In December 2019, the first case of pneumonia of unknown etiology, i.e., coronavirus disease 2019 (COVID-19), occurred in Wuhan South China Seafood Market (
Journal of Geographical Sciences, Volume. 30, Issue 12, 1985(2020)
Spatiotemporal analysis of COVID-19 risk in Guangdong Province based on population migration
Population migration, especially population inflow from epidemic areas, is a key source of the risk related to the coronavirus disease 2019 (COVID-19) epidemic. This paper selects Guangdong Province, China, for a case study. It utilizes big data on population migration and the geospatial analysis technique to develop a model to achieve spatiotemporal analysis of COVID-19 risk. The model takes into consideration the risk differential between the source cities of population migration as well as the heterogeneity in the socioeconomic characteristics of the destination cities of population migration. It further incorporates a time-lag process based on the time distribution of the onset of the imported cases. In theory, the model will be able to predict the evolutional trend and spatial distribution of the COVID-19 risk for a certain time period in the future and provide support for advanced planning and targeted prevention measures. The research findings indicate the following: (1) The COVID-19 epidemic in Guangdong Province reached a turning point on January 29, 2020, after which it showed a gradual decreasing trend. (2) Based on the time-lag analysis of the onset of the imported cases, it is common for a time interval to exist between case importation and illness onset, and the proportion of the cases with an interval of 1-14 days is relatively high. (3) There is evident spatial heterogeneity in the epidemic risk; the risk varies significantly between different areas based on their imported risk, susceptibility risk, and ability to prevent the spread. (4) The degree of connectedness and the scale of population migration between Guangdong’s prefecture-level cities and their counterparts in the source regions of the epidemic, as well as the transportation and location factors of the cities in Guangdong, have a significant impact on the risk classification of the cities in Guangdong. The first-tier cities - Shenzhen and Guangzhou - are high-risk regions. The cities in the Pearl River Delta that are adjacent to Shenzhen and Guangzhou, including Dongguan, Foshan, Huizhou, Zhuhai, Zhongshan, are medium-risk cities. The eastern, northern, and western parts of Guangdong, which are outside of the metropolitan areas of the Pearl River Delta, are considered to have low risks. Therefore, the government should develop prevention and control measures that are specific to different regions based on their risk classification to enable targeted prevention and ensure the smooth operation of society.
1 Introduction
In December 2019, the first case of pneumonia of unknown etiology, i.e., coronavirus disease 2019 (COVID-19), occurred in Wuhan South China Seafood Market (
This type of infectious disease, which spreads rapidly over a wide area, be it SARS - another coronavirus disease like COVID-19 - influenza caused by the avian influenza virus (
Given the human-to-human transmission of COVID-19, before effective vaccines or medicines are available, the most effective ways of preventing the spread are still the traditional measures, including isolating infected persons, tracking close contacts, and restricting mass gatherings (
Nevertheless, the potential risk of COVID-19 spread due to population migration has caught the attention of Chinese governments. After Wuhan “shut down” its entire city on January 23, 2020, to block the spread of the epidemic, other prefecture-level cities (referred to as “cities” in the rest of the paper) in Hubei Province implemented the same lockdown measure, including Huanggang, Jingmen, Xiaogan, and Ezhou. However, before the lockdown, more than 5 million residents or visitors had left Wuhan for places across China, becoming potential spreaders of the virus. The existing confirmed cases in other provinces are all related to this population migration.
With the rapid advancement and popularization of the internet and mobile devices, using big data to realize emergency situational awareness and decision-making support (
This study focuses on the COVID-19 risk posed by the population that came (or returned) to Guangdong during the 2020 Spring Festival period. It uses the population migration big data and the GIS spatial analysis technology, takes into consideration the risk differential among source regions of the population inflows as well as the socioeconomic disparities between the receiving regions of population migration, and constructs a model for analyzing the spatiotemporal distribution of the epidemic risk in Guangdong from the perspectives of risk input and diffusion. The model can be used to reveal the spatiotemporal evolution of the potential epidemic spread risk imposed by the migrating population from Hubei Province. The overall purpose of this study is to use Guangdong’s experience as an example to facilitate the development and implementation of prevention measures that are specific to different regions based on their risk classification and support decision making for preventing the spread of COVID-19 in China.
2 Research methods and data sources
2.1 Overview of the study area and research framework
Guangdong boasts the largest mobile population in China. It has a total out-of-province migratory labor force of 16.69 million, of whom 2 million are from Hubei. According to the internet population migration index, it is estimated that between January 25, 2020 and February 6, 2020, a total of 190,000 people came to Guangdong from Hubei. The period of January 25 - January 29 saw the most population migration. The main source cities of the migration included Wuhan, Jingzhou, Xiangyang, Huanggang, Xiaogan, Suizhou, and Xianning, accounting for 85% of the total population that moved from Hubei to Guangdong. The major population-receiving cities in Guangdong were Shenzhen, Dongguan, Guangzhou, Foshan, Huizhou, and Zhongshan, accounting for 91% of the inflow population from Hubei. As of March 5, 2020, the total number of confirmed COVID-19 cases in Guangdong reached 1351, second only to Hubei’s 67,592 cases. Cases imported from Hubei accounted for the vast majority of the confirmed cases, and diffusion was mainly due to family gatherings. Therefore, preventing importing cases from outside Guangdong and diffusion within the province is still the basic strategy of containing the epidemic in Guangdong.
Based on the above basic prevention and control strategy, this paper develops an analytic framework from the perspectives of risk input and diffusion: (1) Using the big data of population migration and the geospatial analysis technique, the paper introduces the indicators of imported risk and diffusion risk to develop the spatiotemporal analytical model for the COVID-19 risk in Guangdong. (2) Based on the interval distribution of the onset of the imported cases, a lag period is further incorporated into the model to construct a spatiotemporal analytical model that takes into consideration the time-lag effect. The goal is to analyze and forecast the evolution patterns and spatiotemporal configuration of the COVID-19 risk. The research framework is illustrated in
Figure 1.
2.2 Data sources
Data for the matrix of population migration from cities in Hubei to cities in Guangdong are based on mobile phone signaling data supplied by mobile service providers. Mobile phone numbers from Hubei were tracked for their movement to ascertain the population that migrated into each city in Guangdong. The incidence rates in cities in Hubei come from the data published daily by the National Health Commission of China. The sizes of the permanent and mobile populations and the numbers of health care institutions, hospital beds, health care workers, and registered physicians were obtained from the statistical yearbooks of the cities in Guangdong. The number of industrial firms was taken from the China Industrial Enterprise Database; the spatial distribution of the firms was obtained through spatial geocoding. Traffic-volume data were based on the real-time information regarding toll collection by the expressway network. The epidemiological data on confirmed cases are based on information published by Shenzhen Health Commission (SHC) as of March 5, 2020, regarding 416 patients who contacted COVID-19. The information includes the date when each patient came to Shenzhen and the date of illness onset.
2.3 Research methods
2.3.1 Imported risk
To measure the size of the inflow population and to classify the risk levels of source cities, this study uses the matrix of population migration from cities in Hubei to cities in Guangdong and the incidence rates of cities in Hubei as indicators, respectively. Given Hubei was the source region with the most serious epidemic, the inflow population from Hubei is viewed as the imported risk. Further, considering the spatial heterogeneity in the intensity of the epidemic across cities in Hubei, the incidence rates of these cities are used as the weight of the imported risk. The imported risk (
where
2.3.2 Time-lag effect of imported risk
The characteristics of COVID-19, e.g., infectivity during the latent period, are correlated with the number of newly confirmed cases and the number of new infections (
Explanation of some variables
Explanation of some variables
|
The lag between the import of a case and illness onset is denoted by
2.3.3 Diffusion risk
The diffusion risk is illustrated in
where $Risk_{diffusion}^{j}$ denotes the size of population (people) that may contract COVID-19 in city
where $\beta _{diffusion-1}^{i}$ denotes the susceptibility risk coefficient of city
Statistical analysis of the indicator variables
Statistical analysis of the indicator variables
|
2.3.4 Epidemic risk
The formula for epidemic risk is
where $Ris{{k}^{i}}$ denotes the overall epidemic risk of city
3 Spatiotemporal analysis of the COVID-19 risk in Guangdong from a geographic perspective
3.1 Time lag between case import and illness onset of imported cases
Since the novel coronavirus that caused the COVID-19 epidemic can be infectious during the latent period, after the imported risk is determined, we need to apply parameters that account for the time-lag effect to adjust the time distribution of the COVID-19 spread in Guangdong. Based on the review of imported cases in Shenzhen, we find that there is generally a common interval between the import of a case and the onset of illness. We examine this time interval for each imported case in Shenzhen and calculate the probability distribution of the time-lag period (
Figure 2.
3.2 Spatiotemporal distribution of imported risk
Taking into account the time interval of disease onset among the imported cases, the daily imported risk for each city in Guangdong is calculated using formula (1). Then the imported risk is adjusted through formula (2), whereby the probability distribution of the lag period of cases imported into Shenzhen is used as the weight for the adjustment. The results are presented in
Figure 3.
The daily imported risk can be roughly divided into three phases: (1) January 1 - January 10, 2020: the daily imported risk grew rapidly, especially in Shenzhen and Guangzhou. (2) January 10 - January 23, 2020: the daily imported risk fluctuated but overall was on the decline. (3) The daily imported risk declined dramatically. A before-and-after analysis intuitively indicates that the shutdown of Wuhan, a compulsory measure restricting population mobility, on January 23, 2020, played a significant part in breaking the spread of the pandemic.
The cumulative imported risk is shown in
Figure 4.
The results of the spatial analysis of the imported risk are shown in
Figure 5.
3.3 Diffusion risk coefficient
The spatial analysis of the susceptibility and prevention indicators indicates that both types of indicators have significant spatial heterogeneity. The heterogeneity of the susceptibility indicators is mainly demonstrated by the following facts: The permanent population is mainly clustered in the metropolitan cities of the Pearl River Delta and the core cities in eastern and western Guangdong. The mobile population is mainly concentrated in Shenzhen, Guangzhou, and Dongguan. Industrial firms are mainly located in the core cities of the Pearl River Delta metropolitan area surrounding the Pearl River estuary. And the traffic volume is mainly concentrated in the metropolitan cities of the Pearl River Delta.
The heterogeneity of the prevention indicators is demonstrated by the fact that health care workers are mainly concentrated in the metropolitan cities of the Pearl River Delta, especially Guangzhou, Shenzhen, Foshan, and Dongguan. Spatial heterogeneity in health care institutions and hospital beds is less significant. The spatial analysis of the diffusion risk coefficient indicates that the global Moran’s
Based on the analysis of the susceptibility risk coefficients and prevention risk coefficients, the diffusion risk coefficients for the cities across Guangdong are calculated based on formula (4). The diffusion risk coefficient has significant spatial heterogeneity due to the differences in population, industrialization, and traffic volume across the cities (
Figure 6.
3.4 Analysis of the spatiotemporal evolution of the epidemic risk
Based on formula (5) and the lag period of the imported cases, the daily diffusion risks of the cities across Guangdong are shown in
Figure 7.
The diffusion risk and imported risk show significant spatial heterogeneity. Specifically, the simulated epidemic growth process of the cities in Guangdong demonstrates temporal heterogeneity. During the early period of COVID-19 spread, cities with high population mobility and a high concentration of industrial firms had a higher growth rate of confirmed cases. For instance, from January 1, 2020, the daily diffusion risk of Guangzhou and Shenzhen grew through January 9, after which it started to decline, with fluctuations. During the middle and late stages of the COVID-19 spread, the diffusion risk declined rapidly. These results suggest that, while these cities had higher diffusion risks during the early phase of the epidemic spread, they also had higher prevention risk coefficients and could still reduce imported cases through effective prevention measures. Foshan, Dongguan, and Zhuhai experienced similar evolution patterns to those of Shenzhen and Guangzhou. Some cities, such as Huizhou and Zhongshan, had few imported cases and a low growth rate at the early stage of the epidemic, but they experienced significant growth in imported cases during the middle stage of the epidemic, indicating these cities had higher diffusion risks during this middle stage. Other cities had few imported cases during the early stage of the epidemic and did not experience rapid growth during the entire epidemic, mainly because they had lower diffusion risk coefficients.
Based on the daily diffusion risk of cities in Guangdong, the cumulative risk of these cities is calculated (
Figure 8.
Figure 9.
4 Conclusion and discussion
4.1 Conclusion
Population mobility, especially population inflows from epidemic areas, has been the main source of the spread of the COVID-19 epidemic. This paper focuses on the epidemic risk posed by Guangdong’s migratory population, utilizes big data on population migration, and applies the geospatial analysis technique to develop a model to perform spatiotemporal analyses of COVID-19 risk. The model takes into consideration the risk differential between the source cities of population migration, as well as the heterogeneity in the socioeconomic characteristics of the destination cities. It further incorporates a time-lag process based on the time period of imported COVID-19 cases. In theory, the model will be able to predict the trend of the evolution and spatial distribution of COVID-19 risk for a certain time period in the future, as well as provide support for advanced planning and targeted prevention.
The results of the simulated imported risks and diffusion risks indicate that the COVID- 19 epidemic in Guangdong passed the turning point on January 29, 2020 and entered a stable stage. Based on the probability distribution of the lag period of imported COVID-19 cases, the common practice of 14-day isolation of migratory populations adopted across China is effective for most confirmed cases or virus carriers. The lag period for a few cases or virus carriers is more than 14 days, so consideration should be given to extending the isolation period. The simulated results for the imported risk and diffusion risk indicate that there is significant spatiotemporal heterogeneity in the COVID-19 epidemic risk in Guangdong; the risk varies considerably between cities depending on the imported risk, susceptibility risk, and prevention risk. Shenzhen and Guangzhou are the high-risk regions; other cities in the Pearl River Delta, including Dongguan, Foshan, Huizhou, Zhuhai, and Zhongshan, have medium risks; and the cities in eastern, western, and northern ports of Guangdong that are outside of the Pearl River Delta have low risks.
4.2 Discussion
With a new infectious disease, it usually takes a long time for specific vaccines and medical treatment that fundamentally cure the disease to be developed. At this time, direct and effective emergency prevention and control measures are still isolation prevention and control strategies based on population flow restriction. Tian
The negative impacts of restricting population mobility on productivity, peoples’ daily lives, and the economy should not be ignored. Therefore, it is imperative to effectively grasp the spatiotemporal pattern of the spread and implement hierarchical prevention measures based on the spatial heterogeneity of the risk so as to achieve localized, targeted prevention and ensure the smooth operation of society. Currently, the most commonly employed model for COVID-19 risk is the susceptible-exposed-infectious-recovered model. Its main idea is to divide the population into the suspected high-risk population, the exposed population, the infected population, and the recovered population and identify the pattern of spread by examining the mechanism through which the disease is transmitted from one group to another. It has a high requirement for its parameters. This paper only considers the exposed population and infected population, which is a limitation of the paper: On one hand, this decision was due to the insufficiency of available data; on the other, the goal that guided this research was to realize spatial division of the COVID-19 risk and differentiated adoption of prevention levels, and this is why the paper focuses on infected population, an important indicator for evaluating the epidemic situation.
Further, this study incorporates the accurate incidence rates of the cities in Hubei, the lag period of the imported cases, the risk differential between source regions for population migration, and the variance in the socioeconomic situations of the destination regions, to make up for the above-mentioned deficiency in epidemic risk analysis. In addition, this paper does not consider the potential impact of undocumented infected cases. The existence of a large number of such asymptomatic virus carriers who can be super spreaders may cause rapid spread of the COVID-19 virus (
In summary, amid a major public health emergency such as COVID-19, emergency management and control usually incorporate expertise from multiple disciplines, such as pathology, epidemiology, geoinformatics, psychology, and behavioral sciences. This paper integrates epidemiology and geoinformatics and performs some seminal work in modeling the spatiotemporal distribution of COVID-19 epidemic risk. Limited by data availability, however, the analysis is only performed on prefecture-level cities. In fact, with the internet and the almost 100% coverage of cell phone signaling, it is completely possible, through geoinformation techniques, to accurately mine the data of key activity trajectories and activity hotspots of key groups of people and set the level of prevention measures on the inflow population using communities, enterprises, and institutions as the unit. This would alleviate the socioeconomic losses due to excessive prevention and control measures. Therefore, in the future, while ensuring that privacy is strictly protected and confidentiality protocols are followed, it is worthwhile to explore methods and systems for targeted prevention and control by making full use of large spatial data such as cell phone signals and improving the detection level of asymptomatic cases.
At present, the epidemic situation of COVID-19 has shown a global trend. An approach that integrates elements of coordination, classification, and collaboration Zhao
[4] CaoZhidong, ZengDajun, ZhengXiaolong et al. Spatio-temporal evolution of Beijing 2003 SARS epidemic. Scientia Sinica (Terrae), 40, 776-788(2010).
[9] DingSibao, ZhaoWei, XiangWei. Analyzing SARS: Geographical diffusion and hindrance in China. Human Geography, 19, 74-78(2004).
[11] GengMengjie, KamranKHAN, RenXiang et al. Assessing the risk of MERS importation from South Korea into cities of China: A retrospective study. Chinese Science Bulletin, 61, 1016-1024(2016).
[20] WangZheng, CaiDi, LiShan et al. On season risk of the prevalence of SARS in China. Geographical Reasearch, 22, 541-550(2003).
[26] ZhouChenghu, PeiTao, DuYuyan et al. Big data analysis on COVID-19 epidemic and suggestions on regional prevention and control policy. Bulletin of Chinese Academy of Sciences, 35, 200-203(2020).
Get Citation
Copy Citation Text
Yuyao YE, Changjian WANG, Hong’ou ZHANG, Ji YANG, Zhengqian LIU, Kangmin WU, Yingbin DENG. Spatiotemporal analysis of COVID-19 risk in Guangdong Province based on population migration[J]. Journal of Geographical Sciences, 2020, 30(12): 1985
Category: Research Articles
Received: Mar. 9, 2020
Accepted: Sep. 28, 2020
Published Online: May. 7, 2021
The Author Email: