Spatiotemporal characteristics of ozone and PM2.5 pollution in Jilin City: insights from observations in 2020 and 2023

Abstract

Ground-level ozone (O₃) and fine particulate matter (PM_2.5) remain central air quality challenges in China. Still, most evidence comes from national, regional, or megacity-scale studies, leaving medium-sized industrial cities in Northeast China poorly resolved. In Jilin City, where petrochemical, transport, and energy-related emissions interact with complex terrain and warm season meteorology, the coupled behavior of O₃ and PM_2.5 requires clearer diagnosis. This study aimed to quantify the spatiotemporal variability, meteorological sensitivity, and first-order attribution of warm season O₃ and PM_2.5 during the COVID-19 low-emission baseline and post-pandemic recovery period. We hypothesized that O₃ and PM_2.5 would respond through decoupled mechanisms—with O₃ showing stronger meteorological sensitivity, while PM_2.5 was only weakly explained by the selected meteorological predictors—a pattern that is statistically consistent with, but not directly attributable to, emission changes, given the study design. Using 3-km WRF–CMAQ simulations, surface observations, correlation analysis, multiple linear regression, and meteorological normalization, we compared May–July pollution patterns between 2020 and 2023. Model evaluation showed acceptable performance for meteorological variables and PM_2.5 but systematic underestimation of O₃, indicating uncertainty in the absolute magnitude of simulated O₃. Observed O₃ increased in May and June but declined slightly in July, with persistent hotspots in Panshi, whereas PM_2.5 increased in May, changed little in June, and declined markedly in July, while remaining relatively elevated in Huadian, Chuanying, and Jiaohe. District-level anomaly and decomposition analyses further indicated a uniform increase in May O₃, heterogeneous June responses, and clearer pollutant-specific response patterns across urban and periurban districts. Meteorological predictors explained approximately 35% of O₃ variability, dominated by temperature, but < 1% of PM_2.5 variability; therefore, PM_2.5 driver interpretation should be treated qualitatively. Normalization suggested a meteorologically associated May O₃ enhancement and an unexplained residual June O₃ enhancement, highlighting O₃–PM_2.5 decoupling; however, because both meteorology and emissions changed simultaneously between the two simulation years, these signals reflect associations rather than controlled causal attribution.

Graphical abstract