公众灾害情绪感知,基于DBSCAN空间聚类,归纳公众灾害感知的时空分布特征,并从物理建成环境和公众社会属性两方面探究影响公众灾害感知特征的因素。揭示极端天气下公众的灾害感知特征,及其与建成环境及社会属性之间的影响关联,为城市韧性规划设计提供指引。

In recent years,there has been a surge in the frequency of extreme weather events globally.As one of the primary urban disaster-affected entities,the public expresses their subjective perceptions of the disaster processes through social media platforms such as Weibo,profoundly reflecting and influencing the emergency responses and disaster developments.Consequently,investigating public disaster perception from the perspective and methodologies of Urban Ergonomics holds significant implications for enhancing urban resilience.This study utilises Weibo data from the Typhoon"Mangkhut"in Shenzhen during September 2018 as a case study.Employing the Latent Dirichlet Allocation(LDA)model for text clustering and the SnowNLP model for sentiment analysis,the research derives public disaster thematic perceptions and emotional sentiments.Through spatial clustering using the Density-Based Spatial Clustering of Applications with Noise(DBSCAN)algorithm,the spatio-temperal distribution features of public disaster perceptions are summarised.Furthermore,the study explores factors influencing the characteristics of public disaster perception from two perspectives:the physical built environment and public social attributes.The findings reveal distinctive characteristics of public disaster perception under extreme weather conditions and establish associations between these perceptions and the built environment and social attributes.This research serves as a guide for urban resilience planning and design in the context of extreme weather events.