Effects of COVID-19 on Urban Population Flow in China

Abstract

The COVID-19 epidemic has become a Public Health Emergency of International Concern. Thus, this sudden health incident has brought great risk and pressure to the city with dense population flow. A deep understanding of the migration characteristics and laws of the urban population in China will play a very positive role in the prevention and control of the epidemic situation. Based on Baidu location-based service (LBS) big data, using complex networks method and geographic visualization tools, this paper explores the spatial structure evolution of population flow network (PFN) in 368 cities of China under different traffic control situations. Effective distance models and linear regression models were established to analyze how the population flow across cities affects the spread of the epidemic. Our findings show that: (1) the scope of population flow is closely related to the administrative level of the city and the traffic control policies in various cities which adjust with the epidemic situation; The PFN mainly presents the hierarchical structure dominated by the urban hierarchy and the regional isolation structure adjacent to the geographical location.(2) through the analysis network topology structure of PFN, it is found that only the first stage has a large clustering coefficient and a relatively short average path length, which conforms to the characteristics of small world network. The epidemic situation has a great impact on the network topology in other stages, and the network structure tends to be centralized. (3) The overall migration scale of the whole country decreased by 36.85% compared with the same period of last year's lunar calendar, and a further reduction of 78.52% in the nationwide traffic control stage after the festival. (4) Finally, based on the comparison of the effective distance and the spatial distance from the Wuhan to other destination cities, it is demonstrated that there is a higher correlation between the effective distance and the epidemic spread both in Hubei province and the whole country.

Keywords: COVID-19; China; complex networks analysis; effective distance model; urban population flow.

Figure 1

The methodological framework of the geospatial network analysis. PFN: population flow network.

Figure 2

Comparison of trends of population mobility in China in 2020 and 2019. In order to facilitate comparative study, we convert the two sets of data in a unified way according to the lunar calendar, correspondingly, the data for 2019 are from 12 January 2019 to 4 April 2019. (a–e) represent five stages, which we distinguished earlier.

Figure 3

Spatial pattern evolution of PFN in China. Note: due to the asymmetric direction of the villages in PFN, there are two arcs between every two nodes. We set clockwise as the direction of population flow. Baidu Qianxi provide “Baidu migration index (BMI)” to indicate the size of inter-city migration, we convert BMI into the number of population movements using the actual number of inter-city/within-city population flows, provided by official WeChat accounts of Baidu Qianxi. When doing OD network analysis, we need to delete some unimportant links to intuitively and clearly present the information in this figure, so we take BMI equal to 0.003, that is, the actual migrant population is 100 (the rounded data) as the research threshold (a–e).

Figure 4

Factors correlated with confirmed COVID-19 cases. The relationship between the log-transformed geographical distance from Wuhan to nearly 300 cities and the log-transformed number of confirmed cases (a); (b–f), relationship over time between the log-transformed number of confirmed cases and the effective distance from Wuhan to destination cities.