This project collected PM2.5 simultaneously at the Banqiao, Zhongming, Douliu, Chiayi, Xiaogang, and Hualien sites of the Environmental Protection Administration (EPA) once every six days. Moreover, near-traffic-source samplings were conducted at the Datung site in Taipei city and the Taiwan Boulevard site in Taichung city. The collected samples were analyzed for PM2.5 mass concentrations, water-soluble inorganic ions, carbonaceous contents, and metal elements to derive pollution causes and source contributions. Meanwhile, factors influencing atmospheric visibility were also studied to provide environmental protection institutions for setting up their control measures. The regular sampling results of PM2.5 mass concentrations were consistent with the EPA’s routine sampling by other teams without apparent bias. The reconstructed PM2.5 chemical composition fractions ranged from 84 to 93%, which were enough to account for varying characteristics of mass concentrations. Seasonal variations showed that high PM2.5 mass concentrations frequently occurred in winter, gradually decreased in spring, and started increasing in spring. High PM2.5 and NO3- concentrations were still distributed over the sites southward central Taiwan. High NO3- concentrations indicated the significance of controlling their precursor sources. The analyzed metal elements showed markers of boiler combustion or biomass burning particles occurring pervasively at the six sites but with higher concentrations in the southern sites. The OC and EC concentrations from the near-traffic-source samplings at the Taiwan Boulevard and the Datung sites were higher than the nearby regular sites. The measured metal elements were mainly associated with the suspended dust and the abrasion of vehicle components. The metal element concentrations of high health risk, such as Pb, As, and Cd, were not extraordinarily high. During sampling, the double-filter corrections for the volatilization and absorption of water-soluble inorganic ions and carbonaceous contents were conducted persistently from 2017 to 2021. PM2.5 mass concentrations were prevented from underestimation by 5-8.5% on average after offsetting volatilization and absorption. The variation trend of PM2.5 chemical composition has been used for assessing source control efficacy over the last five years. The rapid decrease of SO42- concentration indicated an effective SOx control, while a slowly decreasing trend of OC concentration might be related to diversified source contributions. The increase of NO3- and other characteristic species implied a significant influence from industrial activities. In contrast, the persistent decrease of EC concentration represented an effective control for Diesel vehicles. The analysis of chemical composition fraction for high PM2.5 concentration showed that eastern and northern Taiwan should improve emissions from mobile sources and sulfur-fuel boilers in winter and spring. As for the areas from central to southern Taiwan, continued control for NOx and VOCs emission sources is necessary. Sulfur-fuel boilers are the main control target in summer and early autumn. Except the Douliu site, the OC and PM2.5 concentrations can be reduced if the mobile sources are effectively controlled. The previous derivation was based on OC were mostly primary from OC apportionment in the recent five years. From the spatial similarity analysis of the various sites, the control of SOx can be similar in all areas, while the control for NOx and VOCs emission sources needs to have diversified control measures and strength at specific locations. For those heavy metals (As, Cd, Cr, and Pb) levied for air pollution fees from 2018, their concentrations had a downward trend but were slightly different from PM2.5. It is noted that the Cd and Cr concentrations at the Xiaogang site are the highest among all sites. In the recent five years, ten source factors were apportioned from PMF receptor modeling on PM2.5 chemical composition. The highest three source factors are “sulfates,” “nitrates,” and “vehicle emissions.” In addition, the other source factors from high pollution days can indicate local source distinction. The estimated atmospheric visibility from chemical composition revealed that sulfates and organic matter contributed stably in contrast to the predominant contribution of nitrates from central to southern Taiwan in winter and spring. The CMAQ model simulation on high concentration events showed that an assessment of effective reduction ratio for NOx and VOCs and the associated source control measures would reduce O3 formation and NO3- concentration. The high PM2.5 concentration days were mainly influenced by local pollution with a long duration of low wind speed to result in pollution accumulation. Boiler combustion emissions were a significant influencing factor in all areas from analyzing area pollution characteristics. Under the influence of Covid-19, SO42- concentrations were exceedingly low in 2020, rebounding in 2021, but were still below the extension trend lines at all sites. The contributions from mobile emission sources were apparently reduced but were less in stationary source emissions resulting in a small variation or even higher PM2.5 concentration. From the cross-analysis of monitoring data at the Fugue air-quality monitoring site and the chemical composition at the Hualien site, this project found a significant reduction in transboundary pollution transport with a noticeable decrease in boiler marker species. In conclusion, suggestions are made for consistent stationary source control with a tightening control for mobile sources to reduce PM2.5 pollution. At the present stage, an assessment on effective reduction ratio for NOx and VOCs and the associated source control measures is necessary to reduce O3 formation, NO3- concentration, and PM2.5 high-concentration event. The accomplishment will eventually improve atmospheric visibility from central to southern Taiwan in the pollution seasons.

PM2.5 chemical composition, Temporal and spatial distributions, Near- traffic-source PM2.5, PM2.5 sources, Atmospheric visibility