The petroleum refining industry in China is a major contributor to the national economy and a significant source of ambient volatile organic compounds (VOCs). The development history of China’s refineries was investigated for the period 1949–2018, and future development trends were predicted until 2030. The historical VOC emissions from 1949 to 2018 were estimated based on source-specific emission factors, and the emissions in 2025 and 2030 were predicted under the business-as-usual (BAU), alternative control (AC), and accelerated control (ACC) scenarios. Each scenario consisted of a policy and a technical scenario. VOC emissions from refineries increased from 0.53 Gg in 1949 to 1.12 Tg in 2018; fugitive emissions were always the most significant sources of VOCs (40.0–43.9%), followed by end-of-pipe (28.4–31.3%), tank storage (18.3–25.3%), and wastewater treatment (5.8–6.6%) emissions. Provinces in the coastal area have experienced more VOC emissions than inland areas, and Eastern China currently has the highest VOC emissions from refineries. By 2030, China could reduce its current VOC emissions by 5.4%, 35.7%, and 62.5% under the BAU, AC, and ACC scenarios, respectively. The main pressure for reducing VOC emission from China’s refineries will come predominantly from Northeastern China, followed by Eastern and Northern China. The improvement of the production processes, enhancing the airtightness of equipment and containers, and implementation of improved leak detection and repair system are the more effective measures in reducing VOC emissions, accounting for more than 40% of the total reduction. In addition, the penetration and removal rate of control measures for end-of-pipe sources should be further strengthened.

Identifying and quantifying the major sources of atmospheric particulate matter (PM) is essential for the development of pollution mitigation strategies to protect public health. However, urban PM is affected by local primary emissions, transport, and secondary formation; therefore, advanced methods are needed to elucidate the complex sources and transport patterns. Here, an improved source apportionment method was developed by incorporating the receptor model, Lagrangian simulation, and emissions inventories to quantify PM2.5 sources for an industrial city in China. PM2.5 data including ions, metals, organic carbon, and elemental carbon were obtained by analyzing 1 year of sampling results at urban and rural sites. This method identified coal combustion (30.64%), fugitive dust (13.25%), and vehicles (12.51%) as major primary sources. Secondary sources, including sulfate, nitrate, and secondary organic aerosols also contributed strongly (25.28%–30.76% in total) over urban and rural areas. Hebei Province was the major regional source contributor (43.05%–57.51%) except for fugitive dust, on which Inner Mongolia had a greater impact (43.51%). The megacities of Beijing and Tianjin exerted strong regional impacts on the secondary nitrate and secondary organic aerosols factors, contributing 11.32% and 15.65%, respectively. Pollution events were driven largely by secondary inorganic aerosols, highlighting the importance of reducing precursor emissions at the regional scale, particularly in the Beijing–Tianjin–Hebei region. Overall, our results demonstrate that this novel method offers good flexibility and efficiency for quantifying PM2.5 sources and regional contributions, and that it can be extended to other cities.

Firework/firecracker (FF) burning can significantly deteriorate air quality, whereas little is known about its influences on the elemental composition and associated health risks. Fine particles (PM2.5) and trace elements were characterized based on a multi-site campaign at Chifeng, China around 2016 Chinese Spring Festival (SF). Severe pollution levels average of 57.70 μg m−3 were observed during the SF with maximum to 471.00 μg m−3 shortly after the intensive FF activities. Largely enhanced PM2.5-bound metals were found in both urban and rural sites especially for K (8.27±5.36 μg m−3) and Al (2.36±1.41 μg m−3). Ba and Sr as the tracer of fireworks also increased more than 20-fold compared to non-SF period. Accordingly, FF burning factor identified via PMF model contributed significantly to the total elemental mass (71.34±24.94%) during the SF. Its major impacts on both crustal elements as Al, Ca, K and heavy metals as Cr, Cu and Pb were both identified. Elevated non-cancer risks (0.76 to children, 0.11 to adults) and cancer risks (3.96 × 10−6) were assessed during the SF, with As, Cd, Pb exerted the most adverse threats. The FF burning contributed the second largest share of the health threats after coal combustion, accounted for 28.35% and 12.64% of non-cancer risks for children and adults, respectively, and 10.03% of cancer risks, respectively. This study provided scientific evidences for stricter firework/firecracker regulations to protect public health.

Trace metals in atmospheric particulate matter (PM) are a serious threat to public health. Although pollution from toxic metals has been investigated in many Chinese cities, the spatial and temporal patterns in PM2.5 remain largely unknown. Long-term PM2.5 field sampling in 11 cities, combined with a systemic literature survey covering 51 cities, provides the first comprehensive database of 21 PM2.5-bound trace metals in China. Our results revealed that PM2.5 elemental compositions varied greatly, with generally higher levels in North China, especially for crustal elements. Pollution with Cr, As, and Cd was most serious, with 61, 38, and 16 sites, respectively, surpassing national standards, including some in rural areas. Local emissions, particularly from metallurgical industries, were the dominant factors driving the distribution in polluted cities such as Hengyang, Yuncheng, and Baiyin, which are mainly in North and Central China. Elevated As, Cd, and Cr levels in Yunnan, Guizhou Province within Southwest China were attributed to the high metal content of local coal. Diverse temporal trends of various elements that differed among regions indicated the complexity of emission patterns across the country. The results demonstrated high non-carcinogenic risks for those exposed to trace metals, especially for children and residents of heavily cities highly polluted with As, Pb, or Mn. The estimated carcinogenic risks ranged from 6.61 × 10−6 to 1.92 × 10−4 throughout China, with As being the highest priority element for control, followed by Cr and Cd. Regional diversity in major toxic metals was also revealed, highlighting the need for regional mitigation policies to protect vulnerable populations.

In order to evaluate the volatile organic compounds (VOCs) pollution characteristics in Chengdu and to identify their sources, ambient air sample collection and measurement were conducted at 28 sampling sites covering all districts/counties of Chengdu from May 2016 to January 2017. Meanwhile, a county-level anthropogenic speciated VOCs emission inventory was established by “bottom-up” method for 2016. Then, a comparison was made between the VOCs emissions, spatial variations, and source structures derived from the measurement and emission inventory. Ambient measurements showed that the annual average mixing ratios of VOCs in Chengdu were 57.54 ppbv (12.36 to 456.04 ppbv), of which mainly dominated by alkanes (38.8%) and OVOCs (22.0%). The ambient VOCs in Chengdu have distinct spatiotemporal characteristics, with a high concentration in January at the middle-northern part of the city and a low concentration in September at the southwestern part. The spatial distribution of VOCs estimated by the emission inventory was in good agreement with ambient measurements. Comparison of individual VOCs emissions indicated that the emissions of non-methane hydrocarbon species agreed within ±100% between the two methods. Both positive matrix factorization (PMF) model results and emission inventory showed that vehicle emissions were the major contributor of anthropogenic VOCs in Chengdu (31% and 37%), followed by solvent utilization (26% and 27%) and industrial processes (23% and 30%). The large discrepancies were found between the relative contribution of combustion sources, and the PMF resolved more contributions (20%) than the emission inventory (6%). Overall, this study demonstrates that measurement results and emission inventory were in a good agreement. However, to improve the reliability of the emission inventory, we suggest significant revision on source profiles of oxygenated volatile organic compounds (OVOCs) and halocarbons, as well as more detailed investigation should be made in terms of energy consumption in household.

This paper compiled a new speciated NMVOCs emission inventory for the industrial sources at the county-level by using a bottom-up approach in 2016, as well as estimated the ozone formation potential (OFP) and investigated its spatial characteristics in China. Results indicated that the total NMVOCs emissions from industrial sources estimated as 21.04 Tg in 2016. The five major source categories including “production of VOCs”, “storage and transportation”, “industrial processes using VOCs as raw material”, “processes using VOCs-containing products”, and “industrial fossil fuel combustion processes” generated 1.92 Tg, 0.94 Tg, 6.54 Tg, 10.04 Tg, and 1.60 Tg VOCs, respectively, in 2016. According to our estimates, aromatics were the largest contributor of industrial NMVOCs emissions in 2016, accounting for 36% of total NMVOCs, followed by Alkanes (29%), OVOCs (22%), Alkenes (7%), Halocarbons (4%), and Alkynes (2%). Styrene, m/p-xylene, ethylbenzene, toluene, and ethyl acetate were the top five VOC species from industrial sources in terms of abundance in 2016. Aromatics have a high potential for ozone formation, and accounted for 70% of total OFP, followed by Alkenes (14%), Alkanes (10%), and OVOCs (6%). Styrene, p-xylene, toluene, ethylbenzene, 1,3-butadiene were the five species that had the largest potential to form ozone, and plastic industry, coke industry, household appliances industry, and architectural decoration were the key contributing sectors. The emissions displayed distinct spatial characteristics, with significantly higher emissions and OFPs in coastal regions than in other inland areas of China.

Coke production is a significant source of ambient volatile organic compound emissions; thus, stringent control measures must be applied. We fully characterized the trends in volatile organic compound emissions by the coking industry in China between 1949 and 2016 based on a factory-based database and process-specific emission factors. We then projected the reduction potentials in these emissions if different control policies were implemented in 2020 based on three emission scenarios. The results indicate that: (1) the emission factor of volatile organic compounds for coke plants under uncontrolled conditions was 3.065 g/kg coke, and benzene, toluene, and acetone were the most abundant emission species. (2) The annual volatile organic compound emissions from the coking industry increased by an order of magnitude from 3.38 Gg in 1949 to 1376.54 Gg in 2016. The emissions show distinct spatial characteristics, with significantly higher emissions in northern China than in other areas. (3) Compared to the uncontrolled scenario, if basic or more stringent control measures were fully implemented in China in 2020, then volatile organic compound emissions would be reduced by 59% or 82%, respectively. (4) Controlling coke oven flue gases through efficient combustion, sealing and cleaning the openings of coke ovens, and using gas blanketing or carbon absorbers in by-product facilities were the most effective technologies for controlling volatile organic compound emissions from coke production.

Biomass burning (BB) seriously affect air pollution, human health and global climate. A severe pollution episode (PE) caused by BB was investigated in the southern Sichuan Basin (SSB), one of the most polluted areas in China. Hourly variations in criteria air pollutants (PM2.5, PM10, SO2, NO2, CO, and O3), chemical components, and sources of PM2.5 before, during, and after the severe regional air PE were characterized at three sites, namely Neijiang (NJ), Zigong (ZG), and Yibin (YB). The results showed that combination of intensive pollution from BB, stable meteorological conditions, and the basin topography caused this severe regional PE in the SSB. The average daily concentrations of PM2.5 during the PE were 1.8–6 times those measured during the periods before and after the PE, and 4.0–7.4 times that of World Health Organization air quality guidelines in the SSB. The highest PM levels occurred in ZG, where the peak values of PM2.5 and PM10 reached 536 μg m−3 and 578 μg m−3 at night, respectively. PM10, NO2, and CO also increased dramatically at night in the SSB. O3 formation was affected by BB, showing lower levels at night but higher levels in the day during the PE than before and after the PE, whereas SO2 levels were not affected. Sulfate–nitrate–ammonium in PM2.5 was the main chemical compositions before the PE, whereas organic matter (OM) and K+ became characteristics compositions during and after the PE. Higher OC/EC and Kexcess/EC ratios were observed during the PE and Kexcess/EC ratio was a better indicator of BB in the SSB than OC/EC ratio. The results of a positive matrix factorization model indicated that BB was the most significant contributor to PM2.5 during the PE, accounting for 58% in NJ, 65% in ZG, and 56% in YB. Backward trajectory analysis confirmed that the SSB is susceptible to pollutants from Chongqing and other surrounding cities, especially in ZG and NJ, due to the unique topography of the basin. Our findings suggest that BB in the basin topography can cause severe regional air pollution events at night, thus supporting the critical need for BB control in the basin to improve regional air quality.

Carbonaceous aerosols are linked to severe haze and health effects, while its origins remain still unclear over China. PM2.5 samples covering four seasons from Jan. 2016 to Jan. 2017 were collected at six sites in Chifeng, a representative agro-pastoral transitional zone of North China focusing on the characteristics and sources of organic carbon (OC) and elemental carbon (EC). The annual averages of OC, EC were 9.00 ± 7.24 μg m−3, 1.06 ± 0.79 μg m−3 with site Songshan in coal mining region exhibited significantly enhanced levels. The residential heating emissions, air stagnation, and secondary organic formation all contributed the higher OC, EC levels in winter. Meanwhile, the impacts from open biomass burning were most intensive in spring. The retroplumes via Lagrangian model highlighted a strong seasonality of regional sources which had more impacts on EC increases. The Positive Matrix Factorization (PMF) model resolved six primary sources, namely, coal combustion, biomass burning, industrial processes, oil combustion, fugitive dust, and fireworks. Coal combustion and biomass burning comprised large fractions of OC (30.57%, 30.40%) and EC (23.26%, 38.47%) across the sites, while contributions of industrial processes and oil combustion clearly increased in the sites near industrial sources as smelters. PMF and EC tracer method gave well correlated (r=0.65) estimates of Secondary OC (SOC). The proportion of coal combustion and SOC were more enhanced along with PM2.5 elevation compared to other sources, suggesting their importances during the pollution events.

\textlessp\textgreater\textlessstrong\textgreaterAbstract.\textless/strong\textgreater Improving the accuracy of the anthropogenic volatile organic compound (VOC) emission inventory is essential for reducing air pollution. In this study, we established an emission inventory of anthropogenic VOCs in the Beijing–Tianjin–Hebei (BTH) region of China for 2015 based on the emission factor (EF) method. Online ambient VOC observations were conducted in one urban area of Beijing in January, April, July, and October, which, respectively, represented winter, spring, summer, and autumn in 2015. Furthermore, the developed emission inventory was evaluated by a comprehensive verification system based on the measurements and satellite retrieval results. Firstly, emissions of the individual species of the emission inventory were evaluated according to the ambient measurements and emission ratios versus carbon monoxide (CO). Secondly, the source structure of the emission inventory was evaluated using source appointment with the Positive Matrix Factorization (PMF) model. Thirdly, the spatial and temporal distribution of the developed emission inventory was evaluated by a satellite-derived emission inventory. According to the results of the emission inventory, the total anthropogenic VOC emissions in the BTH region were 3277.66 Gg in 2015. Online measurements showed that the average mixing ratio of VOCs in Beijing was approximately 49.94 ppbv in 2015, ranging from 10.67 to 245.54 ppbv. The annual emissions for 51 of 56 kinds of non-methane hydrocarbon species derived from the measurements agreed within \textlessspan class="inline-formula"\textgreater±100\textless/span\textgreater % with the results of the emission inventory. Based on the PMF results and the emission inventory, it is evident that vehicle-related emissions dominate the composition of anthropogenic VOCs in Beijing. The spatial correlation between the emission inventory and satellite inversion result was significant (\textlessspan class="inline-formula"\textgreater\textitp<0.01\textless/span\textgreater) with a correlation coefficient of 0.75. However, there were discrepancies between the relative contributions of fuel combustion, emissions of oxygenated VOCs (OVOCs), and halocarbons from the measurements and inventory. To obtain a more accurate emission inventory, we propose the investigation of the household coal consumption, the adjustment of EFs based on the latest pollution control policies, and the verification of the source profiles of OVOCs and halocarbons.\textless/p\textgreater

Volatile organic compounds (VOCs) pollution, which is closely linked to photochemical smog and secondary organic aerosols, has become a severe concern in China. Therefore, we compiled a new high-resolution emission inventory for the industrial non-methane Volatile organic compounds (NMVOCs) using “bottom-up” approaches throughout 2010 and 2016. In this work, the industrial sources were divided into five major categories, and 108 specific sources, as well as an emission factor database, was developed for industrial NMVOCs. Results indicated that the total NMVOCs emissions from industrial sources increased from 16.88 Tg in 2010 to 21.04 Tg in 2016 at an annual average rate of 3.7%. The five major source categories including “production of VOCs”, “storage and transportation”, “industrial processes using VOCs as raw material”, “processes using VOCs-containing products”, and “fossil fuel combustion” generated 1.92 Tg, 0.94 Tg, 6.54 Tg, 10.04 Tg, and 1.60 Tg NMVOCs, respectively, in 2016. Coke production, plastic manufacturing, raw medicine industry, and architectural decoration were the primary sources of industrial NMVOCs and emissions of these sources increased by 140 Gg, 190 Gg, 640 Gg, and 700 Gg between 2010 and 2016. The emissions displayed distinct spatial characteristics, with significantly higher emissions in the Beijing-Tianjin-Hebei region, the Pearl River Delta, the Yangtze River Delta, and the Cheng-Yu region than in other areas. Shandong, Guangdong, Jiangsu, Zhejiang, and Henan were the top five provinces with the highest NMVOCs emissions, while the emission hotspots in the county-level were mainly distributed in Guangzhou urban area, Shanghai Pudong New Area, Hangzhou urban area, and Shenzhen urban area. The emissions in Henan province, Hubei province, and Cheng-Yu region increased significantly during the study period. Instead, emissions in some counties of Zhejiang province and Hebei province decreased than in 2010.

Trace elements in atmospheric particular matter play a significant role in controlling aerosolbehavior, and can thereby endanger air quality. Here, the comprehensive investigation on the elemental characteristics and sources in fine and coarse particles at Chifeng was presented. The daily samples of particular matter (PM2.5 and PM10) were collected at six sites for a one-year period, and concentrations of 19 elements were analyzed. Results showed that Al, K, Ca, Fe were the most concentrated elements, in both PM2.5 and PM10. The crustal elements mainly in coarse particles (PM2.5–10) presented higher levels during March to May, due to the increased dust suspension in springtime. The highly enriched elements as Pb, Cd mainly in fine particles (PM2.5) presented elevated levels in cold seasons, related to the increased emissions of coal combustion for heating. Site Songshan had significantly higher Pb, As, Cd levels, ascribing to the influence of coal mining. The influences of metallurgy industries on Fe, Cu, Zn levels in both size fractions were also observed. Positive matrix factorization (PMF) identified four common sources for trace elements in both fine and coarse fractions, namely fugitive dust, coal mining, a mixed industrial factor with iron and zinc, and copper smelting. The factors of coal combustion, biomass burning, oil combustion, vehicle emission and fireworks were merely obtained for fine particles. The crustal elements were mainly related to the impact of fugitive dust, while the notable impacts of coal combustion and iron/steel production were also confirmed. Cu was attributed to copper smelting in both sizes, while the major sources of Zn varied from vehicle emission (44.3%), coal combustion (32.1%) in PM2.5 to mixed industrial factor (89.3%) in PM2.5–10. Although coal combustion, coal mining and copper smelting contributed <20% of the total elemental concentrations, they were responsible for >80% of the toxic elements Pb, As, Cd.

Air quality monitoring networks play a significant role in identifying the spatiotemporal patterns of air pollution, and they need to be deployed efficiently, with a minimum number of sites. The revision and optimal adjustment of existing monitoring networks is crucial for cities that have undergone rapid urban expansion and experience temporal variations in pollution patterns. The approach based on the Weather Research and Forecasting–California PUFF (WRF-CALPUFF) model and genetic algorithm (GA) was developed to design an optimal monitoring network. The maximization of coverage with minimum overlap and the ability to detect violations of standards were developed as the design objectives for redistributed networks. The non-dominated sorting genetic algorithm was applied to optimize the network size and site locations simultaneously for Shijiazhuang city, one of the most polluted cities in China. The assessment on the current network identified the insufficient spatial coverage of SO2 and NO2 monitoring for the expanding city. The optimization results showed that significant improvements were achieved in multiple objectives by redistributing the original network. Efficient coverage of the resulting designs improved to 60.99% and 76.06% of the urban area for SO2 and NO2, respectively. The redistributing design for multi-pollutant including 8 sites was also proposed, with the spatial representation covered 52.30% of the urban area and the overlapped areas decreased by 85.87% compared with the original network. The abilities to detect violations of standards were not improved as much as the other two objectives due to the conflicting nature between the multiple objectives. Additionally, the results demonstrated that the algorithm was slightly sensitive to the parameter settings, with the number of generations presented the most significant effect. Overall, our study presents an effective and feasible procedure for air quality network optimization at a city scale.

Fine particulate matter (PM2.5), largely composed of secondary organic aerosol (SOA), is currently one of the most intractable environmental problems in China. As crucial precursors for SOA, understanding the formation propensity of various volatile organic compound (VOC) species and sources is useful for pollution control. In this work, we estimated the SOA formation potential (SOAP) of anthropogenic VOC emissions based on an improved speciated VOC emission inventory and investigated its distribution in China. According to our estimates, toluene had the largest SOAP, followed by n-dodecane, m-/p-xylene, styrene, n-decane, and n-undecane, while passenger cars, chemical fiber manufacturing, asphalt paving, and building coating were the top five SOAP-contributing sources nationwide. The spatial distribution of SOAP in China shows a distinct pattern of high values in the southeast and low values in the northwest. Beijing–Tianjin–Hebei and surroundings, the Yangtze River Delta, Pearl River Delta, and Sichuan–Chongqing District were found to have the highest SOAP, particularly in urban areas. The major SOAP-contributing species and sources differed among these regions, which was attributed to local industrial and energy structures. Our results suggest that to mitigate PM2.5 pollution in China, more efficient SOAP-based control measures should be implemented instead of current emissions-based policies, and VOC control strategies should be adapted to local conditions.

Chengdu is a megacity in the southwest of China with high ozone (O3) mixing ratio. Observation of volatile organic compounds (VOCs), NO2 and O3 with high temporal resolution was conducted in Chengdu to investigate the chemical processes and causes of high O3 levels. The hourly mixing ratios of VOCs, NO2, and O3 were monitored by an online system from 28 August to 7 October, 2016. According to meteorological conditions, Chengdu, with relative warm weather and low wind speed, is favorable to O3 formation. Part of the O3 in Chengdu may be transported from the downtown area. In O3 episodes, the average mixing ratios of NO2 and O3 were 20.20 ppbv and 47.95 ppbv, respectively. In non-O3 episodes, the average mixing ratios of NO2 and O3 were 16.38 ppbv and 35.15 ppbv, respectively. The average mixing ratio of total VOCs (TVOCs) was 40.29 ppbv in non-O3episodes, which was lower than that in O3 episodes (53.19 ppbv). Alkenes comprised 51.7% of the total O3 formation potential (OFP) in Chengdu, followed by aromatics which accounted for 24.2%. Ethylene, trans-pentene, propene, and BTEX (benzene, ethylbenzene, toluene, m/p-xylene, o-xylene) were also major contributors to the OFP in Chengdu. In O3 episodes, intensive secondary formations were observed during the campaign. Oxygenated VOCs (OVOCs), such as acetone, Methylethylketone (MEK), and Methylvinylketone (MVK) were abundant. Isoprene rapidly converted to MVK and Methacrolein (MACR) during O3 episodes. Acetone was mainly the oxidant of C3-C5 hydrocarbons.

Chongqing is the largest megacity in southwest China and has a mountainous and humid climate. Online measurements of 96 volatile organic compound (VOC) species were performed at the three sites JYS, CJZ, and NQ, which are located in the northern, central, and southern sections of the Chongqing urban district, respectively. The measurements were performed from August to September 2015, at a time interval of 1 h. The spatiotemporal variation of VOC sources in Chongqing was characterized by combining the positive matrix factorization (PMF) model with the online measurement data. The average total VOC mixing ratios of the CJZ, NQ, and JYS sites were 41.2, 34.1, and 23.0 ppbv, respectively. The mixing ratios of tracers of incomplete combustion, exhibited obvious bimodal profiles at the CJZ and NQ sites, whereas those at the JYS site exhibited little change throughout the day. Isoprene at the three sites followed a similar pattern of average diurnal variations in mixing ratios, with minimums before sunrise and maximums at noon. The dominant sources of acetaldehyde and acetone were secondary anthropogenic sourceand aged air mass transport, respectively, in the city of Chongqing. Seven sources were apportioned to the results of PMF calculation using spatiotemporal VOCs composition data. The Vehicle-related sources were the largest contributor at CJZ and NQ, contributing 44% and 37% of the total VOC mixing ratios, respectively, and exhibited clear diurnal variations. Aged background air, with 68% of total VOC emissions, dominated the VOC emissions at JYS. Solvent utilization was a very important contributor at NQ and coincided with the higher levels of aromatics. O3 formation was generally VOC-limited at NQ and CJZ, and was NOx-limited and transition region alternatively at JYS. Alkenes were important for the O3formation at CJZ, and both alkenes and aromatics were important for the O3 formation at NQ.

Seasonal pattern of transport pathways and potential sources of PM2.5 in Chengdu during 2012–2013 were investigated based on hourly PM2.5 data, backward trajectories, clustering analysis, potential source contribution function (PSCF), and concentration-weighted trajectory (CWT) method. The annual hourly mean PM2.5 concentration in Chengdu was 97.4 mg·m–3. 5, 5, 5 and 3 mean clusters were generated in four seasons, respectively. Short-distance air masses, which travelled within the Sichuan Basin with no specific source direction and relatively high PM2.5 loadings (>80 mg·m–3) appeared as important pathways in all seasons. These short pathways indicated that emissions from both local and surrounding regions of Chengdu contributed significantly to PM2.5 pollution. The cities in southern Chengdu were major potential sources with PSCF>0.6 and CWT>90 mg·m–3. The northeastern pathway prevailed throughout the year with higher frequency in autumn and winter and lower frequency in spring and summer. In spring, long-range transport from southern Xinjiang was a representative dust invasion path to Chengdu, and the CWT values along the path were 30-60 mg·m–3. Long-range transport was also observed in autumn from southeastern Xinjiang along a northwesterly pathway, and in winter from the Tibetan Plateau along a westerly pathway. In summer, the potential source regions of Chengdu were smaller than those in other seasons, and no long-range transport pathway was observed. Results of PSCF and CWT indicated that regions in Qinghai and Tibet contributed to PM2.5 pollution in Chengdu as well, and their CWT values increased to above 30 mg·m–3 in winter.

Chongqing, the largest megacity in southwest China, faces serious aerosol pollution but lacks information on particle characteristics and its sources. Official data released by Chongqing Environmental Protection Bureau demonstrated that urban PM10 concentrations decreased remarkably from 150 μg m− 3 in 2000 to 90 μg m− 3 in 2012. However, only several peer-reviewed studies paid attention to local fine particle (PM2.5) pollution. In the study, PM2.5 samples were obtained and subjected to chemical analysis in an urban site of the city during 2012 to 2013. The annual mean PM10 and PM2.5 concentrations in urban Chongqing were 103.9 ± 52.5 and 75.4 ± 42.2 μg m− 3, respectively. PM2.5 showed a distinct seasonality of high concentration in winter and similar levels in other seasons. The average OC/EC (organic carbon/element carbon) ratio was 3.7 with more high-OC/EC ratio sources contribution in autumn and winter. The varying sources and formation mechanisms resulted in SO42 − and NH4+ peaks in both summer and winter, whereas high nitrate concentration was only observed in winter. In the average mass closure, PM2.5 was composed of 23.0% SO42 −, 11.7% NO3−, 10.9% NH4+, 30.8% OM (organic matter), 5.2% EC, 8.2% mineral dust, 0.6% TEO (trace elements), 1.0% Cl− and 1.1% K+, while exhibiting large seasonal variability. Using positive matrix factorization (PMF), six sources were apportioned in PM2.5: secondary inorganic aerosols, coal combustion, other industrial pollution, soil dust, vehicular emission, and metallurgical industry. The annual mean contribution of above sources to PM2.5 was 37.5, 22.0, 17.5, 11.0, 9.8 and 2.2%, respectively. Coal combustion was identified by As tracer and dominated the primary sources of PM2.5, while the two different industrial sources were characterized by Cr and Mo, Co, Ni, and Se, respectively. The study is of great importance in characterizing the historical trends, current chemical characteristics and sources of fine particles in urban Chongqing.