Regional Characteristics of Air Pollutants during Heavy Haze Events in the Yangtze River Delta , China

There were 6 severe haze events over a large area of the Yangtze River Delta (YRD) region in January 2013. In this study, based on the hourly concentrations of trace gases and PM2.5 at 10 observation stations (8 city stations, 1 regional background station and 1 island station) during Jan. 1–31, 2013 as well as the concentrations of water-soluble ions at 5 stations (4 city stations and 1 regional background station) during Jan. 18–24, 2013 in the YRD region, the regional characteristics of the air pollutants during heavy haze episodes were investigated in combination with the atmospheric circulation patterns. The concentrations of PM2.5 on haze days were 1.6–2.4-fold higher than on clear days. The concentration of PM2.5, SO2, NO2 and CO increased significantly, with average values of 128.6, 48.5, 78.1 μg m and 1.5 mg m on haze days, and were 64.6, 36, 52.5 μg m and 1.1 mg m on clear days. The PM2.5 concentration of ten observation sites had positive correlations with CO and NO2, and had weakly negative correlations with O3. The sources of PM2.5, SO2, NO2 and CO were strong in inland cities and weak in coastal cities, and the sources of O3 were mainly from Wuxi, Suzhou and southeast of An’hui. The mass and water-soluble ion concentrations were both centralized in PM2.1 during the haze events; additionally, the NH4, SO4 and NO3 ions were dominant, constituting 86–90.9% of the total ion concentrations in PM2.1. The mass spectra of NH4, K, Cl, SO4, F, NO2 and NO3 had unimodal distributions. The secondary formations of sulfate dominated on haze days, and the nitrate oxidation rates were relatively high for inland cities and low for coastal cities.


INTRODUCTION
In recent years, haze pollution has attracted substantial attention due to its significant effects on visibility and public health in China (Wu et al., 2007;Tie et al., 2009;Zhang et al., 2015).In addition, heterogeneous reactions between trace gases and aerosols are significant during haze events, and they can alter the physical and chemical characteristics of aerosols, influencing the ecosystem and climate (Chameides et al., 1999;Cheng et al., 2013).Trace gases and atmospheric aerosols are tightly connected to each other via physical, chemical, meteorological and biological processes that occur in the atmosphere and at theatmosphere-biosphere interface (see, e.g., Seinfeld and Pandis, 2012).Previous studies have reported that the levels of secondary inorganic and organic aerosols increase sharply during haze events (Zhu et al., 2011;Huang et al., 2014), both of which have high mass abundance and high extinction cross-sections in the visible wavelength diameter range (Hand et al., 2007).
Haze pollution caused by heavy particulate matter (PM) loadings causes significant damage in China.A gradual reduction of 2.1 km per decade in the annual mean visibility was revealed in China from 1990-2005(Che et al., 2007).As one of the six largest city clusters in the world, the average visual range of the Yangtze River Delta (YRD) observed by meteorological stations had a consistent decrease in visibility of 2.4 km per decade from < 25 km to < 20 km from the 1980s to 2010s (Che et al., 2009;Gao et al., 2011).Pollutant concentrations and their chemical compositions change significantly on haze days.Static atmospheric stability conditions are thought to play a major role in the high concentrations of PM 2.5 and trace gases on pollution days (Ding et al., 2013a, b;Wang et al., 2014c).Aneja et al. (2004) reported that secondary species can be formed by the oxidation of primary species via O 3 and then impair the visibility.Rastogi et al. (2014) observed that the scattering species (SO 4 2-, NO 3 -and OC) contribute 50% to the PM 2.5 mass, whereas absorbing species (EC) contribute only 4% on haze days.Wang et al. (2005) found that the formations of SO 4 2− and NO 3 − were determined largely by temperature and NH 4 + in Beijing.Sun et al. (2006) found that the concentrations of elements and water-soluble ions (WSIs) (K+, SO 4 2-and NO 3 -) on haze-fog days were tenfold higher than those on clear days.It has been reported that the secondary organic aerosols (SOAs) increased from 35 to 63% from unpolluted and polluted days in Beijing (Zhang et al., 2014a).Tian et al. (2014) discovered that secondary inorganic aerosols and organic matters increased significantly on haze days as well.
In January 2013, several of the most severe haze events on record swept across most of the east-central cities and covered a quarter of the total land area in China.The regional characteristics of particulate matter (PM) pollution during those haze events have been analyzed by observations of the satellite-and surface-based aerosol optical depth (AOD) and model simulations (Boynard et al., 2014;Che et al., 2014;Tao et al., 2014;Wang et al., 2014b).Wang et al. (2014a) and Zhang et al. (2014b) studied the influence of meteorological factors on this haze pollution by observation data and model simulations.Moreover, the chemical components and size distributions of aerosols have been analyzed during haze events (Cheng et al., 2014;Huang et al., 2014;Tian et al., 2014;Zhang et al., 2014a).However, the regional characteristics of PM 2.5 and trace gases and the size distributions of WSIs have rarely been investigated.In this study, based on the concentrations of trace gases and hourly concentrations of PM 2.5 at 10 observation stations (8 city stations, 1 regional background stations and 1 island station) from Jan. 1-31, 2013 as well as the size distributions of WSIs at 5 stations (4 city stations and 1 regional background stations) during Jan.18-24, 2013 in the Yangtze River Delta (YRD) region, the regional characteristics of air pollutants during heavy haze episodes were investigated in combination with the atmospheric circulation patterns.

Observation Stations
The YRD is an alluvial plain located at the eastern coast of China, which is the biggest economic region in China and is one of the world's advanced manufacturing bases with economic gross income constituting approximately 20% of China's Gross Domestic Product (GDP).According to the regional planning of the Yangtze River Delta approved by China's State Council in 2010, the area of the YRD region, including Shanghai, Jiangsu Province and Zhejiang Province, accounts for 2.19% of China, covering 210,700 km 2 .The YRD region belongs to a typical subtropical monsoon climate.In winter, the temperature is low, while the humidity is usually high, leading to a high frequency of haze events.
The 10 stations are located in Nanjing, Wuxi, Suzhou, Huzhou, Jiaxing, Hangzhou, Lin'an, Shaoxing, Ningbo and Zhoushan, which are the major cities in the YRD region (Fig. 1).One regional background station is located in Lin'an, and one island station is located in Zhoushan (Table 1).In  this paper haze days were defined as that with visibility of less than 10 km and relative humidity of less than 90% and there is no precipitation (Sun et al., 2006;Wu et al., 2007;Deng et al., 2008), clear day was considered to occur when the visibility exceeded 10 km and the daily PM 2.5 concentration was less than 75µg m -3 (the Chinese national secondary standard: GB3095-2012; Rao et al., 2016).

Instrumentation and Experiment Descriptions
Trace O 3 , SO 2 , NO 2 and CO gases were measured with a resolution of 1 h using online analyzers (Thermo Instruments, TEI 49i, 43i, 42i and 48i, respectively), and PM 2.5 mass concentrations were measured with a resolution of 1 h using a mass analyzer (Thermo SHARP-5030).These instruments have been used in many other studies (Petäjä et al., 2013;Herrmann et al., 2014), and Ding et al. (2013a) provided more detailed descriptions of the instrumentations.
Observation of aerosol particles was performed using a 9-stage Anderson-type aerosol sampler (Anderson 2000 Inc., USA) with size ranges of < 0.43, 0.43-0.65, 0.65-1.1, 1.1-2.1, 2.1-3.3, 3.3-4.7, 4.7-5.8 5.8-9.0 and 9.0-10.0µm for water-soluble ionic components.The flow rate required by the Anderson-type aerosol sampler is 28.3 L min -1 .The sampler was operated with a 80 mm Teflon filter (Whatman, Clifton, England) for water-soluble ionic components, the membranes were weighed with a Mettler Toledo MX-5 microbalance after constant temperature and humidity treatment before and after sampling, and the microbalance was calibrated using a standard weight.The weight difference before and after sampling is the particle weight.
The PM 2.5 and trace gases data collection was performed from 1 January to 31 January 2013 at the 10 observation stations.The size-segregated aerosol particles were continuously collected for 23 h and analyzed by IC at Nanjing, Suzhou, Hangzhou, Lin'an and Ningbo form 18 January to 24 January 2013.

Ground Weather and NECP Data
In the present study, we selected visibility and relative humidity (RH) data at four times (2:00, 8:00, 14:00 and 20:00) on a single day from 55 ground-based weather stations from Jan. 1-31, 2013 and during 1980-2009 (http://cdc.cma.g ov.cn), and we calculated the daily mean value.

PSCF Analysis
Ashbaugh et al. (1985) developed the potential source contribution function (PSCF).It has been applied in a series of studies over a variety of scales (Lucey et al., 2001;Wang et al., 2009).In this study, we calculated the 48-hour backward trajectory for each hour at the height of 500 m at the ten observation stations, and meteorological data were collected from the GDAS archive (http://ready.arl.noaa.gov/HYSPLIT.php).
The calculated trajectories were used to analyze possible long-range sources contributing to select the pollution concentration observed in YRD region's atmosphere.For this purpose, potential source contribution function (PSCF) analysis (Lucey et al., 2001) was employed.The objective of this method is to develop a probability field which suggests the likely source locations of the measured episodic pollution concentrations.The PSCF of an element χ in subregion ij is given by Eq. ( 1): where; n ij is the total number of trajectory segments in the ij th sub-region during the entire study period, m ij is the total number of "episode or polluted'' trajectory segments in the same ij th sub-region during the same period, and W(n ij ) is an arbitrary weight function .
Since the PSCF is computed as a ratio of the counts of selected events (m ij ) to the counts of all events (n ij ), it is expected that m ij will be relatively smaller than n ij .Values related to sparse trajectory coverage of the more distant grid cells may result in PSCF xj with high uncertainty in the apparent high value (Zeng and Hopke, 1989).For large values of N, there is more statistical stability in the calculated value.Thus, to reduce the effect of small values of n ij , an arbitrary weight function W(n ij ) is multiplied into the PSCF value to better reflect the uncertainty in the values for these cells.W(n ij ) is the amount of endpoints per cell, calculated by dividing the PSCF grid to total number of trajectory endpoints.The weight function used is given in Eq. ( 2

The Sulfur Oxidation Ratio (SOR) and Nitrogen Oxidation Ratio (NOR)
The concentrations of SO 4 2-and NO 3 -depended on the concentrations of SO 2 and NO 2 and the atmosphere conversion of SO 2 to SO 4 2-and NO 2 to NO 3 -.The sulfur and nitrogen oxidation ratios were defined as follows (Chen et al., 2003;Wang et al., 2005):

Synoptic Conditions and Air Quality Index (AQI) during the Haze Pollution Events
A continuous and wide range of heavy haze pollution episodes occurred in January 2013, which was due to the specific atmospheric circulation.The range included two troughs and one ridge in middle-high latitudes of the Eurasian region with a strong blocking anticyclone in Western Siberia and a cold vortex at the high altitude of East Asia.The visibility was much higher than the mean climatological value during Jan.1-5, and it was up to 15.8 km on Jan.1, which was related to strong cold air passing through.Strong cold air favored the pollutants' diffusion, the average concentration of PM 2.5 on Jan. 1-4 was 64.6 ± 22.1 µg m -3 , with the lowest value of 28.6 µg m -3 during 14:00 on Jan. 2 to 7:00 on Jan. 3. Fig. 2 shows the average AQI were below 100 and was the lowest value of 64 on Jan. 3 in the YRD cities, Nantong and Yangzhou had the lowest AQI value of 48 and 39.It has been reported that the temperature dropped by 5-9°C (Guan et al., 2013) and that the wind speed increased greatly.We found that the cold air belonged to the horizontal trough, longitudinal style.However, it was dominated by a weak ridge or a straight westerly at 500 hPa on Jan. 6; additionally, there was anticyclonic circulation at 850 hPa and the ground isobar was relatively sparse, indicating that the wind speed was small and the weather was clear on the low level of the YRD region.As a result, it was easy to keep the water vapor at a saturation state.Guan et al. (2013) calculated that the RH was larger than 90%, even up to 100%, in local areas at the altitude of 2 m, and it obviously decreased with height in the north China, Jiangnan area, western south China, and southwest areas.Fig. 2(a) shows that the RH was higher than the mean climatological value (73.3%) during the haze events, and it was up to 85.5% on Jan. 1-14.It was controlled by a cold high-pressure field at the ground level.Air pollutants in the YRD region is likely to be easily accumulated with a straight circulation at high altitude and a uniform pressure field at the ground in January.Stable weather condition and high RH favored the accumulation of PM 2.5 .Fig. 3 shows the PM 2.5 concentration increased continuously during Jan. 3-14 in various cities in the YRD region.For example, the maximum PM 2.5 value of 300 µg m -3 occurred at 4:00 on Jan. 13.High PM 2.5 level deteriorated the air quality.Fig. 2 shows the AQI values were mostly above 100 and up to 267 and 268 in Wuxi and Nanjing.
The temperature was stable at the middle-high latitude due to the confrontation of cold air masses to the warm air masses before Jan.13.A strong, cold air pressed southward on Jan. 16-17, resulting in an increase of the visibility, exceeding 10 km (less than the mean climatological value) on Jan. 17-19.Fig. 2 shows the air quality was improved quickly when cold air passed through.The AQI values decreased below 100 for major cities on Jan. 17.The PM 2.5 concentration was below 60 µg m -3 except Hangzhou, Lin'an and Jiaxing, with an average concentration of 65.5 µg m -3 .The PM 2.5 level rose again after the impacts of cold air finished, the haze process occurred as well.The temperature stabilized again on Jan. 22-31.Fig. 4 shows that the ground pressure field and ground speed from the north were both smaller than the mean climatological value, which indicated that the strength of cold air in January was weak compared with the previous observations.The conditions of warm temperature and weak cold air supplied sufficient water vapor and   favored a stable atmospheric stratification on the ground, which were beneficial to haze formation in January.

Variations of the PM 2.5 and Trace Gases Time Series of the PM 2.5 and Trace Gases
Fig. 3 shows that the PM 2.5 was the principle pollutant during the 6 haze events; additionally, the CO and NO 2 concentrations increased at different levels.The concentration of PM 2.5 , SO 2 , NO 2 and CO increased significantly, with average values of 128.6, 48.5, 78.1 µg m -3 and 1.5 mg m -3 on haze days, and were 64.6, 36, 52.5 µg m -3 and 1.1 mg m -3 on clear days.However, the O 3 concentration had small difference on haze and clear days, with values of 26.9 and 24.1 µg m -3 .Considerable particulate matters had strong scattering ability for solar radiation, leading to weak photochemical reactions.Which indicated that haze event may be caused by secondary aerosols through heterogeneous reaction under high humidity in winter.In addition, Table S1 shows that the correlation coefficients between PM 2.5 and O x (NO 2 + O 3 ) were relative low for major cities in the TRD region, which were mostly below 0.4, excepting for Lin'an and Wuxi with values above 0.5 and were even lower on haze days (Lin'an: 0.47, Wuxi: 0.46, the other cities: 0.17-0.31).Therefore, we considered that the photochemical reactions weakened during haze episodes.Table S1 demonstrates that the correlation coefficients of the hourly PM 2.5 with CO and NO 2 were in the range of 0.44-0.77and 0.57-0.69,with an average of 0.66 ± 0.11 and 0.64 ± 0.05 for different cities, respectively.CO is a product of incomplete fossil fuel combustion and/or biomass and biofuel burning, and it is mainly emitted from vehicle exhaust in urban areas (Gou et al., 2004).Boogaard et al. (2012) observed that the NO 2 concentration increased with the increase in the number of road vehicles in the Netherlands.Vehicle exhaust, such as CO, NO and NO 2 , may accumulate near the ground under unfavorable conditions during haze events from section 3.1.
Fig. 3 shows the PM 2.5 concentration changed greatly during haze episodes.The maximum level of PM 2.5 was 332, 320, 305, 324 and 349 µg m -3 in Nanjing, Huzhou, Jiaxing, Suzhou and Wuxi, and was even up to 292 and 252 µg m -3 for Lin'an and Zhoushan.Hence, the haze events in the YRD region revealed a significantly regional feature.In addition, the concentrations of CO and NO 2 changed similarly with PM 2.5 .The maximum level of CO was 4.2, 5.3, 3.3, 3.0, 3.3 and 3.1 mg•m -3 in Hangzhou, Suzhou, Nanjing, Ningbo, Shaoxing and Jiaxing.The maximum level of NO 2 was 240, 225, 218, 170, 169 and 184 µg m -3 in Suzhou, Wuxi, Shaoxing, Hangzhou, Huzhou and Nanjing.Those cities in the YRD region had large population and developed industry, numerous pollutants was thus emitted by vehicle exhaust and industrial process.Table 2 shows large pollutant concentrations during the first three haze episodes, with the maximum average PM 2.5 level of 152 µg m -3 in HD3.Fig. 2 exhibited high relative humidity mostly above 80% in HD3, indicating that high RH favored the fine particle formation on haze days.The SO 2 concentrations also increased during haze events (Fig. 3).However, the correlation coefficients of SO 2 with PM 2.5 were relatively low, which were mostly < 0.5 from Table S1.Fig. 3(a) reveals that the 6 haze events can be classified into the SO 2 -rich (HD1 and HD2) and SO 2 -poor (HD3, HD4, HD5 and HD6) types.Table 2 shows that the SO 2 levels of the SO 2 -rich type were generally high, with values that were greater than 60 µg m -3 , which were > 1.8fold higher than those on clear days.However, the levels were similar to those on clear days for the SO 2 -poor haze types type, with values of approximately 40 µg m -3 .The relative humidity was mostly above 80% during the HD3 episode (Fig. 2), promoting the formation of secondary aerosols via the reaction of SO 2 and NO 2 on the surface of aerosols and increasing the aerosols hygroscopicity.As a result, more trace gases are converted into secondary aerosols (Zhu et al., 2011).In addition, He et al. (2014) reported that the transformation process of SO 2 to sulfate could be advanced under low O 3 and high NO 2 level conditions.We found that SO 4 2-was dominant among the WSIs in PM 2.1 in the YRD region (Fig. 1), constituting 36.1-41.5% of the total ion concentrations.Table 2 shows that NO 2 concentrations were commonly high on haze days, with a maximum of 95.3 µg m -3 on HD2, and they were 1.4-1.8-foldhigher than on clear days.The O 3 levels were relatively low on haze days, and they were sometimes lower than those on clear days (Fig. 3 and Table 2).Table S1 lists that the correlation coefficients of PM 2.5 with O 3 ; they were low, with an average of -0.18 ± 0.14, demonstrating that O 3 had a small contribution to the PM 2.5 mass.

PSCF Analysis of the PM 2.5 and Trace Gases
Figs. 5 and S2-S5 give the potential sources and source strength of PM 2.5 , SO 2 , NO 2 , CO and O 3 , deep color refers to great impacts of the corresponding area to the pollutants.A high banding PSCF of PM 2.5 value (> 0.9) distributed at Nanjing, Lishui and Yixing, and a high regional PSCF value (0.7-0.9) existed at Wuxi, Suzhou, Jiaxing, Hangzhou, Shaoxing and Huzhou.These cities belong to economically developed areas, with numerous particulate sources.Fig. 5 shows the PSCF value for PM 2.5 in Taihu was much lower than the surrounding areas although the Taihu situated at the centre of the YRD region, which illustrated that Taihu had obvious impacts on the air quality in the YRD region.Moreover, a low PSCF value for PM 2.5 occurred at Ningbo, Lin'an and Zhoushan.Ningbo was adjacent to Donghai and Hangzhou Bay, which favored the pollution diffusion, the PSCF value for PM 2.5 was low as result.Zhoushan belongs to sea inland with less industrial sources, leading to the minimum PSCF values.As a regional background station in the YRD region, Lin'an had less industrial sources; however, the PSCF value for PM 2.5 was mostly above 0.6, indicating a regional PM 2.5 pollution in the YRD region.
Fig. S2 shows the PSCF value for SO 2 was high (> 0.9) in Nanjing, Wuxi, Suzhou, Jiaxing, Hangzhou and Huzhou.The source distribution of SO 2 over Taihu was consistent to the surrounding areas, suggesting to a uniform distribution for SO 2 than PM 2.5 .Furthermore, the PSCF value was relative large in the centre cities of the YRD region and decreased with the increasing distance from the centre city.The PSCF distribution of NO 2 was similar to SO 2 from Fig. S3, PSCF value decreased significantly with the increasing distance from the centre city, which might be probably associated with residence time of the pollutants.Fig. S4 shows the PSCF distribution of CO was similar to that of PM 2.5 , which illustrated a common source for the two parts.
Fig. S5 indicates that the sources of O 3 were significantly different from the other components, and included Wuxi, Suzhou and southeast of An'hui.Additionally, the source strengths were mostly below 0.5, illustrating a small contribution of O 3 in these pollution areas.High PSCF values were focused on Wuhu and Xuancheng, where the PM 2.5 level was low and the precursors of CO, NO 2 and SO 2 had medium level.Strong photochemical reaction occurred under intense solar radiation due to weak PM 2.5 extinction effect.Therefore, the PSCF value for O 3 was low in the centre cities due to high PM 2.5 concentrations.

Size Distributions of the Mass and Water-Soluble Ion Concentrations Spectral Distributions of the Mass and Water-Soluble Ion Concentrations
Fig. 6 shows that the mass concentrations were mostly centralized for particle sizes smaller than 2.1 µm.The respective PM 2.1 concentrations were 90.5, 121.5, 108.9, 76.6 and 89.5 µg m -3 and the respective ratios of PM 2.1 /PM 10 were 63.8%, 65.7%, 65.4%, 50.3% and 68.6% in Nanjing, Suzhou, Lin'an, Ningbo and Hangzhou.Low values in Ningbo were from missing sampling particles with sizes < 0.43 µm.The concentrations of WSIs in PM 2.1 were 70. 1, 79.7, 44.4, 53.6 and 47.5 µg m -3 in PM 2.1 , accounting for 77.3%, 77.0%, 81.0%, 74.4% and 79.4%, respectively, of the total ion concentrations in Nanjing, Suzhou, Lin'an, Ningbo and Hangzhou, which was similar to the distributions of the  PM mass.The proportions of water-soluble ion concentrations in PM 2.1 to the total PM 2.1 mass were 77.5%, 65.6%, 40.8%, 70.0% and 53.1% in Nanjing, Suzhou, Lin'an, Ningbo and Hangzhou, respectively.Fig. 1 indicates that the NH 4 + , SO 4 2-and NO 3 -ions were dominant, constituting 86-90.9% of the ion concentrations in PM 2.1 , demonstrating that secondary ions were predominant, which is consistent with the above analysis.
Fig. 7 shows that the mass spectra of different cities exhibited unimodal distributions, with peaks at 0.43-0.65 µm.However, there were still slight differences among the distributions for various cities due to their locations (Fig. 7(a)).For example, small peaks at 4.7-5.8µm and 9.0-10.0µm were present in the coastal city of Ningbo, which were related to the large wind speeds stirring up high levels of coarse particles.Moreover, there was a maximum peak in Lin'an, with the value of 134.5 µg m -3 µm -1 .The high particle concentration at 0.43-0.65 µm was focused on Jan. 18-20 in Lin'an, with values of 149.2, 203.6 and 140.5 µg m -3 µm -1 .The particle concentrations in the corresponding size segment were147.9, 92.8 and 164.7 µg m -3 µm -1 in Hangzhou.According to the ground weather map published by the Hong Kong Observatory (http://gb.weather.gov.hk), it was controlled by easterly flow on Jan. 17, by westerly flow on Jan. 18-19 and by easterly flow under peripheral of low pressure on Jan. 20 over Hangzhou.It can be seen that high particle concentration was due to the foreign transport.Additionally, the high RH in the mountainous area of Lin'an favored the fine particles' growth and accumulation.Both of which caused to the high particle concentration at 0.43-0.65 µm on Jan. 18-20 in Lin'an compared to the other cities.As a regional background station, fine particles probably originated from external transportation from low-level local sources, illustrating that the haze pollution in the YRD region had regional features.
The spectral distributions of WSIs varied greatly with different cities (Figs. 2-, F -, NO 2 -and NO 3 -were similar in various cities, with concentrations focused in particle size smaller than 2.1 µm.As a coastal city, Ningbo had a higher RH, leading to the shifting to larger size segment for the secondary WSIs, ions of SO 4 2-, NO 3 -and NH 4 + at ~2 µm had high concentrations as result.Ions of F -and NO 2 -in winter were mainly from coal consumption and waste Incineration.Fig. S6 shows the concentration of F -ion in Lin'an was approximate to the other cities, and was even larger than Suzhou in the particle size of 0.43-0.65 µm, suggesting to the regional pollution feature.The F -concentration was the largest in the new particle size segment in Nanjing from Fig. S6, which may Fig. 7. Spectral distributions of the mass and water-soluble ion concentrations during observation period.be due to the coal burning process.In Lin'an, the NO 2 concentration in fine particle size segment was the largest from Fig. S6, the NOR in PM 2.1 and PM 1.1 was 0.18 and 0.14 from Table 3, which was slightly lower than Nanjing, this phenomena demonstrated that the conversion of NO 2 to nitrate might be an important nitrate source.Ca 2+ and Mg 2+ , from local sources, varied with different cities, and there were mostly low concentrations on haze days.The peaks of Na + from sea salt mainly appeared in the large size segments in Hanzhou and Ningbo. Fig. 8 shows that the ratios of NH 4 + in coarse and fine particles were distinct, which were mostly below 0.2 in PM 2.1-10 and above 0.2 in PM 2.1 and PM 1.1 .The ratios were 0.2-0.7 in coarse particles and 0.2-0.5 in fine particles for NO 3 -, and they were identical for the two types of particles for SO 4 2-, with values in the range of 0.2-0.6.Fig. 2 shows that the relative humidity was high on Jan. 18-24 in the YRD region, which favored the ternary nucleation of NH 3 -H 2 SO 4 -H 2 O, increasing the NH 4 + ratio in fine particles.In addition, low O 3 and high NO 2 levels were also observed on Jan. 18-24, which favored the conversion of SO 2 to SO 4 2through the catalysis of NO 2 under high RH (He et al., 2014).As a result, the ratios of NO 3 -in fine particles decreased.

The Sulfur Oxidation Ratio (SOR) and Nitrogen Oxidation Ratio (NOR)
Table 3 shows that SOR and NOR were relatively low, with values in the range of 0.04-0.08 and 0.02-0.05 in PM 2.1-10 on haze days in different cities, indicating that the conversions of SO 2 and NO 2 to sulfate and nitrate can be ignored in coarse particles.Related studies found that the conversion of SO 2 to sulfate may occur if SOR > 0.1 (Kaneyasu et al., 1995;Zhang et al., 2011).SOR in PM 2.1 and PM 1.1 were 0.21-0.35and 0.11-0.28,respectively, demonstrating an occurrence of secondary formation of sulfate.However, NOR in PM 2.1 and PM 1.1 had large discrepancies among different cities, which were relatively high in Nanjing, Suzhou and Lin'an, with values above 0.1, and were low in Hangzhou and Ningbo (Table 3).Fig. S3 shows the NO 2 source were mainly distributed at Wuxi, Suzhou, Jiaxing and Hangzhou, and also occurred in a small range of Nanjing.While the NOR of fine particles was the largest in Nanjing and the lowest in Hangzhou, which demonstrated that the conversion of NO 2 to nitrate did not necessarily occur when the potential sources of NO 2 were strong.Zhoushan.The sources of SO 2 and NO 2 are similar, which was relative large in the centre cities of the YRD region and decreased with the increasing distance from the centre city.The sources of O 3 were significantly different from the sources of other components; the sources were mainly Wuxi, Suzhou and southeast of An'hui.
Water-soluble ion concentrations accounted for 77.3%, 77.0%, 81.0%, 74.4% and 79.4% of the total ion concentrations in PM 2.1 ; additionally, ions of NH 4 + , SO 4 2and NO 3 -were dominant, constituting 86-90.9% of the total ion concentrations in PM 2.1 for these five cities.The mass concentration, NH 4 + , K + , Cl -, SO 4 2-F -, NO 2 -and NO 3 spectra of different cities exhibited unimodal distributions, with peaks at 0.43-0.65 µm.The ratios of NH 4 + in coarse and fine particles were distinct and were mostly below 0.2 in PM 2.1-10 and above 0.2 in PM 2.1 and PM 1.1 .The ratios for NO 3 -were 0.2-0.7 in coarse particles and 0.2-0.5 in fine particles.The SOR in PM 2.1 and PM 1.1 were 0.21-0.35and 0.11-0.28,demonstrating an occurrence of secondary formation of sulfate.However, the NOR in PM 2.1 and PM 1.1 had great discrepancies among different cities, which were relatively large in Nanjing, Suzhou and Lin'an, with values above 0.1, and small in Hangzhou and Ningbo.

Fig. 1 .
Fig. 1.The distributions of observation stations and the ratios of water soluble ions in PM 2.1.

Fig. 2 .
Fig. 2. Time series of the visibility, relative humidity and AQI of major cities in the YRD region.

Fig. 3 .
Fig. 3. Time series of stacked plot of PM 2.5 , CO, NO 2 , SO 2 and O 3 for major cities in the YRD region during observation period.

Fig. 4 .
Fig. 4. The differences in the pressure and wind fields between the mean climatological value in January (1980-2012) and January 2013

Fig. 5 .
Fig. 5. PM 2.5 distributions of the PSCF analysis by 48 h back-trajectory at a height of 500 m during observation period.

Fig. 8 .
Fig. 8. Ternary diagrams for the ratio of sulfate, nitrate and ammonium during observation period.

Table 1 .
The geographic information for the observation stations.

Table 2 .
The average concentrations of trace gases and PM 2.5 on clear and haze days in January 2013.