Abstract Air pollution in the form of PM2.5 decreases life expectancy considering its contribution to morbidity and mortality. Therefore, scientific and accurate PM2.5 exposure risk assessment is essential. However, the considering daily/hourly mean PM2.5 concentration and overlooking population mobility in exposure risk assessments result in underestimation of its adverse effects. Thus, using social media data and exceedance PM2.5 concentration, two novel indicators named hourly exceedance PM2.5 exposure (HEPE) and daily cumulative variation of exceedance PM2.5 exposure (DCEPE) are developed in our study. Spatiotemporal variation analysis of HEPE showed that the first exceedance PM2.5 exposure risk was observed at 10:00; this lasted till the end of the day. According to the standard deviation ellipse (SDE) method analysis results, at first, the major spatial tendency direction was northeast-southwest, with the average center (AE) located in the Yuangang Township. Then, the HEPE in the western study area increased dramatically. The major spatial tendency direction shifted from northeast-southwest to east-west. Consequently, the AE of HEPE shifted to central study area. The spatiotemporal variation characteristics led us to investigate the mechanisms. A bivariate LISA was applied to detect the spatial association between DCEPE and city functional zones (CFZs). Results showed that highly spatial associations were found between the DCEPE and CFZs in the southwest of the study area. Residential neighborhoods and transportation services showed a closer relationship with the spatial distribution of DCEPE. Based on these results, we found increasing public health threats posed by PM2.5. Thus, HEPE is an essential factor to assess air pollution exposure risk. Moreover, more attention should be paid to the spatial association between DCEPE and city functions, which is important for the development of air pollution mitigation strategies.