key: cord-0828149-qe7jubor authors: Lu, Debin; Xiao, Wu; Xu, Guoyu; Ha, Lin; Yang, Dongyang title: Spatiotemporal patterns and influencing factors of human migration networks in China during COVID-19 date: 2021-10-14 journal: Geography and Sustainability DOI: 10.1016/j.geosus.2021.10.001 sha: 8439bd823c384764a9b1f8c64f7b49221cc5817e doc_id: 828149 cord_uid: qe7jubor The social lockdowns and strict control measures initiated to combat the COVID-19 pandemic have had an impact on human migration. In this study, big data was used to analyze spatial patterns of population migration in 369 Chinese cities during the COVID-19 outbreak and to identify determinants of population migration. We found that the overall migration intensity decreased by 39.87% compared to the same period in 2019 prior to the COVID-19 outbreak. COVID-19 severely affected human migration. The public holidays and weekends have impacted human migration from the perspective of time scale. The spatial pattern of China's population distribution presents a diamond structure that is dense in the east and sparse in the west, which is bounded by the hu line and the cities such as Beijing, Shanghai, Guangzhou and Chengdu as nodes to connect. There is a strong consistency between the population distribution center and the level of urban development. The urban human migration network is centered on provincial capitals or municipalities at the regional scale, showing a prominent "center-periphery" structure. COVID-19 dispersed the forces of human migration in time and changed the direction of human migration in space. But it did not change the pattern of national migration. The most critical factors influencing mass migration are income levels and traditional culture. This study reveals the impacts of major public health emergencies on conventional migration patterns and provides a scientific theoretical reference for COVID-19 prevention and control. COVID-19 has swept the world and disrupted people's everyday life You et al., 2020; Huang et al., 2020) . COVID-19 is highly contagious and can be transmitted from person to person (Ghinai et al., 2020) . When it broke out in China just before the Spring Festival, China underwent a large-scale and long-distance population movement. Fortunately, the measures of the "lockdown" were adopted under the scientific decision-making of the government and strong popular support. By the end of March 2020, the domestic pandemic had been stopped in Wuhan, effectively curbing the spread of COVID-19 (Tian et al., 2020; Sulaymon et al., 2021) . Critical phased results were achieved in preventing and controlling the pandemic (Huang, 2020; . However, the current international pandemic is still undergoing rapid evolution, and there are also some small-scale outbreaks domestically. Therefore, analyzing the direction and proportion of population movement before and after COVID-19 outbreaks and studying migration after the human intervention are of great value to the pandemic prevention and control. Throughout the past, various plagues such as smallpox (Brown, 1965) , measles (Dörig et al., 1993) , black plague (Herlihy, 1997) , cholera (Chin et al., 2011) and malaria (Murray et al., 2012) have occurred. In recent years, the Ebola virus (Feldmann and Klenk, 1996) , SARS (Severe Acute Respiratory Syndrome) (Rota et al., 2003) and H7N9 influenza (Zhou et al., 2013) have also emerged. However, the large-scale COVID-19 outbreak superimposed on the Chinese Spring Festival is rare. Human migration may cause the virus to spread geographically (Jia et al., 2020; and become a flashpoint to further spread the pandemic (Shi, 2020) . Because the traditional measurement of the floating population is based on small sample surveys and the coverage is small, it is difficult to track the population (Wei et al., 2018; Pan and Lai, 2019) . In addition, it is difficult to identify the proportion and direction of population movement. Thus there is little research on human migration under major public safety incidents like COVID-19. The relationship between human migration and COVID-19 remains unknown, especially the daily proportion and direction of human migration. In recent years, the rapid development of geographic information systems (GIS), remote sensing (RS) and global positioning system (GPS) technology, mobile internet, internet of things (IoT) and other technologies, has made the integration and continuous observation of human spatiotemporal behavior data, including geographic location, social attributes, movement trajectory, migration process and interaction mode possible. (Ke et al., 2015; Liu and Shi, 2016) . High-spatiotemporal-resolution human migration data represented by Baidu, Tencent and Gaode migration data have complemented conventional survey methods and are being applied to research on the regularity of human migration Cui et al., 2020) . Furthermore, the "flow space" has been successfully identified by interweaving physical infrastructure and virtual networks (Castells, 2001; Liu and Shi, 2016) , revealing the interaction of human migration between cities such as thepopulation migration related to the Chinese Spring Festival. Before the Spring Festival, people go to their hometowns and return to the city after the Spring Festival and Lantern Festival. The human migration data collected by big data methods can be superimposed with COVID-19 pandemic data to get a time and space perspective. Due to the large-scale COVID-19 outbreak, the human migration patterns were significantly different characteristics. Understanding the trajectory of population movement has a fundamental role in judging the distribution area of the pandemic and blocking transmission channels for pandemic control. It also provides other countries with human migration patterns under the interference of COVID-19 to facilitate the control of COVID-19. To reveal the characteristics of the short-term impact of COVID-19 on population migration in China, Baidu migration data were applied to construct a network of human migration between cities, and time series methods were employed to measure the scale and regularity of human migration. Chinese cities from 1 January 2020 to 31 March 2020. According to statistics, most cities cover more than 95%, and a small number of cities cover more than 85%. This helps to guarantee the validity and representativeness of the data. However, since the proportion of urban immigration or emigration is relative, it is impossible to compare cities directly. Therefore, it is necessary to multiply the comparable migration intensity index by the relative proportion of the immigration or emigration intensity to obtain a comparable rate between cities. In order to reveal the dynamic mechanism of population migration, the influencing factors of human migration reflect city size, income and the economic development level, including the number of COVID-19 infections, GDP (Ten thousand yuan), primary industry accounts for the proportion of GDP(%), the second industry accounts for the proportion of GDP (%), the third industry accounts for the proportion of GDP (%), population, the number of industrial enterprises above a designated size, average wages of employees in urban areas (Yuan), and urban area (km 2 Baidu migration reflects the ratio of population inflow and outflow between cities (Liu and Shi, 2016) . We set the number of people who emigrated from the city to on a certain day as , the number of people who immigrated from the city to as , and the sums of emigrated and immigrated population are ∑ 1 and ∑ 1 , respectively, so the emigration and immigration data provided by Baidu migration are = / ∑ 1 and = / ∑ 1 . Through the directed weighted asymmetric matrix (Chan, 2015; Zhao et al., 2017; Yang et al., 2020) , we constructed 96 daily 369×369 bidirectional matrices during the research period. (1) The proportion of net population inflow to the city and on a certain day is: Where > 0, it indicates the direction of population inflow and conversely < 0 indicates the direction of population outflow. The labor import and export cities can be judged according to the net population inflow in different periods before and after the holidays. Migration changes over time and time series methods can be used to reveal these changes (Taylor and Letham, 2018) . The time series of human migration are recorded as ( = 1,2, ⋯ , ), and is constituted of a trend term , seasonal term , holiday effect ℎ and residual term . We use a multiplicative combination model. If Here the trend model is where: is the growth rate; has the rate adjustments; is the offset parameter; and is set to − to make the function continuous. We can approximate arbitrary smooth seasonal effects with For the series with the year as the period ( = 365.25), = 10; for the series with the week as the period ( = 7), = 3. Here the holiday model is: where: represents the number of holidays; represents a period before and after holidays; and ~(0, 2 ). The COVID-19 pandemic coincided with the traditional Chinese "Spring The most active period of population flows in China is during the Spring Festival. Generally, people return home before New Year's Eve. When the sixth day of the New Year and Lantern Festival ends, people began to return to their workplaces. After the Lantern Festival, most migration has subsided (Wei et al., 2018) . To reveal the impact of holidays and weekends on migration during the pandemic, time series methods were used to analyze labor import and export. The results are shown in Figure 3 . As Figure 3 shows, migration is affected by holidays and weekends. After the net outflow of labor import cities reached its lowest value on New Year's Eve, it began to recover gradually. After reaching its peak on the sixth day of the New Year, it began to show weekly fluctuations. Population flows were low during working days and peaked on weekends. The overall net inflows was on the rise. On the contrary, the net inflow of labor export cities began to decline after reaching the highest value on New Year's Eve, and began to show weekly fluctuations after reaching the lowest point on the sixth day of the New Year. Overall, the net inflow shown a downward trend. Affected by the COVID-19 pandemic this year, the short-distance travel caused by visiting relatives, friends, and travel during the Spring Festival has not risen significantly. The fluctuation of human migration was mainly caused by returning to work. This "center-periphery" spatial model can be divided into two types. The first is the hierarchical structure dominated by the urban hierarchy. The megacities cities (Beijing, Shanghai, Guangzhou, and Chengdu) control regional population migration. Their radiation intensity and breadth of connections have broken through administrative boundaries. Those are the core cities in the province (city) and are connected to the population in a trans-provincial region. It has a strong appeal to the population of distant cities. The second is the proximity effect dominated by the geographical location. The population flow is controlled in non-core cities and provincial capitals, which affects the inflow of urban population in or around the province. Compared with Beijing, Shanghai, Guangzhou and Chengdu, the influence of other provincial capitals is relatively weak. Overall, key cities in the country (provincial capitals or municipalities) are the main nodes of population migration, with important functions of agglomeration, diffusion and transit, and they are in a key position in urban population migration . With changes in urban economic linkage activities and the improvement of traffic conditions, the attractiveness to population and radiation range of core cities will become wider and the spatial friction of economic linkages will become smaller and smaller (Pan et al., 2019) . Wuhan City is where China's first COVID-19 case was discovered. It had population migration during the Pre-Lockdown, however the number of people moving in and out was very small During-Lockdown and Post-Lockdown. There was no obvious "center-periphery" spatial structure feature caused by traffic control or the lockdown policy. On the contrary, although the intensity of population migration in the Pearl River Delta and the Yangtze River Delta declined compared with the same period in 2019, they still showed a trend of concentrated and contiguous net population inflows and presented an obvious employment-driven pattern. The density of the population migration between cities in a region often determines the city's economic development. As can be seen from Figure 6 , China's population migration network presents a spatial structure with Beijing, Shanghai, Guangzhou and Chengdu as the apex, Wuhan as the center and two axes running north and south. It is shown as a "diamond" structure that is supported by a cross-shaped skeleton formed by connecting the east-west axis. In addition, there are two migration routes that have broken through the "diamond" structure. One is the northwest route from Lanzhou to Urumqi and the other is the northeast route from Beijing to Shenyang to Changchun to Harbin (Pan et al., 2019; Shi, 2020) . Flow and direction are the basic attributes of population migration. Statistics show that the top ten cities carry about 30% of the total migration which means that nearly one-third of China's population movement occurrs in only ten cities. These However, northeastern and western border cities are less attractive with smaller and more confined migration. The proportion of the population outflow in these provinces is also low. Urban populations in or around the provinces migrate to provincial capital cities or regional cities. Large cities have a strong gravitational force and a large range of influence, affecting regional urban population migration. Figure During-Lockdown, the government strictly controlled population movement, but the demand for employment and for the return of people who travelled before the Spring Festival was urgent, and the migration of urban populations where the pandemic was not serious was still active. In Post-Lockdown, migration was affected by the number of COVID-19 infections. In cities with the severe COVID-19 numbers, control measures were more strict, the resumption rates lower, and the corresponding population migration was lower. In addition, population migration in small cities dispersed over time, while the immigration and emigration of the population between large cities offset each other, and urban population inflows was relatively low. Overall, COVID-19 delayed population migration, dispersed the proportion of population migration in time, and interfered with the direction of population migration spatially. However, income levels and the demand for employment were the main driving mechanisms of population migration after the Spring Festival. Income levels were an important mechanism for stimulating the population flows from third and fourth-tier central and western cities to first-tier eastern cities. In addition, people often return home for the New Year due to the traditional culture (Wei et al., 2018) . Fig. 7 The impact mechanism of population flows The population migration network is bounded by the Hu line, which is dense in the east and sparse in the west, with Beijing, Shanghai, Guangzhou, Chengdu and Wuhan as nodes. It shows a rhombus structure as the basic spatial feature of population migration at a national scale (Liu and Shi, 2016; Zhao et al., 2017; Jiang et al., 2017; Pan et al., 2019; . Due to the outbreak of COVID-19, large-scale social intercourse has been confined, which is unprecedented. This analysis provides a case study of migration patterns of China's urban population during COIVD-19. We found that COVID-19, as a public safety emergency, greatly reduced the strength of population migration. Although Baidu's migration data had advantages that traditional data cannot match, it also has limitations related to data acquisition (Liu and Shi, 2016) . Baidu migration has many users, but there are unconnected. The time for two positioning requests is fixed at eight hours and staying in the destination city for four hours is recorded as migration. However, this may result in: missing long-distance migration; repeated calculation errors; disassembled travel routes; and problem with fully identifying user travel. In addition, to protect user privacy, it is difficult to obtain social attributes such as occupation, gender, age, travel purpose and duration times (Pan et al., 2019) . Furthermore it is impossible to distinguish migrant workers, visiting relatives and students travelling home for their holidays. However, as a source of daily population flows between cities, it is unmatched by any other data source. In the future, multi-source geographic big data should be integrated to explore the social attributes of population migration. Baidu migration applies LBS technology to record the daily dynamic and real-time population flows between cities, becoming important for characterizing geographic behavior. Baidu's daily migration data from 1 January to 31 March 2020, were used in this manuscript. These data were divided into the Pre-Lockdown, During-Lock down and Post-Lock down periods. Time series and geographic network methods were adopted to study migration and its influencing factors under a major public safety incident. The 2020 national migration strength decreased by 39.87% compared with the same period in 2019. The pandemic severely affected population migration. After the "lockdown" in Wuhan, population flows were disrupted with outflows effectively controlled. In terms of timing, population migration was affected by holidays and weekends. The national population migration network is bounded by the Hu line which is dense in the east and sparse in the west, showing a diamond structure with Beijing, Shanghai, Guangzhou, Chengdu and Wuhan as the nodes. The urban population migration network is centered on the provincial capital or municipality, showing an obvious "center-periphery" structure. Generally, the attractiveness of provincial capital cities is weak, which affects the population migration to provinces. The attractiveness of large cities is strong and affects regional urban population migration. Income is the most important factor affecting large-scale population migration. COVID-19 dispersed the intensity of migration over time and changed the spatial direction of migration. The study helps to support a balanced national, regional economic development pattern for the control of COVID-19. The authors declare that no known financial interests or personal relationships influenced the research reported in this paper. Complications of smallpox vaccination Space of flows, space of places: Materials for a theory of urbanism in the information age. The city reader When human networks collide: the degree distributions of hyper-networks The origin of the Haitian cholera outbreak strain The spatial-temporal dynamics of daily intercity mobility in the Yangtze River Delta: An analysis using big data Development Research Centre of the State Council (DRC). (2019) Migrant people, changing city The human CD46 molecule is a receptor for measles virus (Edmonston strain) Marburg and Ebola viruses First known person-to-person transmission of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in the USA The Black Death and the transformation of the west Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. The lancet Inspirations of the Wuhan Lockdown on Global Urban Public Health under the Outbreak of COVID-19 Pandemic Population flow drives spatio-temporal distribution of COVID-19 in China Research on China's urban population mobility network: Based on Baidu migration big data Architecture and key issues for human space-time behavior data observation Public Pandemic Prevention and Control Services Based on Big Data of Spatiotemporal Location Make Cities Smarter Spatial pattern of population daily flow among cities based on ICT: A case study of "Baidu Migration Global malaria mortality between 1980 and 2010: a systematic analysis Spatial pattern of population mobility among cities in China: Case study of the National Day plus Mid-Autumn Festival based on Tencent migration data Characterization of a novel coronavirus associated with severe acute respiratory syndrome. science The impact of spring festival migration on coronavirus pandemic COVID-19 pandemic in Wuhan: Ambient air quality and the relationships between criteria air pollutants and meteorological variables before, during, and after lockdown Forecasting at scale. The American Statistician An investigation of transmission control measures during the first 50 days of the COVID-19 pandemic in China The rich-club phenomenon of China's population flow network during the country's spring festival The geographical destination distribution and effect of outflow population of Wuhan when the outbreak of the 2019-nCoV Pneumonia Spatiotemporal Patterns of Population Mobility and its Determinants in Chinese Cities Based on Assessment of monthly economic losses in Wuhan under the lockdown against COVID-19 Unprecedented disruption of lives and work: Health, distress and life satisfaction of working adults in China one month into the COVID-19 outbreak Spatial patterns and determinant factors of population flow networks in China: Analysis on Tencent Location Big Data Alter-based centrality and power of Chinese city network using inter-provincial population flow This work was sponsored by Natural Science Foundation of Henan (No. 202300410076).