关键词: 城市韧性, 社交媒体, 时空间分析, 暴雨洪涝, 大数据

Abstract:

Recently, user-generated content with spatial inference such as the check-in data in social media has been used in disaster research and planning. The volunteered geographical information has been used in representing the spatio-temporal patterns of people’s response towards large-scale hazards. This study focus on a small-scale but repeatedly occurred hazard – rainstorms and associated floods (RF). RF has posed a tremendous challenge to urban resilience, especially for cities in developing countries experiencing rapid land use changes. Using the RF in Nanjing City as a case, Sina Weibo check-in data during July 1-21, 2016 was collected and their content were analyzed. Specifically, the RF-related keywords were compiled and used to identify RF-related tweets from background check-in records. Then sentiment analysis was applied to show attitude expressed in each FR-related keywords. Then, the public awareness index and public sentiment index were used for revealing the spatio-temporal patterns of pubic responses towards RF. Our findings show that: 1) Temporally, the majority of RF-related tweets were posted during the rainstorm period; within a day, the RF concern was most discernable during the morning and evening peak hours; 2) Spatially, high public concern towards RF was found in areas within the main city and three new districts; and the major transport hubs and new districts with low altitude and rapid urban construction in recent years showed a higher level of concern on RF; 3) Both the temporal and spatial analysis suggest that RF adversely influence people’s travel-related behavior the most. Based on the knowledge gained from this analysis, policy implications were proposed to increase the urban resilience towards RF covering both hardware and software parts. It is suggested that these critical facilities including drainage facilities and other underground infrastructure, transport systems, and communications systems require advanced planning and long-term maintenance, especially for new city districts and key regional and urban transport hubs. Furthermore, it is of significance that information management and easily accessible communication function well during the RF, making the government, industry and the general public well-informed about the ongoing situation. Methodologically, this study further confirms that social media data can provide valuable real-time spatial information on understanding people’s responses towards RF, which could be used by local governments in emergency management. To enhance urban resilience, a quick decision making based on collection and analysis of social media data could be a good example of the big data use in smart city development.

Key words: urban resilience, social media, spatio-temporal analysis, rainstorm and associated floods, big data