Investigating the spatiotemporal pattern of urban vibrancy and its determinants: Spatial big data analyses in Beijing, China

Highlights

• Weibo check-ins are used to represent urban vibrancy in Beijing at subdistrict level.

• We investigated point of interests, socioeconomic and locational factors’ impact on urban vibrancy.

• Spatial big data provides in-depth understanding of urban vibrancy and its determinants.

• Multiscale geographically weighted regression analysis was applied.

• Determinants’ impacts varying over places and scales.

Abstract

Investigating urban vibrancy and factors that impact urban vibrancy aids the understanding of urban land use policies, provides solid foundation for scientific urban planning. The boom in information and communication technologies and the advancement of big data extraction provides new sources of data and make it possible to measure and analyze urban vibrancy at a finer spatial and temporal scale. This study aims to portray the spatiotemporal variation patterns of urban vibrancy in 24 h and investigate the potential influence mechanism of it. The central districts of Beijing consisting of 135 subdistricts are selected as the study area. Massive and spontaneous geo-tagged check-in data released from social media platforms has attracted increasing attentions in urban vibrancy studies because it reflects well people’s activities at a certain time, which is a good proxy for urban vibrancy. This study hence uses the check-in data from Weibo, the largest microblogging platform in China, to proxy urban vibrancy. We also extract from multisource spatial big data to explore potential determinants of urban vibrancy. This study seeks to reveal the global and local varying impacts of different factors on urban vibrancy by employ spatial lag model (SLM) and multiscale geographically weighted regression (MGWR) model. Results show that the increase in the number of different point of interests (POIs) improves urban vibrancy. Their effects on vibrancy vary at different times but have no obvious spatial scale variation. Splitting effect and attraction effect of land use diversity are introduced to explain its significantly negative effect on the intensity and fluctuation of urban vibrancy. It requires the wisdom of urban planners to balance these two effects of land use diversity in the process of urban construction. The guidance strategy of “highlighting the main functions and enriching the auxiliary functions” is helpful to build vibrant cities. Socioeconomical conditions, location and accessibility have different spatial scale effects on urban vibrancy at subdistrict level. These findings enable us to have a deeper understanding of the variation patterns and influence mechanism of urban vibrancy in China’s megacities and benefit the urban land use policy research and management community.

Introduction

In the last two decades, China has undergone dramatic increase in both urban population and built-up areas (He et al., 2018, Huang et al., 2017, Wang et al., 2012). Such rapid urban development causes scholars to worry about the disorderly urban sprawl in some areas might lead to inefficiency of creating vibrant urban spaces (Batty, 2016, Jin et al., 2017, Xia et al., 2020). The government has also noticed the potential problems in the process of urbanization and proposes to improve urban vibrancy as an important strategic task in the 13th Five-year Plan. Since Jacobs (1961) put forward the concept of urban vibrancy, it has attracted extensive scholarly attentions from different disciplines, such as urban planning, social sciences and geographical information sciences (Delclos-Alio et al., 2019, Gehl, 1971, Klemek, 2007, Maas, 1984, Montgomery, 1998, Sung and Lee, 2015). Researchers generally believe that urban vibrancy reflects the interaction between the various daily human activities and existing urban facilities, which plays a critical role in promoting comprehensive, coordinated, and sustainable urban development (Kang, 2020, Li et al., 2020b, Mouratidis and Poortinga, 2020, Wu et al., 2018a). Vibrant urban spaces tend to support diverse human activities and facilitate social interactions, bring a diversity of benefits, such as improving the people’s subjective feelings of urban life, attracting investment and talents, enhancing economic competitiveness, and achieving social sustainability (Meng and Xing, 2019, Pinquart and Sorensen, 2000, Zhang et al., 2020). On the contrary, the lack of urban vibrancy may lead to a series of socioeconomic issues and become serious impediments to urban development (Jin et al., 2017, Li et al., 2020b, Woodworth and Wallace, 2017). The studies on urban vibrancy will help urban planners and decision-makers make effective plans to achieve sustainable urban development (Laman et al., 2019, Meng and Xing, 2019). Hence it is imperative for urban scholars to understand the patterns of urban vibrancy and identify the factors that affect urban vibrancy, especially at finer spatial scales.

Data availability used to be the primary issue for quantitative analysis of urban vibrancy (Sung and Lee, 2015, Ye et al., 2018). Most scholars conducted their investigations of urban vibrancy at the scale of the neighborhood through qualitative methods, such as field observations and interviews (Filion and Hammond, 2003, Powe, 2012, Ravenscroft, 2000). However, with the rapid development of information and communication technologies, many data sources emerge to provide previously unavailable information about urban dynamics and open up new opportunities in the studies of urban vibrancy (Garcia-Palomares et al., 2018). In the era of big data, internet users are not only recipients of information but also producers of vast amounts of data. With the permission of users, many social media apps can collect tremendous amount of information along with accurate locational information when they post a message, leaving so-called “digital geographic footprints” (Garcia-Palomares et al., 2018, Tu et al., 2020). The digital geographic footprints are intermingled with their offline daily life and provide new ways of investigating the interaction between human activities and urban environments (Blanford et al., 2015, Garcia-Palomares et al., 2018, Shelton et al., 2015). Compared with traditional data derived from statistical census and surveys, these geo-tagged big data show significant advantages for us to further investigate urban vibrancy because it contains large sample size and have high penetration with strong timeliness (Li et al., 2020b, Xia et al., 2020, Ye et al., 2018). With geo-tagged social media, large-scale and fine-grained datasets become more readily available. These datasets open a new chapter in urban studies in the big data era that could advance the theoretical understanding and practical management of sustainable urban development through the investigation of the highly dynamic but ultimately consistent urban vibrancy. Many recent studies have tried to analyze spatial big data from different sources to further explore urban vibrancy (Delclos-Alio et al., 2019, Lu et al., 2019, Meng and Xing, 2019, Tang et al., 2018, Wu et al., 2018a, Zeng et al., 2018). Our current study follows suit the trend of scholarly work but attempt to provide a more in-depth analysis of urban vibrancy with advanced spatiotemporal analytical techniques to contribute to the scientific understanding of sustainable urban development.

This article aims to further the understanding of the spatiotemporal variation pattern of urban vibrancy at subdistrict level and explore the relationship between urban vibrancy and its determinants. In addition to explore the potential mechanisms of urban vibrancy, this study also proposes that such mechanisms might vary in space. Specifically, this study attempts to make contributions to urban studies in several ways. First, we analyze and distinguish the physical features and socioeconomic attributions of spatial big data from different sources based on our own investigation and reference to previous studies (Huang et al., 2020, Li et al., 2020a, Tu et al., 2020). This is the foundation to choose more reliable indicators to represent urban vibrancy. Because urban vibrancy is a people-centered rather than material-centered concept, the indicators used to measure urban vibrancy should be able to reflect the intensity of human activities directly. Social media data serves the purpose well. Second, dynamics and variability are two prominent features of urban vibrancy, and the spatial and temporal snapshots of human activities show the varied aspects of daily life in cities. The distinct feature of this study is that we divide 24-hour into 12 slots (2 h per slot) rather than simply regarding a whole day as a snapshot of cities, which could accurately reflect the complex and changeable urban vibrancy from a more subtle time scale. Third, we try to summarize the influencing factors of urban vibrancy into a more systematic research framework and take into consideration both spatial autocorrelation and spatial heterogeneity in empirical analysis so that the analytical paradigm can be applied to other urban settings as well. In this study, we propose that the determinants of urban vibrancy might include four aspects: land function, socioeconomic conditions, accessibility, and location. In terms of empirical research, spatial autoregression model is applied to control the influence of potential spatial autocorrelation. To deal with possible spatially varying relationship between urban vibrancy and its determinants, we employ the multiscale geographically weighted regression (MGWR) to address both the scale effect and locational effect of influencing factors. MGWR could improve the estimation efficiency and accuracy, and its result gives the influence scale of different explanatory variables (Fotheringham et al., 2017, Li and Fotheringham, 2020, Yu et al., 2020). The application of MGWR will deepen our understanding of influence mechanism of urban vibrancy in a more holistic way.

The remainder of this article is organized as follows. Section 2 reviews the relevant literature of urban vibrancy and introduces the research framework. Section 3 gives a detailed description of the study area, data sources and empirical methods. Section 4 reports and analyzes the study results, with specific attention to spatial and temporal variation of urban vibrancy and the impacts of its determinants. The last section concludes this study and discusses future work.