Short-term recovery of outdoor exercise after the lifting of COVID-19 restrictions in Beijing: a GPS-based study of urban environmental differences

Abstract Background Large infection waves can disrupt health-related physical activity even when outdoor spaces remain accessible. However, little is known about how outdoor activity recovers across different urban environments immediately after public health restrictions are lifted. This study examined short-term changes in outdoor exercise in Beijing after the lifting of COVID-19 restrictions, with particular attention to temporal phases, spatial heterogeneity and differences among urban environmental settings. Methods We analysed 196,524 anonymised walking, running and cycling trajectories recorded by the Keep app from 13 December 2022 to 31 March 2023. Activity in January 2022 was used as a reference period. Daily trajectory counts were modelled as time series to identify recovery phases. Spatial clustering, hotspot statistics and trajectory-based counts were used to compare changes across urban green spaces, active transportation corridors and localised sports facilities, and between central and suburban areas. Results Outdoor exercise did not show a simple linear recovery. Four short-term phases were identified: initial stability, brief disruption, rapid rebound and later stabilisation. The main decline occurred after the infection peak, indicating a delayed behavioural response rather than an immediate reduction when restrictions were lifted. Recovery varied markedly by urban setting. After accounting for differences in spatial extent, urban green spaces showed the most stable activity intensity across phases. Active transportation corridors exhibited the largest relative increase during the rapid recovery phase, particularly in central urban areas, whereas localised sports facilities showed a later increase in activity intensity. These patterns describe differences in observed trajectory activity across exercise-space typologies rather than behavioural preferences of population groups. Spatial differences were also evident, central and suburban areas exhibited different recovery patterns in observed exercise trajectories, with variation in the timing and magnitude of activity changes. Conclusions Outdoor exercise recovery after the lifting of COVID-19 restrictions was phased, delayed and spatially uneven. Analysis of spatially normalized activity intensity and trajectory changes revealed differentiated recovery patterns across exercise-space typologies: urban green spaces showed the greatest stability, active transportation corridors showed the strongest increase during early recovery, and localised sports facilities showed higher activity intensity during later recovery. These findings highlight that different exercise-space typologies exhibited distinct recovery patterns after a major public health shock.