Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Resilience and recovery of public transport use during COVID-19

Abstract

To better understand how public transport use varied during the first year of COVID-19, we define and measure travel behavior resilience. With trip records between November 2019 and September 2020 in Kunming, China, we identify people who relied on traveling by subway both before and after the first pandemic wave. We investigate whether and how travelers recover to their pre-pandemic mobility level. We find that public transport use recovered slowly, as urban mobility is a result of urban functionality, transport supply, social context, and inter-personal differences. In general, urban mobility represents a strengthened revisiting tendency during COVID-19, as individual’s trips occur within a more limited space. We confirm that travel behavior resilience differs by groups. Commuters recover travel frequency and length, while older people decrease frequency but retain activity space. The study suggests that policymakers take group heterogeneity and travel behavior resilience into account for transport management and city restoration.

Introduction

Since COVID-19 was recorded in December 2019 in Wuhan, China, the pandemic has been repeatedly reported in cities worldwide. In response to the pandemic, policy interventions including lockdowns, transport system shutdowns, travel restrictions, social distancing, and self-isolation have been imposed by national, regional, and local governments reducing mobility and disrupting social activity1.

With this background, the study of human movement or population mobility has attracted scholars’ attention, as it is highly useful to track infection, identify super-spreading conditions, simulate imported cases, predict pandemic waves, and test intervention effectiveness2,3,4,5,6,7,8. Previous studies have correlated human movement and pandemic dynamics at multiple spatial scales9,10,11,12,13,14,15,16. Although several studies observed a specific city and made predictions during COVID-198,17,18, they have not measured how human movement varies and recovers. In transport, most studies reported aggregated analyses19,20, or surveyed behavior21, attitudes22, or preferences23, but long-term group mobility has not been tracked systematically. Some studies have confirmed a significant modal shift from public transport to other transport modes, and decreasing public transport shares vs. private vehicles have been a challenge for carbon emission reduction and sustainable transportation development18,24,25. Still, how public transport use recovers with the city-life restoration is unclear. To fill this research gap, we analyze how group mobility in the Kunming subway system recovers to the pre-pandemic level. Hence, we focus on transit trips between November 2019 and September 2020 in Kunming, a provincial capital in China (Fig. 1a).

Fig. 1: Study area, analytical framework, and data selection.
figure 1

a Location of Kunming and its subway system. b Travel behavior resilience of public transport use. c The trip rate in Kunming subway system between January and September 2020. (The supply function curve indicates how the public transport system reopened and adapted during the COVID-19 pandemic. The resilience triangle can be calculated with the degree of mobility change, and the periods of reduction and recovery. The trip rate is calculated by the ratio of all transit trips in each week over the weekly average number of trips in November 2019, namely before the COVID-19 pandemic. As we tracked individual trip records, frequent travelers are those who accessed the subway system frequently in November 2019, and their travel frequency recovered in September 2020. The remaining travelers in November 2019 are tracked as infrequent travelers. Note that data in June and August 2020 are unavailable.).

After lifting lockdown interventions, the restoration of city life can be described from both supply and demand perspectives. On the supply side, public facilities such as shopping malls, public libraries, and public transport systems reopened, eventually providing the same service level as before COVID-19 to ensure basic urban function. Furthermore, the capability of public transport systems to adapt and reopen during and after a natural disaster or system shutdown indicates system resilience from the supply side26,27, and system function loss during this process has been measured with a resilience triangle28. The recovery of public transport supply indicates that subway systems reopened and train frequency gradually recovered to pre-pandemic levels (see the supply function curve in Fig. 1b).

From the demand side, full recovery means after the city lockdown people have returned to previous behavior or have adapted interventions (such as mask-wearing) in their daily life to the extent they behave as they did prior to the pandemic. This demand for recovery is related to “behavioral resilience” in psychology, which means positive adaptation after negative socioeconomic changes, traumatic life events, community grief and loss, and environmental pressure29. During the recovery period, travel behavior is affected by service facilities, social context, transport supply, and individual psychological state.

Travel behavior occurs over time and space, motivated by individual extrinsic requirements and intrinsic needs30, and varies based on socioeconomic background, preferences, and transport supply. In brief, inter-personal differences make travel behavior spatially and temporally diverse. Traditional travel surveys usually include these factors, but it is difficult to retrospectively collect accurate data after the fact, which limits a diagnosis of travel behavior change due to events. This paper defines the concept of travel behavior resilience and applies it to COVID-19, depending on both individual preferences in the before times and individual experience during the pandemic, showing how travel behavior co-evolves with the restoration of city-life and reopening of the public transport system.

With the onset of COVID-19 in a city, human mobility rapidly reduces as both the transport system supply is contracted (to discourage travel and because staff themselves are affected) and the willingness to travel declines, and it gradually recovers after the pandemic dissipates. The recovery of transport system supply can be explained with models from engineering resilience, while from the demand side, resilience can be understood with social-ecological resilience31. For public transport use, travel behavior resilience can be understood as the process by which transport supply and demand co-evolve and achieve temporary equilibrium32. In this process, some temporary equilibrium points can be achieved as shown in Fig. 1b This process begins when the level of public transport use starts to reduce and ends when it stops increasing and stays in a stable status. The stable status can be captured when public transport use recovers to pre-pandemic levels (i.e., the pre-existing equilibrium) or co-evolves with the supply to a new equilibrium. In the latter case, people’s travel structure may change, as avoidable travel can be replaced by online shopping, working from home, etc. Also, some trips by public transport can be replaced by other travel modes. Here, we offer a longitudinal analysis tracking mobility group differences for the former case. We measure whether and how public transport use recovers to its pre-pandemic mobility level. As Fig. 1b shows, travel behavior resilience can be measured with a triangle denoting to what degree mobility (amount of travel) drops, compared to the pre-pandemic level. We draw the resilience triangle with mobility reduction and recovery periods, and the quantity of mobility change. In general, a smaller triangle indicates greater resilience, the capability to adapt to a particular negative disruption.

During a pandemic, people first decide whether to revisit some, or all, of the same places as before, thus restoring their activity space and travel duration, or to visit unfamiliar places instead. We posit that if travel duration restores faster than activity space, it indicates people are returning to their most frequent activities (such as commuting) but not the infrequent social and discretionary shopping activities at dispersed locations. In Fig. 1b, the mobility curve can be drawn with diverse indicators including trips, travel days, total trip distance, activity space, and places visited during the study period.

As aforementioned, during the recovery stage, we expect travel behavior to be more cautious and confined to familiar places. Commuter travel is urgent and mostly fixed, and many governments aim to maintain commuting trips of essential workers as much as possible to sustain urban services and city functionality. In contrast, vulnerable groups (such as older people) tend to decrease their travel in order to minimize pandemic exposure. Therefore, we posit the following hypotheses to investigate travel behavior resilience during the COVID-19 period.

Hypothesis 1.

The indicator denoting whether to engage in a trip recovers faster than other mobility indicators such as trip distance, traveled days, activity space, and visited locations (stations).

Hypothesis 2.

Mobility indicators measuring the spatial dimension such as activity space recover slower than others measuring travel duration such as trip distance.

Hypothesis 3.

The trip proportion of revisiting familiar places increases, while that of exploring unfamiliar places decreases.

Hypothesis 4.

Commuters present greater travel behavior resilience, while the elderly show lesser resilience.

As suggested in recent studies33,34,35,36,37, geolocation data and mobility data can correlate human activities with both the first wave of pandemic transmissibility and second-wave infection. Transit trip records can be used to investigate travel behavior resilience. The reasons are below. During COVID-19, intercity mobility encountered more restrictions than intra-urban, as long-distance human movement can amplify localized pandemic outbreaks9,11,38. After the COVID-19 outbreak, the order in which travel restrictions were lifted tended to be intra-urban, regional, and then global. Also, the infection risk by walking, cycling, or private driving (and even more importantly, the perceived risk) is significantly lower than public transport due to its collective nature of mobility in a confined space and the infection mechanism of COVID-1939. Consequently, a significant modal shift from public transport to other transport modes has been seen18,24,25.

This paper focuses on the study of Kunming (Fig. 1a), a gateway city and international trade hub in Southwest China with a 2020 population of about 7 million. In 2020, the Kunming subway delivered 22.22% of intra-urban trips40. It is a representative city with a developing transit system and substantial transit demand. Kunming has been involved in the pandemic when COVID-19 cases were confirmed continuously between 22 January and 19 February 2020 (Fig. 1c), and it employed lockdown interventions between 28 January and 29 February 2020 (Supplementary Note 1).

As people who rely on public transport are frequent travelers that are the crucial group with regular spatiotemporal patterns41, this paper focuses on a longitudinal analysis of them. To this end, we collected trips paid by smartcards or virtual payment platforms (e.g., mobile phone applications) from the Kunming subway system between November 2019 and September 2020 (Supplementary Note 2).

Trips from 1 to 28 November 2019 are used to identify frequent travelers before COVID-19, representing the pre-pandemic mobility level. According to weekly calculations, half of the trips in the Kunming subway system are generated by passengers who access the subway 3 days per week or more. We define frequent travelers as those who traveled at least 3 days each week during the four consecutive weeks before the pandemic. Comparing the same individual’s trip records in September 2020, those who still traveled at least 3 days each week are defined as subjects. This allows us to investigate whether and how their mobility recovers. Overall, this study focuses on 16,403 subjects who generated over 3.5 million trip records in ten months. It is substantial sample size for a longitudinal study of travel behavior and urban analytics42.

The number of trips by all subjects fell during the COVID-19 outbreak, when random wandering, as well as repeated travel, significantly declined (Fig. 1c). In the post-outbreak period, the number of trips in September 2020 for infrequent travelers was about 40% of the pre-pandemic mobility level, while frequent travelers had nearly recovered. Recall the concept of travel behavior resilience, we focus on the analysis of frequent travelers to better understand how public transport use recovers. As shown in Fig. 2, 21.87% of frequent travelers identified before the pandemic still frequently accessed transit service in September 2020 using the same identity (smartcards or mobile phones). We classify subjects into four groups, commuters (36%), elderly (5.4%, above 60 years old), students (4.2%), and others (54.4%). The sampling process can be seen in Supplementary Note 3.

Fig. 2: Distributions of frequent travelers in Kunming subway system.
figure 2

a November 2019, b September 2020, c Subjects during COVID-19. Frequent travelers are those who traveled at least 3 days each week for 4 consecutive weeks. Subjects are frequent travelers in both November 2019 and September 2020. They are classified into four mobility groups and their payment methods are recorded.

Results

Sequential variations of public transport use

During COVID-19, public transport use experienced three phases, drastic reduction, rapid growth, and stabilization. First, a drastic reduction presented according to a significant decline in mobility indicators (Figs. 3 and 4a), after the first case was confirmed on 22 January 2020 in Kunming. The drastic reduction period took about 3 weeks. It is worth noting that mobility indicators in Fig. 3b–f are calculated based on the number of frequent travelers who traveled each week (Fig. 3a). Second, after the city restarted, regular activity mobility indicators began to increase. At the end of Week 10 (21 March 2020), the rates of mobility indicators all recovered to over half of the pre-pandemic level, except for students’ as schools were all closed due to the pandemic. Third, the rapid growth ended as the recovery rate of mobility indicators gradually restored to the pre-pandemic level. To summarize, these sequential variations reinforce a recent study43, which confirms the sampling validation with discontinuous data.

Fig. 3: Mobility indicators by groups in 2020.
figure 3

a subjects who traveled during the pandemic, b trips, c traveled days, d total trip distance, e activity space, and f visited stations. All indicators are reported by weeks and compared to the pre-pandemic mobility level (i.e., weekly average in November 2019). Note that bar plotting represents the pre-pandemic mobility level.

Fig. 4: t test analysis and trip distribution by station rank.
figure 4

a Week-to-week t test of mobility indicators for commuters, elderly, students, and other frequent travelers. For example, w1-0 indicates a t test analysis comparing week 1 (from 19 to 25 January) with week 0 (from 12 to 18 January 2020). Individual trips in continuous 19 weeks between January and May are used. Trips from 1 to 28 July and those from 1 to 28 September are used, and they are weeks 24–27 and weeks 32–35. N, D, Q, T, A, and S denote variations of subjects who traveled, traveled days, trips, total distance, activity space, and stations visited per week. ***p < 0.001, **p < 0.01, *0.01 ≤ p< 0.05. b Trips by station rank. c Trip proportions by station rank. b and c are drawn with monthly analysis. For a month, we sort out trips by subjects. At the individual level, we rank stations by trips and calculate/the trip proportion generated at each station. We obtain the Top 4 stations for each subject. b and c show that individual trips usually concentrate on their own top 1 and 2 stations.

Resilience by mobility indicators

Compared to the pre-pandemic mobility level, the proportion of subjects who traveled after the first wave of the pandemic varies first, while other indicators are concomitant and follow in the travel willingness variation. Evidence can be seen below. During the early sharp reduction, the proportion of subjects who traveled went down faster than other indicators, and the proportion decreased to 3.34% (Fig. 3), implying the preliminary impact of the COVID-19 outbreak on willingness, permission, or ability to travel. Once travel began to rapidly restore, the p value of subjects traveled is higher than other mobility indicators in the t test analysis (in Supplementary Tables 1–5). During the stabilization, the recovery rate of subjects who traveled returned to the pre-pandemic level in Week 23 (from 1 September 2020), followed by traveled days in Week 25, trips, and total distance in Week 26. In brief, these mobility indicators present stronger follow-on nature and three-phase variation, which verifies Hypothesis 1.

Activity space and stations visited did not return to the pre-pandemic mobility level by 28 September 2020, the end of the study period, their recovery rates are 94.72% and 97.43%. For each mobility group, we formulate 23 week-to-week models in the t test analysis. In a longitudinal comparison, activity space models were less statistically significant than other indicators, with only 10 out of 23 models for commuters, 5 for the elderly, 4 for students, and 10 for others. In a parallel comparison, the p value of activity space is less than other indicators. These phenomena corroborate Hypothesis 2.

Revisitation law during COVID-19

Previous studies find that most urban movement is revisiting, including trips such as commuting, going to school, or shopping at the nearby supermarket, rather than exploration44,45. To measure the revisiting tendency of public transport use during COVID-19, we analyze the number of times each station was visited. For each subject, we rank stations by trips boarding/alighting at each station and also calculate the proportion generated at each station. Their monthly distribution can be seen in Fig. 4b, c, namely the pre-pandemic distribution (November 2019), the distribution during the COVID-19 (March 2020), and September 2020. Overall, the Top 4 stations for each frequent traveler receive ~95% of all trips, of which 71.5% of all trips are generated at the Top 2 stations, indicating a very high revisiting tendency by urban transit users. Moreover, the revisiting tendency strengthened during recovery, as trips were more concentrated at the Top 2 stations than pre-pandemic (i.e., 78.3%). Moreover, while trips/week in September 2020 were restored to 103.65% of pre-pandemic levels (Fig. 3), a more concentrated pattern can be seen, with the activity space at only 94.72% of the before times. It indicates that the destination diversity of transit trips declined. These findings are consistent with Hypothesis 3. We conclude a revisitation law during the COVID-19 pandemic, namely the revisiting tendency has been strengthened.

Resilience comparison between mobility groups

Travel behavior resilience differs by mobility groups (Fig. 3 and 4a). Commuters’ six mobility indicators show an increasing tendency earlier with higher p values in Week 4 (from 9 February 2020). Furthermore, commuters’ mobility continuously increased for ~6 weeks, indicating longer rapid growth. We further identify each commuter’s station nearest their workplace with trips in November 2019 according to individual spatiotemporal regularity of commuting trips. In addition, the standard deviations of total distance and activity space increase during the stabilization phase (Table 1), indicating people’s activity engaged in a larger space once the COVID-19 pandemic was under control. Moreover, commuters tend to engage in more trips within a smaller activity space, while the elderly and others tend to engage in fewer trips within a larger activity space.

Table 1 Mobility indicators during three sequential phases by groups.

Recovery duration and pre-pandemic mobility level

According to Table 1, the number of trips and travel days in a week by commuters tend to be more stable with a smaller standard deviation. The traveled days by commuters returned to about five days in the stabilization phase, whereas it was 3 days/week in the reduction and 2 days/week in the rapid growth. Although the traveled days by older people and others also increased, they are ~4 days/week in the stabilization.

Furthermore, we analyze the proportion of commuting trips over the total number of trips by an individual each week and conduct a week-to-week analysis (Table 2). The proportion of commuting trips significantly increased for 11 consecutive weeks until the end of April 2020, a longer rapid growth than indicators in Fig. 4a. However, the proportion of commuting trips was 80.60% before the pandemic, while it remained at 71.08% in the stabilization period. This phenomenon indicates that intervention policies such as work from home and flexible schedules can remain effective. For frequent travelers, commuting is still the largest component of public transport use even during the COVID-19 period. As people who rely on public transport are usually low to medium-income residents or essential workers, their commuting recovery is of necessity and urgency.

Table 2 t test analysis of commuting trips.

As COVID-19 puts older people at a higher risk of infection34,46, their travel behavior tends to be more cautious, presenting a lower level of resilience. In comparison with commuters or other frequent travelers, the rapid growth of elderly mobility triggered two weeks later and remained lower. Meanwhile, the recovery rates of indicators for the elderly were ~80% slower (Fig. 3). Moreover, the Kunming subway offers elderly concessions for two groups (ages between 60 and 69 and those older than 69). We find that the older group, who relied on traveling by transit more before the pandemic, presents a lower level of travel behavior resilience.

Recall the resilience triangle of public transport use (Fig. 1b), a comparison of mobility reduction and the recovery period can offer insights into the lower resilience of the elderly. According to curves of subjects who traveled, commuters experienced shorter reduction and recovery periods than the elderly. As for the quantity of mobility change, the proportion of commuters who traveled after the first wave of COVID-19 decreased to 4.40% of the original value, while that of the elderly dropped to 2.60%. Hence, we conclude that commuters have a smaller resilience triangle than the elderly. Similarly, days, trips, total distance, and stations visited by commuters also present smaller resilience triangles. These phenomena corroborate Hypothesis 4. In addition, the mobility level of students’ public transport use eventually tracked well, following the school closure in spring, subsequent reopening, and fall opening, which can be a control group when trip records are not continuous.

According to the resilience triangle of travel behavior (Fig. 1b), we select subjects whose trips, total distance, or activity space have decreased to zero since the COVID-19 outbreak, and then track when the mobility indicators of these subjects recover to the pre-pandemic level respectively. With this selection, the shorter the recovery duration, the greater the resilience. The correlations between recovery duration and pre-pandemic mobility level are shown in Fig. 5. Overall, the recovery duration decreases with the trips/week before the pandemic, and so do total distance and activity space. These phenomena denote a higher level of pre-pandemic mobility needs longer restoration duration, as people focus on maintaining essential trips. To some extent, the study aligns with previous findings of increasing car-based trips and decreasing transit trips during the COVID-19 pandemic25,31,47.

Fig. 5: Correlations between travel recovery duration and pre-pandemic mobility level.
figure 5

ac After the first COVID-19 wave, subjects whose trips in a week decreased to zero are selected (n = 15,395). Their average number of trips/week before COVID-19 are calculated. The recovery duration is defined as from the week when trips decreased to zero to the week when trips recover to the pre-pandemic level. The regression analysis is conducted with trips/week before COVID-19 (x) and the recovery duration (y). We obtain ycom = 0.49x + 8.13, R2 = 0.71; yeld = 0.42x + 10.55, R2 = 0.82; yoth = 0.22x + 10.74, R2 = 0.48. d, f Subjects whose total distance decreased to zero during COVID-19 are selected (n = 15,402). The regression is performed with total distance/week before COVID-19 (x) and recovery duration (y). The results are given below ycom = 0.01x + 11.61, R2 = 0.44;yeld = 0.02x + 13.06, R2 = 0.17; yoth = 0.01x + 12.41, R2 = 0.39. gi subjects whose activity space decreased to zero during the first pandemic wave are selected (n = 15,402). The regression is performed with activity space/week before COVID-19 (x) and recovery duration (y). We obtain ycom = 0.05x + 12.80, R2 = 0.71; yeld = 0.04x + 9.29, R2 = 0.12; yoth = 0.03x + 12.88, R2 = 0.43. Note that confidence intervals at the 95% level are shown.

In terms of mobility group difference, the recovery duration of trips and total distance by the elderly is longer, while their activity space restores faster. This indicates that older people tend to decrease travel frequency, but not the confines of their activity space in the post-COVID-19 period. Older people prefer visiting public parks and medical facilities more than other groups48, and their trips are of less urgency and fixity, their travel recovery tends to be slower, presenting the lower resilience.

In contrast, commuters’ travel frequency and total distance recover faster than their activity space. These phenomena indicate that commuters retain their commuting trips, but reduce their social activity and travel purpose diversity.

Discussion

This research proposes a general measurement of travel behavior resilience according to group mobility curves, which is used to study how public transport use recovers as people restore their daily urban activities during a pandemic. We examine the travel behavior resilience of frequent public transport users in Kunming, China over a 10-month period. First, even in a city that only experienced one COVID-19 infection wave, the recovery of public transport use took about half a year. Although this paper limits the investigation to a specific city in China, the experience is still valuable and it finds the necessity of long-term interventions before the recovery of human mobility49. Second, the recovery of public transport use by different mobility groups took different amounts of time. In groups (such as older people) at high risk of infection the pandemic disruption, the destination diversity significantly declined. This implies the revisiting nature of urban mobility has been intensified, while exploration behavior has weakened. Consequently, individual activity space tends to be smaller, decreasing to 95% of pre-pandemic levels.

With the findings above, our research has policy implications for urban studies, transport management, and social science during COVID-1918. According to the travel behavior resilience analysis, we suggest that the travel restriction or intervention during the COVID-19 pandemic should consider the diverse needs of mobility groups. Considering the slow recovery of urban mobility and decreasing diversity of trip purposes, policy should consider whether and how to stimulate intra-urban mobility in order to facilitate city restoration. These policies should differ by age group, as retirees have very different travel patterns and virus susceptibility than younger groups. Furthermore, the case study of Kunming offers a pandemic experience, when most cities experienced multiple COVID-19 waves, with a discrepancy between policy intervention and severity of infection and mortality. Currently, the public and government acceptability of stringent restrictions on mobility and public transport operator has been depleted to varying levels. Under this context, the Kunming experience is still representative of cities in China and other countries that stick to the stringent zero-COVID restrictions, and it may be useful to understand to what extent and how long the stringent restrictions affect public transport use.

This research suggests future directions. Although travel behavior resilience can be defined as an equilibrium between travel demand and supply, the analysis only focuses on cases when the public transport use of frequent travelers recovered to the pre-pandemic level. We could further develop a quantitative method of defining the re-equilibrium during COVID-19 between travel demand and supply if the travel demand has not recovered to the pre-pandemic level. Here, we only confirmed the slow recovery of public transport use, but how it affects traffic congestion and carbon emission is unclear. Its influence on carbon emission needs more assessment, which is significant for sustainable urban transportation in the post-pandemic period. One limitation occurs, as we only obtained data over a 10-month period. A dataset over 12 months could be better to avoid the influence of seasonable variability.

Methods

We collect mobility indicators at the individual level. Individual activity space is defined by the minimum convex polygon confined by all alighting and boarding stations visited50, and the number of distinct alighting and boarding stations visited is counted as stations visited. A weekly t test analysis is conducted for subjects by indicators.

We calculate mobility indicators by weeks. At the individual scale, in a week, we calculate the number of traveled days, the number of trips, the total trip distance, the activity space, and the number of stations visited. Individual activity space is defined by the minimum convex polygon confined by all alighting and boarding stations visited50, and the number of distinct alighting and boarding stations visited is counted as stations visited. This calculation is conducted with data between 1 and 28 November 2019, and weekly average indicators are obtained to present pre-pandemic mobility levels. For the data between January and May, July, and September 2020, the weekly calculation is conducted. To further analyze the collective pattern by mobility groups, we used the following methods.

For a subject i who accessed the subway system in the week studied, we set wi = 1, otherwise wi = 0, a notation list is in Supplementary Table 5. For a mobility group, the proportion of subjects who traveled is defined as subjects who accessed the subway in the week studied over the total number of subjects n, it is given by

$$N = \mathop {\sum }\limits_{i = 1}^n w_i/n,$$
(1)

where n = 16403.

Rate of change of traveled days D is defined as the average traveled days by subjects traveled in the week studied over the average traveled days by all subjects in the week before the pandemic, it is given by

$$D = \frac{{\mathop {\int }\nolimits_{i = 1}^n d_i/\mathop {\int }\nolimits_{i = 1}^n w_i}}{{\mathop {\int }\nolimits_{i = 1}^n d_i^0/n}}$$
(2)

where \(d_i^0\) is the number of traveled days by subject i in the week before the pandemic, the number of traveled days by a subject in the week studied is di. Note that we use data between 1 and 28 November 2019 to calculate weekly average indicators.

Similarly, the rate of change for trips in a week is given below

$$Q = \frac{{\mathop {\int }\nolimits_{i = 1}^n q_i/\mathop {\int }\nolimits_{i = 1}^n w_i}}{{\mathop {\int }\nolimits_{i = 1}^n q_i^0/n}}$$
(3)

where \(q_i^0\) is the number of trips in a week by a subject before the pandemic, the number of trips by a subject in the week studied is qi. Note that we also use data between 1 and 28 November 2019 to calculate weekly average indicators.

The rate of change for trip distance is given as

$$T = \frac{{\mathop {\int }\nolimits_{i = 1}^n t_i/\mathop {\int }\nolimits_{i = 1}^n w_i}}{{\mathop {\int }\nolimits_{i = 1}^n t_i^0/n}}$$
(4)

where \(t_i^0\) is the total distance by a subject in a week before the pandemic, the total distance by a subject in the week studied is ti.

The rate of change for activity space is given as

$$A = \frac{{\mathop {\int }\nolimits_{i = 1}^n a_i/\mathop {\int }\nolimits_{i = 1}^n w_i}}{{\mathop {\int }\nolimits_{i = 1}^n a_i^0/n}}$$
(5)

where \(a_i^0\) is activity space by a subject in a week before the pandemic, the activity space by a subject in the week studied is ai.

The rate of change for stations visited is given by

$$S = \frac{{\mathop {\int }\nolimits_{i = 1}^n s_i/\mathop {\int }\nolimits_{i = 1}^n w_i}}{{\mathop {\int }\nolimits_{i = 1}^n s_i^0/n}}$$
(6)

where \(s_i^0\) is the number of stations visited by a subject in a week before the pandemic, the number of stations visited by a subject in the week studied is si.

Data availability

The data generated and analyzed during this study are described in the data record https://doi.org/10.6084/m9.figshare.19771396. The data of weekly mobility indicators are openly available as part of the figshare metadata record in the file of “mobility indicators”. Mobility indicators of commuters, elderly, students, and others are available. For each mobility groups, whether an individual traveled or not, trips, traveled days, trip distance, activity space, and visited stations in each week are recorded. The authors confirm that no individual privacy information is included.

Code availability

There are no openly available code packages generated in this study.

References

  1. Wei, Y. et al. Spread of COVID-19 in China: analysis from a city-based epidemic and mobility model. Cities 110, 103010 (2021).

    Article  Google Scholar 

  2. Kraemer, M. U. G. et al. The effect of human mobility and control measures on the COVID-19 epidemic in China. Science 368, 493–497 (2020).

    CAS  Article  Google Scholar 

  3. Jia, J. S. et al. Population flow drives spatio-temporal distribution of COVID-19 in China. Nature 582, 389–394 (2020).

    CAS  Article  Google Scholar 

  4. Chang, S. et al. J. Mobility network models of COVID-19 explain inequities and inform reopening. Nature 589, 82–87 (2021).

    CAS  Article  Google Scholar 

  5. Xiong, C. et al. Mobile device data reveal the dynamics in a positive relationship between human mobility and COVID-19 infections. Proc. Natl. Acad. Sci. 117, 27087–27089 (2020).

    CAS  Article  Google Scholar 

  6. Menkir, T. F. et al. Estimating internationally imported cases during the early COVID-19 pandemic. Nat. Commun. 12, 311 (2021).

    CAS  Article  Google Scholar 

  7. Tian, H. et al. An investigation of transmission control measures during the first 50 days of the COVID-19 epidemic in China. Science 368, 638–642 (2020).

    CAS  Article  Google Scholar 

  8. Loo, B. P. Y., Tsoi, K. H., Wong, P. P. Y. & Lai, P. C. Identification of superspreading environment under COVID-19 through human mobility data. Sci. Rep. 11, 4699 (2021).

    CAS  Article  Google Scholar 

  9. Colizza, V., Barrat, A., Barthélemy, M. & Vespignani, A. The role of the airline transportation network in the prediction and predictability of global epidemics. Proc. Natl. Acad. Sci. USA 103, 2015–2020 (2006).

    CAS  Article  Google Scholar 

  10. Meloni, S., Arenas, A. & Moreno, Y. Traffic-driven epidemic spreading in finite-size scale-free networks. Proc. Natl. Acad. Sci. USA 106, 16897–16902 (2009).

    CAS  Article  Google Scholar 

  11. Brockmann, D. & Helbing, D. The hidden geometry of complex, network-driven contagion phenomena. Science 342, 1337–1342 (2013).

    CAS  Article  Google Scholar 

  12. Lai, S. et al. Effect of non-pharmaceutical interventions to contain COVID-19 in China. Nature 585, 410–413 (2020).

    CAS  Article  Google Scholar 

  13. Kim, J. & Kwan, M.-P. The impact of the COVID-19 pandemic on people’s mobility: a longitudinal study of the US from March to September of 2020. J. Transp. Geogr. 93, 103039 (2021).

    Article  Google Scholar 

  14. Persson, J., Parie, J. F. & Feuerriegel, S. Monitoring the COVID-19 epidemic with nationwide telecommunication data. Proc. Natl. Acad. Sci. USA 118, 26 (2021).

    Google Scholar 

  15. Han, X. et al. Quantifying COVID-19 importation risk in a dynamic network of domestic cities and international countries. Proc. Natl. Acad. Sci. USA 118, 31 (2021).

    Google Scholar 

  16. Gibbs, H. et al. Changing travel patterns in China during the early stages of the COVID-19 pandemic. Nat. Commun. 11, 5012 (2020).

    CAS  Article  Google Scholar 

  17. Leung, K., Wu, J. T. & Leung, G. M. Real-time tracking and prediction of COVID-19 infection using digital proxies of population mobility and mixing. Nat. Commun. 12, 1501 (2021).

    CAS  Article  Google Scholar 

  18. Hasselwander, M. et al. Building back better: the COVID-19 pandemic and transport policy implications for a developing megacity. Sustain. Cities Soc 69, 102864 (2021).

    Article  Google Scholar 

  19. Gao, Y. & Levinson, D. COVID-19, travel time reliability, and the emergence of a double-humped peak period. Transport Findings. https://doi.org/10.32866/001c.27013 (2021).

  20. Zhang, Y., Zhang, A. & Wang, J. Exploring the roles of high-speed train, air and coach services in the spread of COVID-19 in China. Transp. Policy 94, 34–42 (2020).

    Article  Google Scholar 

  21. Beck, M. J. & Hensher, D. A. Insights into the impact of COVID-19 on household travel and activities in Australia-the early days under restrictions. Transp. Policy 96, 76–93 (2020).

    Article  Google Scholar 

  22. Mirtich, L., et al. How stable are transport-related attitudes over time? Transport Findings. https://doi.org/10.32866/001c.24556 (2021).

  23. Aoustin, L. & Levinson, D. M. Longing to travel: Commute appreciation during COVID-19. Transport Findings. https://doi.org/10.32866/001c.18523 (2021).

  24. Earley, R. & Newman, P. Transport in the aftermath of COVID-19: Lessons learned and future directions. J. Transp. Technol. 11, 109–127 (2021).

    Google Scholar 

  25. Zhang, J., Hayashi, Y. & Frank, L. D. COVID-19 and transport: Findings from a world-wide expert survey. Transp. Policy 103, 68–85 (2021).

    Article  Google Scholar 

  26. Bešinović, N. Resilience in railway transport systems: a literature review and research agenda. Transp. Rev. 40, 457–478 (2020).

    Article  Google Scholar 

  27. Chopra, S. S., Dillon, T., Bilec, M. M. & Khanna, V. A network-based framework for assessing infrastructure resilience: a case study of the London metro system. J. Royal Soc. Interface 13, 20160113 (2016).

    Article  Google Scholar 

  28. Bevilacqua, M., Ciarapica, F. E. & Marcucci, G. Supply chain resilience triangle: the study and development of a framework. J. Int. Eco. Manag. Eng 11, 2046–2053 (2017).

    Google Scholar 

  29. Bonanno, G. A., Galea, S., Bucciarelli, A. & Vlahov, D. What predicts psychological resilience after disaster? the role of demographics, resources, and life stress. J. Consult. Clin. Psychol. 75, 671–682 (2007).

    Article  Google Scholar 

  30. Mokhtarian, P. L., Salomon, I. & Singer, M. E. What moves us? An interdisciplinary exploration of reasons for traveling. Transp. Rev. 35, 250–274 (2015).

    Article  Google Scholar 

  31. Schwanen, T. ESCAP75: Enhancing the resilience of urban transport in Asian cities after COVID-19: Synthesis of academic study results and general recommendations. https://www.unescap.org/sites/default/d8files/event-documents/Enhancing_the_Resilience_of_Urban_Transport_0.pdf (2021).

  32. Jenelius, E. Rail transport resilience to demand shocks and COVID-19. Preprint at https://www.researchgate.net/publication/343600990 (2020).

  33. Leung, K., Wu, J., Liu, D. & Leung, G. First-wave COVID-19 transmissibility and severity in China outside Hubei after control measures, and second-wave scenario planning: a modelling impact assessment. Lancet (London, England) 395, 1382–1393 (2020).

    CAS  Article  Google Scholar 

  34. Hou, X. et al. Intra-county modeling of COVID-19 infection with human mobility: assessing spatial heterogeneity with business traffic, age, and race. Proc. Natl. Acad. Sci. USA 118, 24 (2021).

    Google Scholar 

  35. Buckee, C. O. et al. Aggregated mobility data could help fight COVID-19. Science 368, 145–146 (2020).

    Article  CAS  Google Scholar 

  36. Charoenwong, B., Kwan, A. & Pursiainen, V. Social connections with COVID-19-affected areas increase compliance with mobility restrictions. Sci. Adv. 6, eabc3054 (2020).

    CAS  Article  Google Scholar 

  37. Kogan, N. E. et al. An early warning approach to monitor COVID-19 activity with multiple digital traces in near real time. Sci. Adv. 7, eabd6989 (2021).

    Article  Google Scholar 

  38. Balcan, D. et al. Multiscale mobility networks and the spatial spreading of infectious diseases. Proc. Natl. Acad. Sci. USA 106, 21484–21489 (2009).

    CAS  Article  Google Scholar 

  39. Zhang, R., Li, Y., Zhang, A. L., Wang, Y. & Molina, M. J. Identifying airborne transmission as the dominant route for the spread of COVID-19. Proc. Natl. Acad. Sci. USA 117, 14857–14863 (2020).

    CAS  Article  Google Scholar 

  40. Ministry of Transport of the People’s Republic of China, R. Statistics reports of urban traffic volume. https://www.mot.gov.cn/tongjishuju/chengshikeyun/ (2020).

  41. Huang, J., Levinson, D., Wang, J., Zhou, J. & Wang, Z.-J. Tracking job and housing dynamics with smartcard data. Proc. Natl. Acad. Sci. USA 115, 12710–12715 (2018).

    CAS  Article  Google Scholar 

  42. Kandt, J. & Batty, M. Smart cities, big data and urban policy: towards urban analytics for the long run. Cities 109, 102992 (2021).

    Article  Google Scholar 

  43. Schlosser, F. et al. COVID-19 lockdown induces disease-mitigating structural changes in mobility networks. Proc. Natl. Acad. Sci. USA 117, 32883–32890 (2020).

    CAS  Article  Google Scholar 

  44. Ecoffet, A., Huizinga, J., Lehman, J., Stanley, K. O. & Clune, J. First return, then explore. Nature 590, 580–586 (2021).

    CAS  Article  Google Scholar 

  45. Richardson, D. B. et al. Spatial turn in health research. Science 339, 1390–1392 (2013).

    CAS  Article  Google Scholar 

  46. Lau, M. S. Y. et al. Characterizing superspreading events and age-specific infectiousness of SARS-CoV-2 transmission in Georgia, USA. Proc. Natl. Acad. Sci. USA 117, 22430–22435 (2020).

    CAS  Article  Google Scholar 

  47. Beck, M. J., Hensher, D. A. & Wei, E. Slowly coming out of COVID-19 restrictions in Australia: implications for working from home and commuting trips by car and public transport. J. Transp. Geogr. 88, 102846 (2020).

    Article  Google Scholar 

  48. Du, F., Mao, L. & Wang, J. Determinants of travel mode choice for seeking healthcare: a comparison between elderly and non-elderly patients. J. Transp. Geogr. 92, 103023 (2021).

    Article  Google Scholar 

  49. Yang, J. et al. Despite vaccination, China needs non-pharmaceutical interventions to prevent widespread outbreaks of COVID-19 in 2021. Nat. Hum. Behav. 5, 1009–1020 (2021).

    Article  Google Scholar 

  50. Patterson, Z. & Farber, S. Potential path areas and activity spaces in application: a review. Transp. Rev. 35, 679–700 (2015).

    Article  Google Scholar 

Download references

Acknowledgements

This work is financially supported by the National Natural Science Foundation of China (Grant Nos. 42121001, 42071147, and 42001123), Youth Innovation Promotion Association of Chinese Academy of Sciences (Grant No. 2021049).

Author information

Authors and Affiliations

Authors

Contributions

J.W. and J.H. conceived the idea and performed the experiments, and they are co-first authors; J.W., J.H., and D.L. designed the experiments; J.W., J.H., and H.Y. analyzed the data; J.W., J.H., and D.L. wrote the paper. All authors reviewed and edited the paper.

Corresponding authors

Correspondence to Jiaoe Wang or Haoran Yang.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Wang, J., Huang, J., Yang, H. et al. Resilience and recovery of public transport use during COVID-19. npj Urban Sustain 2, 18 (2022). https://doi.org/10.1038/s42949-022-00061-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1038/s42949-022-00061-1

Search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing