ZHANG Yngfn, ZHONG Weijing, WANG De , LIN Feng-Tyn

a Research Center of Digital Planning Technology, Shanghai Tongji Urban Planning & Design Institute, Shanghai, 200092, China

b Hangzhou City Planning & Design Academy, Hangzhou, 310012, China

c College of Architecture and Urban Planning, Tongji University, Shanghai, 200092, China

ABSTRACT

In recent years, major cities around the world such as New York in USA,Melbourne in Australia, and Shanghai in China, have planned to boost their nighttime urban vibrancy levels to spur the economy and achieve cultural diversity. The study of nighttime urban vibrancy from the perspective of spatiotemporal characteristics is increasingly being recognized as part of the essential work in the field of urban planning and geography. This research used mobile phone signaling records to measure urban vibrancy in central Shanghai and revealed its spatiotemporal patterns during nighttime.Specifically, this research explored the changes of urban vibrancy within a day, studied the distribution of urban vibrancy during the nighttime, and visually presented the spatiotemporal changes of nighttime urban vibrancy in central Shanghai. Moreover, on the basis of the behavior pattern of each mobile user, we classified nighttime urban vibrancy into three different types:nighttime working vibrancy, nighttime leisure vibrancy, and nighttime floating vibrancy. We then tried to determine how land use affected nighttime leisure vibrancy. The results showed that urban vibrancy in central Shanghai exhibits a periodic pattern over one-day period. A high-level nighttime urban vibrancy belt is present within central Shanghai. Business offices, hotels, entertainment and recreational districts, wholesale markets,and express services contribute most to the vibrancy at nighttime. In addition,the correlation analysis shows that public and commercial facilities generate high levels of nighttime leisure vibrancy than residential facilities. The mixed land use of public and commercial facilities and residential facilities within 500 m is more critical than the mixed use of a single land lot. The research can be a basis for supporting land use planning and providing evidence for policy-making to improve the level of nighttime urban vibrancy in cities.

ARTICLEINFO

Keywords：

Nighttime urban

vibrancy

Mobile phone data

Land use

Public and commercial

facilities

Residential facilities

Central Shanghai

1. Introduction

Since the second half of the 20thcentury, research on urban vibrancy has attracted the attention of many scholars who study urban planning and geography. The concept of urban vibrancy was first introduced by Jacobs (1961),who defined urban vibrancy as “rich diversity of urban life”. Urban vibrancy is closely related to street life over a 24-h period. Maas (1984) indicated that urban vibrancy includes the concerns of a city’s employment opportunities,service infrastructure, building quality, and satisfaction of residents. It is vital to study how urban vibrancy is generated and how urban vibrancy is distributed in cities. Most scholars have focused on how urban structures and urban cultures affect the vibrancy of a city, such as mixed land use, appropriate urban densities, human-scale environments, street connectivities (Lunecke and Mora, 2018), parking policies (Still and Simmonds, 2000), cultural diversities (Chion, 2009), and urban textures (Sedaghati and Farsi, 2016).

The previous literature shows how the built environment has a great impact on urban vibrancy. Improving the built environment will consequently affect human behavior as well as urban vibrancy. Traditionally, due to the technical limitations and cost constraints of data collection, urban vibrancy is commonly measured by the number of pedestrians on the streets during the daytime, which is collected by investigators; or by the number of trips that residents make within a small research zone, which is obtained from travel surveys.

Nightlife is one of the most critical parts of a modern city. According to the data investigated by United Nations,nearly 70% of the world’s population will live in urban areas by 2050 (Population Division of Department of Economic and Social Affairs of the United Nations, 2018). This can be seen as a challenge or an opportunity. To feed and house the increased population, we have to create more food and construction on the same land. However,cities must do more than merely grow higher or lower, i.e., grow into the sky or underground; they can also grow in the use of time by exploring how to make greater use of the evening and nighttime to provide jobs, support community cohesion, and promote social inclusion (Diplomacy and Seijas, 2021).

Urban planning plays a vital role in ensuring a thriving nighttime economy and diverse urban nightlife. London,for instance, in its Cultural Infrastructure Plan, has proposed an urban planning strategy for building various nighttime vibrancy clusters across the Greater London Area, which would thereby improve the mixed land use of public spaces and attract many more visitors (Greater London Authority, 2019). Traditionally, because National Aeronautics and Space Administration (NASA) and Defense Meteorological Satellite Program/Operational Linescan System (DMSP/OLS) night light data represent sparse data sources, they can be used to fully analyze the level of nighttime urban vibrancy, but there are few corresponding research results, especially from the perspective of individual behaviors. Therefore, understanding the characteristics of nighttime urban vibrancy based on individual behaviors and determining what kinds of built environment elements may affect urban vibrancy are both significant research topics.

The development of information and communication technology enables researchers to accurately, meticulously,and dynamically observe how residents interact with urban settings and provides a more detailed approach for understanding large-scale urban spaces (Ratti et al., 2006; Niu et al., 2015; Luo and Zhen, 2019). In the research on urban vibrancy, newer databases and different methodologies have begun to be introduced, such as applications of mobile phone data (Xie et al., 2018); public transit smart card data (Jang, 2010); heat maps provided by Baidu Maps,Google Maps, and Amap (Zhou et al., 2018); Yelp and Dianping datasets (Rahimi et al., 2018); and location-based social media data, such as Sina Weibo (Zhen et al., 2017). Mobile phone data, with their real-time dynamics, good spatial coverage, accurate time coverage, and detailed individual records, provide a new approach for studying urban vibrancy.

In this paper, mobile phone data are used to conduct a quantitative study of nighttime urban vibrancy in central Shanghai, China. The research data and methods are introduced and explained in detail in the second section,including how to use mobile phone data to identify nighttime urban vibrancy. Then, the paper describes the general characteristics of urban vibrancy in central Shanghai, including the changes in urban vibrancy within a day and the spatial distributions of nighttime urban vibrancy. Next, it discusses three different types of nighttime urban vibrancy in central Shanghai: nighttime working vibrancy, nighttime leisure vibrancy, and nighttime floating vibrancy.Finally, the paper analyzes the influences of land use types and mixed land use for generating nighttime leisure vibrancy and attempts to determine the optimal range of mixed land use around a land lot to better improve nighttime urban vibrancy of the land lot.

2. Data and methods

2.1. Study area

The study area consists of the district that is contained within the Outer Ring Expressway of Shanghai, which covers an area of approximately 660 km2and is usually called central Shanghai. This district, with high population and construction density, mixed land use, and a complex built environment, is a typical area of modern metropolis.Correspondingly, Shanghai has the highest mobile penetration rate in China, and cell phone towers in central Shanghai are dense and numerous, ensuring good data quality.

2.2. Research data

The research data include three datasets: the mobile phone data of Shanghai, the 6thNational Population Census of the People’s Republic of China (National Bureau of Statistics of the People’s Republic of China, 2010), and the land use data of Shanghai. The mobile phone data were got from one of the biggest mobile network operators in Shanghai for two weeks (14 d) in 2014, i.e., from 15 March to 28 March. There are about 3.6×104cell phone towers in the whole city, every 100-300 m in central Shanghai and every 1000-3000 m in the suburban area. Whenever a mobile phone connects with cell phone towers, the cell phone tower will track it and generate a record. The average daily records are about 5.0×108to 9.0×108. Each piece of record includes: (1) a user ID; (2) a timestamp; (3) a location ID of the connected cell phone tower; and (4) the communication event type (event ID) (see Table 1). The user ID is anonymous and encrypted. On average, the data record 1.6×107to 1.9×107user IDs every day (it is approximately 70% of the total population of Shanghai (2.4×107) in 2014).

Table 1Recorded mobile phone data examples.

Human behavior is the focus of studying urban vibrancy in urban planning and geography. Mobile phone data,with their real-time dynamics, good spatial coverage, accurate time coverage, detailed individual records, and high device penetration rate, can effectively reflect people’s behavior patterns. Accordingly, in this study, we used the number of mobile users that were recorded by the cell phone towers to measure the level of urban vibrancy. In addition, similar to previous studies (Jiang et al., 2017; Yan et al., 2019), we inferred the home and workplace locations of mobile users based on the accumulated time that the users spent on each cell phone tower. By comparing the number of users whose homes were located in central Shanghai by subdistrict, which is one of the small political divisions in China, as well as the population in the 2010 census, we found there is a statistically significant relationship between the population and the number of mobile users, with a correlation coefficient of 0.943.

2.3. Research methods

2.3.1. Method of drawing time-vibrancy curve

For ten weekdays during the 14-d study period, we counted the average daily records for each cell phone tower in central Shanghai every two hours. The proportion of the records for each period in one day (R_avgj) was calculated as follows:

whereR_avgjis the proportion of the records in periodjin a day to the total records in the whole day (%);kis the cell phone tower ID (k=1, 2, …, 7608);iis the order number of the ten weekdays (i=1, 2, …, 10);jis the start time with an interval of two hours (j=0, 2, …, 22); andTi_jis the number of records of cell phone towerkin thejthperiod of theithday. For instance,R_avg0 represents the proportion of the records from 00:00 to 02:00 (LST) in a day.

We calculatedR_avgj(j=0, 2, …, 22) for each period to obtain the time-vibrancy curve for central Shanghai:

2.3.2 Identifying nighttime urban vibrancy

The communication event types of the mobile phone data include: (1) outgoing, which comprises outgoing calls and sending text messages; (2) incoming, which contains incoming calls and receiving text messages; (3) powering on; (4) powering off; (5) automatic switching; and (6) periodic location update, that is, when the cell towers and mobile devices are not in contact for a period of time, they will connect automatically and periodically.

As shown in Figure 1, overall, incoming event type accounts for 39.69% of all communication event types in this 14-d dataset, followed by periodic location update (21.91%) and automatic switching (25.81%). The forth communication event type is outgoing, which accounts for 9.29%, and the proportions of powering on (2.11%) and powering off (1.20%) are the lowest. For different periods, the proportions of each communication event type are also different. In the evening, periodic location update accounts for approximately 80.00%, and the number of incoming and outgoing event types decreases considerably. For this reason, we only used the recorded data of incoming and outgoing communication events to analyze nighttime urban vibrancy, which is represented by the number of users that were recorded by the cell phone towers. The method used for identifying nighttime urban vibrancy is shown in Figure 2. In addition, to reflect the changes in urban vibrancy during the nighttime, we divided nighttime vibrancy into two time periods: evening urban vibrancy from 22:00 to 24:00 (LST) and late-night urban vibrancy from 02:00 to 04:00, and calculated vibrancy values. Then, we used the kernel density analysis tool in ArcGIS to visually present the distribution of urban vibrancy.

2.3.3. Measuring the level of mixed land use

Because the minimum distance between most cell phone towers is less than 300 m, we divided central Shanghai into several small research grids (500-m length and 500-m width), with a total of 2767 research grids. Then, we created buffers around the selected research grids with distances of 250 m, as shown in Figure 3, and calculated the land use indicators within each buffer (calculating all areas within each buffer, regardless of overlap). Pearson correlation coefficients were used to measure the relationships between land use indicators and nighttime urban vibrancy values. Each research grid and its corresponding buffers constituted a research unit.

Compared with the other types of nighttime vibrancy, nighttime leisure vibrancy is the most complex and diverse and is greatly affected by urban land uses. We chose nighttime leisure vibrancy of weekdays to analyze the correlations between weekday nighttime leisure vibrancy and land use indicators. In this research, we mainly focused on the influence of three types of land use on nighttime leisure vibrancy: residential facilities, public and commercial facilities, and industrial facilities.

The proportions of land areas of residential facilities, public and commercial facilities, and industrial facilities were used in each research unit. These are the three major land use types in a city. The proportions of different land use types reflect the function of a research unit and will further affect the activities around the research unit at night.

The mixed land use of public and commercial facilities and residential facilities is a common strategy that is used to improve urban vibrancy. In this study, we primarily analyzed the impact of these two mixed land use types on nighttime urban vibrancy. The indicator is shown as follows:

where Ij(C_R) is the degree of the mixed land use of public and commercial facilities and residential facilities in research unit j (%); Pcjis the proportion of the land area of public and commercial facilities in research unit j (%);and Prjis the proportion of the land area of residential facilities in research unit j (%). Concerning these two land use types, regardless of which one has a larger area in a research unit, the higher the degree of balance between them, the larger the land use indicator value. If the areas of the two land use types are the same, the indicator value is equal to 1.

3. Results

3.1. General characteristics of urban vibrancy

3.1.1. Time-vibrancy curve

The daily time-vibrancy curve for central Shanghai is shown in Figure 4. During the bedtime period from 02:00 to 06:00, the number of mobile phone data records is the lowest. Specifically, the records from 02:00 to 04:00 and from 04:00 to 06:00 account for 2.77% and 2.87% of the day’s total records, respectively. During 06:00-08:00, the number of records starts to increase, and the proportion rises to 4.73%. This effect indicates the beginning of the daily lives of the residents. After that time, the number of mobile phone data records rapidly increases and then reaches a maximum in the period from 10:00 to 12:00, accounting for 13.09% of the day’s total records. During the lunch break, the number of records decreases slightly in the period during 12:00-14:00, and the proportion decreases to 11.13% of the total. Soon after this point, when people resume work in the afternoon (from 16:00 to 18:00), the number of records rises again and reaches the second peak, which accounts for 13.07% of the total. After people get off work during 18:00-20:00, the number of records gradually declines but still maintains a high proportion for a while (11.53%). These results provide a partial view of the human nightlife and nighttime urban vibrancy in central Shanghai. With regard to the period from 22:00 to 24:00, which is close to the start of the bedtime hours for most people, the number of records sharply declines.

In short, the number of daily mobile phone data records in central Shanghai presents a double peak curve. The number of records is high during working hours and decreases during the lunch period and after work, generally corresponding to people’s daily behavior patterns.

3.1.2. Spatial distribution of nighttime urban vibrancy

The spatial distributions of nighttime urban vibrancy in central Shanghai on weekdays and weekends are shown in Figures 5 and 6.

In general, compared with evening urban vibrancy (22:00-24:00), the late-night urban vibrancy (02:00-04:00) is more centralized. According to the distribution of late-night urban vibrancy, we can classify the central Shanghai into four types of clusters at the spatial scale. The first cluster type with the highest urban vibrancy density is located near the Changshou Road and Changping Road. This cluster mainly consists of business office buildings and is well equipped with commercial and entertainment facilities, such as KTV clubs, restaurants, 24-h convenience stores,and bars. The concentration of those who work at night and those who enjoy the night makes this area to be the most vibrant place in central Shanghai. On the other hand, the buildings in this area are mostly a combination of towers and podiums. The towers are mainly business offices, with some lofts. The podiums usually contain many retail stores, restaurants, and facilities for entertainment and recreation along the streets. This mixed land use building types also contribute to the diversified nightlife and a high level of nighttime urban vibrancy.

The second type of cluster is distributed in the Zhenru District, which is one of the subcenters of central Shanghai and is located in northwestern central Shanghai, the East Huaihai Road, and the Great World. The East Huaihai Road and the Great World are both located in the city center, with mixed land use and high density; there are masses high-rise office buildings, hotels, and attractions. The Zhenru District is an area where wholesale markets and marketplaces with many aquatic and agricultural products are concentrated. Many people work at night in these markets, maintaining a high level of nighttime urban vibrancy.

The third cluster type is a nighttime urban vibrancy belt that involves the Dapuqiao area and the district near the Madang Road. This cluster does not have a visible vibrancy peak, but the overall vibrancy is still strong at night.

The fourth type of vibrancy cluster is mainly distributed outside the Inner Ring Road, such as the area near the Anguo Road and Zhoushan Road, the Zhongxin Plaza, the Helen International area, the area around the Zhabei District Central Hospital, the area near the Shuicheng Road and Xianxia Road, and the commercial facilities and business offices near the Wuzhong Road and Lianhua Road. The places mentioned above have diverse service sectors and urban functions, including community activity centers, shopping malls, office buildings, hospitals,hotels, bars, and clubs.

The spatial distribution of nighttime urban vibrancy on weekends, in general, is similar to that on weekdays.However, there are still two differences between them. The first is that the area with the highest nighttime vibrancy density shifts from the Changshou Road and Changping Road to the Great World, which is a city-level entertainment region. The second difference is that, compared with weekdays, the nighttime urban vibrancy density around the Huashan Road and Hengshan Road, which are the most popular places for drinking and dining, increases considerably on weekends. This result also shows that people engage in more leisure and entertainment activities on weekend nights.

A comparison of the average nighttime urban vibrancy density, maximum nighttime urban vibrancy density, and standard deviation in different periods is shown in Table 2. We find that the late-night urban vibrancy densities on weekdays and weekends are both lower than the evening urban vibrancy densities. Additionally, urban vibrancy density drops quickly on weekday nights, while the change in urban vibrancy from evening to late-night is smaller on weekends. In short, evening urban vibrancy density on weekdays is normally higher than that on weekends in central Shanghai, while late-night urban vibrancy density is greater on weekends than on weekdays.

Table 2Comparison of nighttime vibrancy indicators in different periods.

3.2. Types of nighttime urban vibrancy

3.2.1. Nighttime working vibrancy

This type of nighttime urban vibrancy is generated by people who work at night; its distribution is shown in Figure 7. The area with a high nighttime working vibrancy is spatially scattered and is distributed in a dispersed manner in the area around the Changshou Road and Changping Road, the Cao’an Business Center, the area near the Wuzhonglu Road and Lianhua Road, the Xujiahui area, the Helen International area, and the Zhongtie Tower. Additionally,there is a visible nighttime working vibrancy belt, which consists of the Dapuqiao area, the Shanghai TCMIntegrated Hospital, the Middle Huaihai Road, the Fuzhou Road, the West Beijing Road, and the Huashan Road.Nighttime working vibrancy density throughout this belt is evenly distributed.

3.2.2. Nighttime leisure vibrancy

Nighttime leisure vibrancy is generated by people who enjoy the night. That is, the people who neither live nor work in an area arrive for leisure at night. Compared with the other types of nighttime urban vibrancy, the distributions of nighttime leisure vibrancy are entirely different on weekdays and weekends (Fig. 8). Nighttime leisure vibrancy on weekdays is spatially scattered and distributed outside the city center, while nighttime leisure vibrancy on weekends is mainly concentrated in the city center. In addition, compared with weekdays, nighttime leisure vibrancy on weekends near the Outer Ring Expressway decreases sharply, and a high leisure vibrancy cluster appears around the Jing’an Temple, the Hengshan Road, the Xintiandi Road, and the East Huaihai Road. Overall,the distribution of nighttime leisure vibrancy is significantly affected by the days involved. On weekdays, high leisure vibrancy density levels mainly occur around the residential and employment areas. In contrast, on weekends,the leisure and entertainment areas in the city center attract more people to conduct activities and enjoy the nightlife,resulting it to be a high-level nighttime leisure vibrancy zone.

Fig. 1. Proportion of each communication event type of the mobile phone data of central Shanghai during 15 March-28 March in 2014 (a), and proportion of each communication event type in one day (b).

Fig. 2. Method of identifying nighttime urban vibrancy.

Fig. 3. Diagram of the research unit.

Fig. 4. Daily time-vibrancy curve for central Shanghai.

Fig. 5. Evening urban vibrancy density of central Shanghai (22:00-24:00) on weekdays (a) and weekends (b).

Fig. 6. Late-night urban vibrancy density of central Shanghai (02:00-04:00) on weekdays (a) and weekends (b).

Fig. 7. Nighttime working vibrancy density of central Shanghai on weekdays.

Fig. 8. Nighttime leisure vibrancy density of central Shanghai on weekdays (a) and weekends (b).

3.2.3. Nighttime floating vibrancy

Nighttime floating vibrancy is generated by people who do not have a place of residence or stable workplace in central Shanghai, which means that these people do not have specific behavior patterns, such as drivers, couriers,travelers, and tourists. The distribution of nighttime floating vibrancy (Fig. 9) on weekdays is similar to the distribution of overall nighttime urban vibrancy. On weekends, compared with the overall nighttime urban vibrancy,the nighttime floating vibrancy near the Shanghai Railway Station increases considerably, which causes this area to be the location with the highest nighttime floating vibrancy density in central Shanghai on weekend nights.

Fig. 9. Nighttime floating vibrancy density of central Shanghai on weekdays (a) and weekends (b).

In summary, significant and intense nighttime urban vibrancy usually occurs in areas with high service sector densities, diverse urban functions, and good accessibility. The different characteristics of the three nighttime urban vibrancy types may help planners create a more vibrant nightlife in cities and support the organization of urban functions and the configuration of different kinds of public facilities.

3.3. Impact of land use on nighttime urban vibrancy

The correlation analysis between nighttime leisure vibrancy and land use indicators is shown in Table 3. These correlations are all statistically significant. For the residential facilities and public and commercial facilities, the relationship between nighttime leisure vibrancy and proportion of land area is positive, while for the industrial facilities, the relationship is negative between them. With increasing buffer distance, the correlations between nighttime leisure vibrancy and area proportions within each buffer also become stronger. Specifically, with regard to the research grid itself, the correlation coefficient between nighttime leisure vibrancy and area proportion of public and commercial facilities is equal to 0.345, which is the same for residential facilities. However, as the buffer distances expand in size, the impact of area proportion of public and commercial facilities on nighttime leisure vibrancy is higher than that of residential facilities. We can conclude that for these two similar land use types, the type of land where public and commercial facilities are concentrated may generate higher nighttime leisure vibrancy than the type of land where residential facilities are concentrated.

Table 3Correlation analysis between nighttime leisure vibrancy and land use indicators at different buffer distances.

With increases in the buffer distance, the correlation coefficient between nighttime leisure vibrancy and the mixed land use of public and commercial facilities and residential facilities increases first and then begins to decrease when the buffer distance equals to 500 m. Moreover, when the buffer distance is 250 or 500 m, the correlation coefficient is greater than that when the buffer distance is 0 m. This illustrates that for nighttime leisure vibrancy, the mixed land use of public and commercial facilities and residential facilities within 500 m buffer distance is more important than the mixed use of the land lot itself.

4. Discussion and conclusions

Improving urban vibrancy is one of the focuses of recent research on the built environment. Due to data limitations, previous studies have rarely analyzed the distributions of urban vibrancy at the city level and have seldom discussed how different factors would affect the urban vibrancy at the micro block level. Based on mobile phone data, we conducted this study to show a time-vibrancy curve of central Shanghai, map the distribution of nighttime urban vibrancy, present the spatiotemporal changes in urban vibrancy, and ultimately determine how land use affects urban vibrancy at night.

The research results indicate that urban vibrancy in central Shanghai is periodic at a daily scale, which is consistent with people’s routine lives. Nighttime urban vibrancy is mainly concentrated in commercial facilities and business offices, and there is a belt with high-level nighttime vibrancy density within the Inner Ring Road. Business offices, hotels, entertainment and recreation areas, wholesale markets, and express services contribute most to the nighttime urban vibrancy in central Shanghai. Furthermore, we find that the correlations between nighttime leisure vibrancy and land use indicators, including the proportion of land area of public and commercial facilities, the proportion of land area of residential facilities, the proportion of land area of industrial facilities, and the mixed land use of public and commercial facilities and residential facilities, are all statistically significant and strongly correlated. Specifically, for residential facilities and public and commercial facilities, the relationship between nighttime leisure vibrancy and proportion of land area is positive, while for industrial facilities, the relationship is negative. As the buffer distance expands, the impact of the proportion of land area with public and commercial facilities on nighttime leisure vibrancy is higher than that of the proportion of land area with residential facilities. In terms of the mixed land use level of public and commercial facilities and residential facilities, with the expansion of buffer distance from 0 to 500 m, the correlation coefficient between nighttime leisure vibrancy and the mixed land use level of public and commercial facilities and residential facilities increases first and then gradually decreases. We also find that the mixed land use of residential facilities and public and commercial facilities in the 500 m buffer zone is more important than the mixed use level of a land lot itself.

In general, nighttime leisure vibrancy is positively correlated with the proportion of land area with public and commercial facilities and the proportion of land area with residential facilities. As the buffer distances expand,however, the correlation coefficient between nighttime leisure vibrancy and the proportion of land area with public and commercial facilities becomes larger than that between nighttime leisure vibrancy and the proportion of land area with residential facilities. In other words, public and commercial facilities can improve the nighttime leisure vibrancy more effectively than residential facilities. Therefore, when choosing an area to set off nighttime urban vibrancy, we should give priority to places with masses and concentrations of public and commercial facilities, or we can add this type of land use to a developing zone to create high nighttime urban vibrancy. Moreover, the correlation coefficient between nighttime leisure vibrancy and the mixed land use of public and commercial facilities and residential facilities in the 500 m buffer zone is greater than the coefficient between nighttime leisure vibrancy and the mixed use of the land lot itself. When defining urban planning projects and policies to promote nighttime urban vibrancy, we should pay more attention to the mixed land use of residential facilities and public and commercial facilities within 500 m of the land lot. That is, in order to improve nighttime urban vibrancy of a land lot, increasing the mixed land use level of its surrounding area is more influential than the mixed land use of the land lot itself.

These findings provide a deeper understanding of nighttime urban vibrancy and indicate that nighttime urban vibrancy is closely related to the land use types and their mixed use. Other research also indicates that mixed land use can promote urban vibrancy (Ning, 2016). The quantitative analysis of the impact of area proportions of land use and their mixed use on nighttime vibrancy supports land use planning and the allocation of public facilities and provides some evidences for policy-making to improve nighttime urban vibrancy. This research demonstrates the potential and value of mobile phone data for studying urban vibrancy. We can thus apply this approach to visually present and deeply analyze the human behaviors and activities across an entire city by using hundred-meter level grids, and use these data to assess urban spaces and built environments through a comprehensive, low-costs, and more objective process.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This research was supported by the National Natural Science Foundation of China (41771170).