Particulate Matter Distributions in China during a Winter Period with Frequent Pollution Episodes ( January 2013 )

Particulate matter distributions in China during January 2013 were analyzed using hourly PM2.5 and PM10 concentrations from 74 cities. Five haze episodes occurred in this month. Both PM2.5 and PM10 concentrations increased rapidly at the beginning of January 2013 and remained at high levels throughout the month with monthly average values of 128.7 and 184.4 μg/m, respectively. On January 12, the most polluted day in this month, 13 cites were severely polluted with daily average PM2.5 concentrations greater than 300 μg/m, and 18 cities were heavily polluted with daily average PM2.5 concentrations between 200 and 300 μg/m. These episodes often occurred in a large spatial domain with the North China Plain as the most polluted area, including Jing-Jin-Ji area (Beijing, Tianjin, and Heibei provinces). Both PM2.5 and PM10 had good correlations with ambient CO, NO2, and SO2 concentrations. High PM concentrations often occurred at low wind speeds and high relative humidity. In addition, PM levels in January 2013 were compared with those from other international cities.


INTRODUCTION
Along with fast economic development, especially the rapid increase of fossil fuel consumption, electricity generation, and number of motor vehicles, air pollution has become a severe environmental problem in China (Wang and Hao, 2012).Among different air pollutants, atmospheric particulate matter (PM), especially those with an aerodynamic diameter of 2.5 µm or less (PM 2.5 ), has drawn significant attention.Atmospheric PM plays an important role in urban and regional air pollution (Querol et al., 2004;Shimadera et al., 2013), visibility reduction (Appel et al., 1985;Wang et al., 2012a), and global climate change (Booth et al., 2012;Randles et al., 2013).They can also cause serious adverse health effects.The associations between exposures to fine particles and mortality and morbidity were widely discussed in the past decade (Pope and Dockery, 2006;Wong et al., 2008;Anenberg et al., 2010;Walsh, 2014).Recent studies in China reported increasing incidence rates for cardiovascular and respiratory diseases and intensive care visits due to PM exposure (Kan et al., 2012;Cheng et al., 2013;Yang et al., 2013;Zhang et al., 2013).In 2010, ambient particulate matter pollution (PM 2.5 ) has become the fourth leading risk factor for disability-adjusted life-years in China (Yang et al., 2013).
Many studies investigated PM 2.5 pollution in megacities or regions in China, such as Beijing (He et al., 2001), Shanghai (Wang et al., 2006), the Yangtze River Delta Region (Wang et al., 2012b), and the Pearl River Delta Region (Peng et al., 2011).Since PM pollution in China is frequently occurring as regional events, it is important to characterize nation-wide PM 2.5 concentrations simultaneously.A few studies have analyzed PM 2.5 pollution in a large spatial domain in China based on literature reviews (Yang et al., 2011a), ground level PM 2.5 measurements in 14 cities (Cao et al., 2012), aerosol optical depth (AOD) data from ground sites (Wang et al., 2011) or satellite observations (He et al., 2012).However, these studies have limitations when characterizing nation-wide PM pollution levels, e.g., lack of data from well compared measuring methods, insufficient number of sites, high uncertainties when deriving PM 2.5 concentrations from AOD data.
Since June 2000, the Ministry of Environmental Protection of China (MEP) started to publish a daily air pollutant index (API), an integrated index calculated using daily concentrations of SO 2 , NO 2 , and PM 10 (particulate matter with an aerodynamic diameter of 10 µm or less).API data has been used to estimate PM 10 concentrations in 86 Chinese cities and to analyze the long-term variation of PM 10 across China (Yang, 2009;Qu et al., 2010;Cheng et al., 2013).In DATA SOURCES PM 2.5 , PM 10 , NO 2 , SO 2 , CO, and O 3 hourly concentrations at 484 monitoring sites in 73 cities (without Beijing, the list is given in SI) from December 24 th 2012 to January 31 th 2013 were downloaded from the China National Urban Air Quality Real-time Publishing Platform (http://113.108.142.147:20035/emcpublish/)supported by the MEP.For a given city, data from all sites were averaged to represent its average pollution level.Data from different cities were added to this publishing platform over time.The first city was added on December 15 th .By December 24 th , data from 73 cities (without Beijing) had been added in this platform.On January 1 st 2013, this platform was officially introduced to the public.Since the data from Beijing was not included in this platform until January 17 th , data from Beijing Air Quality Automatic Monitoring System (http://zx.bjmemc.com.cn/)supported by the Beijing Municipal Environmental Monitoring Center (BMEMC) were used instead to characterize the air quality during January 2013.Since September 2012, the Beijing monitoring system publishes hourly concentrations of the above six pollutants at 35 sites including the 12 sites later used by the national publishing platform.For some reason, the hourly PM 2.5 data in Beijing from January 1 st to January 12 th were not collected.For this period, PM 2.5 data released by the U.S. Embassy Beijing Air Quality Monitor (http://beijing.usembassy-china.org.cn/070109air.html) were used to complement the analysis.Before the end of 2012, daily or hourly PM 10 concentration measured by the government was not available to the public.The air pollution index (API) of Beijing from July 2000 to December 2012 was downloaded from the National Daily Air Quality Report (http://datacenter.mep.gov.cn/) supported by the MEP.Daily PM 10 concentrations were then estimated from the API value when PM 10 was the primary pollutant (see SI for more information).The same method has been used previously to study PM 10 pollution in Chinese cities (e.g., Qu et al., 2010;Cheng et al., 2013).Together with the PM 10 data from the BMEMC, the long-term variation of PM 10 in Beijing from July 2000 to December 2013 can be analyzed.
The Tapered Element Oscillating Microbalance (TEOM) and the Automatic Beta Radiation Attenuation Monitor are two kinds of instruments used at these sites to measure PM 2.5 and PM 10 .Their calibration and data quality control are supported by China National Environmental Monitoring Center.The same dataset is also used for air quality compliance practices in China.Beijing's PM 2.5 data from the monitoring system and those from the U.S.

Overview of PM Pollution in January 2013
In January 2013, severe pollution episodes with high particulate matter concentrations occurred in China.As shown in Fig. 1, both PM 2.5 and PM 10 concentrations increased rapidly at the beginning of January 2013 and remained at a high level throughout the month.The same patterns were observed for other pollutants such as NO 2 , SO 2 , and CO (Fig. S-2).Five pollution episodes can be identified from the temporal profiles of these five pollutants, i.e., January 7 th to 8 th (peak on 8 th ), 10 th to 16 th (peak on 12 th ), 18 th to 19 th (peak on 19 th ), 21 th to 23 th (peak on 22 th ), and 27 th to 30 th (peak on 28 th ).As a result of these episodes, the average PM 2.5 and PM 10 concentrations in 74 cities during January were 128.7 and 184.4 µg/m 3 , respectively.For comparison, the average PM 2.5 concentrations in London and Los Angeles during the same month were 17.5 and 15.1 µg/m 3 , respectively.Among these episodes, the one from 10 th to 16 th was the heaviest and lasted the longest.During these seven days, the average PM 2.5 concentration in 74 cities was 159.2 µg/m 3 .36 sites from 24 cities (there  (AQG 2005), respectively.Class II values in CNAAQS apply for residential, commercial, cultural, industrial, and rural areas.In a single box plot, the central rectangle spans the first quartile to the third quartile.The segment inside the rectangle shows the median.The whiskers above and below the box show the locations of the 10 th and 90 th percentiles.The points above and below the whiskers show the 5 th and 95 th percentiles.are 164 sites in total for these cities) reported hourly PM 2.5 concentration higher than 900 µg/m 3 and 125 monitoring sites from 43 cities reported hourly PM 2.5 concentration higher than 500 µg/m 3 .On January 12 th , the most polluted day in this month, 14 cities were severely polluted with daily average PM 2.5 value greater than 300 µg/m 3 and 18 cities were heavily polluted with daily average PM 2.5 concentration between 200 and 300 µg/m 3 .This large-scale regional event happened in middle and eastern China.The most polluted cities were Langfang, Xingtai, Baoding, and Shijiazhuang which are all in Hebei province.Their daily PM 2.5 levels on January 12 th were 718.1, 695.3, 666.8, and 654.6 µg/m 3 , respectively.PM 2.5 pollution in China showed significantly regional characteristics with the North China Plain, including Jing-Jin-Ji area (Beijing, Tianjin, and Heibei province), being the most polluted area (Fig. 2 and Table 1).Fig. 2(a) indicates that a large spatial domain in China was simultaneously experiencing severe PM 2.5 pollution on 12 th January, the most polluted day during the month.Hebei was the most polluted province in China during January 2013.It is noticed that two cities (Zhangjiakou and Chengde) in the north of the Jing-Jin-Ji area reported PM 2.5 concentrations of 50-75 µg/m 3 (light green dots, Fig. 2(a)) whereas the other cities in the Jing-Jin-Ji area reported much higher PM 2.5 concentrations.Zhangjiakou and Chengde are much less developed.In addition, the Jundushan Mountain and the Yanshan Mountain prevent pollutants transporting from other developed area including Beijing.According to the monthly average PM 2.5 concentration in January 2013 (Fig. 2(b) and Table S1), the most polluted six cities (Xingtai, Shijiazhuang, Handan, Baoding, Hengshui, and Langfang) and the 8 th most polluted city (Tangshan) are all in Hebei province.The 7 th , 9 th , and 10 th most polluted cities are Jinan (Shandong province), Zhengzhou (Henan province), and Xi'an (Shaanxi province), respectively.Similar regional distributions of air pollution in China was reported previously using other methods.middle and eastern China as one of the four major haze areas in China (Zhang et al., 2012).PM 10 measurement data from the national monitoring network (Qu et al., 2010;Cheng et al., 2013;Wang et al., 2013) and modeled results on PM 10 and PM 2.5 (Chen et al., 2013;Shimadera et al., 2013) reported consistent results with this study based on PM 2.5 data from the national monitoring network, i.e., PM pollution in middle and eastern China is severer than in any other area in China.PM 2.5 concentration is typically highest in winter due to the unfavorable meteorological conditions and emissions from winter heating in northern China (He et al., 2001;Duan et al., 2006;Jahn et al., 2011;Zhao et al., 2011).
The average PM 2.5 concentration in 14 Chinese cities during January 2003 was found to be 161.7 µg/m 3 (Cao et al., 2012), which is 26% higher than that of 74 Chinese cities in January 2013.However, significantly more public attention occurred during the pollution episodes in January 2013 due to their large spatial coverage, high frequency, long lasting time, and high peak hourly PM 2.5 concentrations.In January 2013, PM 2.5 concentrations in 74 cities frequently violated air quality standards.Daily and annual average PM 2.5 concentrations were included in the new China National Ambient Air Quality Standard (CNAAQS, 2012) and the World Health Organization Air Quality Guideline (AQG, 2005).On January 12 th , the most polluted day in this month, 63 of the 74 cities had daily average PM 2.5 concentrations greater than 75 µg/m 3 (CNNAQS, 2012).On January 3 rd , the cleanest day in this month, 61 of the 74 cities had daily average PM 2.5 concentrations meeting CNAAQS (< 75 µg/m 3 ), while only 5 cities met AQG (< 25 µg/m 3 ).From January 5 th to 31 st , the percentage of cities with daily PM 2.5 concentration satisfying CNNAQS was in the range of 12% to 39%.During the same period, the number of cities with daily PM 2.5 concentrations meeting AQG was no more than 3.During the severest episode (Jan.10 th to 16 th ), over 70% of the 74 cities did not meet CNNAQS.Hourly PM 2.5 concentrations even reached 1000 µg/m 3 in eleven heavily polluted cities including Shijiazhuang, Xingtai, Xi'an, and Wuhan.

Comparison with Other International Cities and Historical Data
Beginning in the last century, air pollution has affected urban environments all over the world.The most wellknown air pollution events include the 1952 London Great Smog and the Los Angles Photochemical Smog.After many years of battling air pollution, these two cities have improved their air quality significantly.While daily PM 2.5 and PM 10 concentrations in London and LA were no more than 35 µg/m 3 in January 2013, those in Chinese cities were much higher.Fig. 3 shows monthly average PM concentrations in Beijing, Shanghai, and Xingtai (the most polluted city in January 2013) comparing to London and Los Angles.PM 2.5 in Xingtai, 324.2 µg/m 3 , was 18 and 21 times higher than those in London and LA, respectively.PM 2.5 in Beijing was 9 and 11 times higher than those in London and LA, respectively, while PM 2.5 in Beijing in 1992-1993 was only 5 times higher than that in Los Angeles in 1986 (Zhang and Friedlander 2000).In January 2013, Shanghai had lower PM pollution than Beijing.Its monthly  (Duan et al., 2006) and winter 2005-2008(Yu et al., 2011););Shanghai: winter 2003-2005(Wang et al., 2006), winter 2005-2009(Zhao et al., 2011), and Dec 2009 (Wang et al., 2012b).W i n t e r 2 0 0 1 -2 0 0 2 W i n t e r 2 0 0 5 -2 0 0 6 W i n t e r 2 0 0 6 -2 0 0 7 W i n t e r 2 0 0 7 -2 0 0 8 W i n t e r 2 0 1 2 -2 0 1 3 W i n t e r 2 0 0 3 -2 0 0 5 W i n t e r 2 0 0 5 -2 0 0 9 Beijing Shanghai London 100 10 average PM 2.5 concentration was 5 and 6 times higher than those in London and LA, respectively.
Historical PM data in London, Beijing, and Shanghai were also included in Fig. 3.The maximum daily PM 10 concentration during the London Great Smog was estimated to be 4460 µg/m 3 and the monthly average PM 10 level in December 1952 was approximately 3000 µg/m 3 (see SI for more information and references).PM 10 concentrations in Chinese cities were much lower than this level.However, it should be noted that the PM 10 level in London in December 1951, one year prior to the Great Smog event, was only 430 µg/m 3 (Authority, 2002), approximately 2 times higher than PM 10 levels in Beijing in January 2013.PM 10 concentration in Xingtai was 534.8 µg/m 3 , even higher than that in London in December 1951.Compared to previously reported Beijing winter PM levels in the last decade, PM 2.5 in Beijing in January 2013 was higher.Continuous PM 2.5 monitoring data collected since 2012 shows a significantly increasing trend with an average increasing rate of 1.48 µg/(m 3 •month) (Fig. 4(a)).PM 2.5 in Shanghai in January 2013 was at a comparable level with those in other periods.This indicates that in Chinese cities like Beijing and Shanghai PM 2.5 pollution did not improve during the past decade.To further explore PM 10 pollution trends, we took Beijing as an example.Long-term PM 10 concentrations from the national monitoring network is shown in Fig. 4(b).During the period from July 2000 to December 2013, January 2013 was not the month with the highest PM 10 concentration.There are twelve months with average PM 10 concentrations higher than January 2013.Three of them occurred in the winter or autumn season, i.e., October 2004, November 2005, and December 2007, while others occurred in March, April, or May which were mainly due to sand storms.The last time the monthly average PM 10 level was higher than January 2013 is May 2008.Linear regression from July 2000 to December 2013 indicates that PM 10 in Beijing was slowly decreasing at a rate of 0.27 µg/(m 3 •month) with fluctuations.This is attributed to control measures for ensuring good air quality during the 2008 Beijing Olympic Games (Wang et al., 2010;Yang et al., 2011b).Accordingly, PM 10 concentrations in the summer of 2008 were at a relatively low level.However, if the linear regression is taken from August 2008 to December 2013, it shows that PM 10 in Beijing was relatively stable during this period.The monthly average PM 10 concentration in January 2013 was 194 ± 132 µg/m 3 , while the highest level in 2007 was 198 ± 134 µg/m 3 which occurred in December.Similar variations in the trend occurred in other Chinese cities.Based on the same PM 10 data set from the national monitoring network, it was found that most of the 86 Chinese cities had annual PM 10 concentrations continually decreasing from 2001 to 2011 (Cheng et al., 2013;Wang et al., 2013).Wang et al. (2013) pointed out that the reduction of annual PM 10 concentration from 2001 to 2005 was relatively fast.While after 2006, the decreasing rate slowed down and even showed a slight increase in 2010 compared to 2009.In addition, some Chinese cities favored by tourists and cities with rapid economically development showed an opposite increasing trend.

Analysis of Pollution Episodes
During smog events in China, PM concentrations often increased together with gaseous pollutants such as CO, SO 2 , and NO 2 .The latter two also serve as the main gaseous precursors for secondary inorganic aerosol.For instance, sulfate, nitrate, and ammonium together accounted for approximately 55% of PM 2.5 mass concentration in Beijing during January 8 th to 14 th , 2013 (Cao et al., 2014).PM showed positive correlations with SO 2 , NO 2 , and CO based on hourly averaged data (Table 2), indicating that Chinese cities have significant emissions from both coal combustion and vehicles.For Beijing in which most large industries have been moved out and the remaining ones face tight standards, PM had much stronger correlation with CO and NO 2 than with SO 2 .Vehicle emissions in Beijing were reported to contribute approximately 71%-85% of ambient CO concentrations (Hao et al., 2000) and 67%-71% of ambient NO x concentrations (Wang et al., 2009).These findings indicate that in Beijing vehicle emissions play an important role in pollution episodes.A recent study reported that the coexistence of NO x and SO 2 leads to rapid conversion of SO 2 to sulfate, the decrease of SO 2 and the increase of PM during episodes in January 2013 (He et al., 2014), which may also explain the stronger correlations between PM and NO 2 .It should be noted that the correlation between PM and gaseous precursors (SO 2 and NO 2 ) is affected by relative humidity (Table 3).High relative humidity enhances the aqueous phase conversion of gas precursors.Therefore, the fitted slope between PM and precursors decreased remarkably with increasing relative humidity.In contrast, the correlation between PM 2.5 and CO was not affected by relative humidity.
Wind speed and relative humidity are two meteorological parameters that show positive correlation with PM concentrations.We took Beijing as an example again and presented daily PM 2.5 and PM 10 concentrations for January 2013 and other winter periods as a function of daily average wind speed and relative humidity (Fig. 5).Both PM 2.5 and PM 10 increase with decreasing wind speed and increasing relative humidity.Low wind speed means unfavorable atmospheric dilution and dispersion conditions that lead to the accumulation of pollutants.High relative humidity can contribute to particle growth through water uptake and aqueous redox chemistry (e.g., the oxidation of sulfur dioxide to sulfate).The monthly average wind speed in January 2013 was the lowest in the past ten years while the monthly average relative humidity in January 2013 was the highest in the past ten years.During the five pollution episodes in January 2013, the average PM 2.5 concentration was 147.9 µg/m 3 with an average wind speed of 1.90 m/s and an average relative humidity of 72.9%.In February 2013, PM 2.5 concentrations were dramatically reduced to 104 µg/m 3 when higher wind speeds (2.27 m/s) and lower relative humidities (50.9%) occurred.Similar relationships between PM and these two meteorological conditions had been previously observed (e.g., He et al., 2001;Sun et al., 2013).At high wind speeds, high PM concentrations, especially PM 10 , may also occur which is mainly due to fugitive dust.Wind-blown dust can increase PM 10 concentrations significantly (Feng et al., 2011).For instance, during the dust storm in Beijing on April 6 th , 2000, PM 10 concentrations reached a level as high as 1500 µg/m 3 and the storm lasted for 14 hours (Xie et al., 2005).Dust was reported to account for ~20% of PM 2.5 in Chinese cities (Cao et al., 2012).Its contribution to PM 2.5 increased to ~42% when frequent sandstorms invaded Beijing in April 2001 (Yang et al., 2011a).The pollution processes in most middle and east China locations are similar during wintertime.Meteorological conditions with low wind speeds, high relative humidities, and low temperatures occur frequently.Domestic heating in winter generates much more atmospheric pollutants than in summertime.The absence of wind leads to the accumulation of pollutants.The temperature inversion with a warm air layer overlying a cold air at ground level also prevents pollutants from good dispersion.Both gas phase and aqueous phase chemical reactions form secondary pollutants.Accordingly, concentrations of both primary and secondary pollutants increase significantly.During haze episodes, PM concentrations typically increase by a factor of 3-5.On January 3 rd , the cleanest day in January 2013, average PM 2.5 and PM 10 concentrations for 74 cities were 59.9 and 86.5 µg/m 3 , respectively, while on January 12 th , they increased by a factor of 3.4 and 3.2, respectively.This is consistent with previous reports that PM 2.5 concentrations can be 3 times higher on smoggy days than clear days in north China (Li et al., 2011).However, these pollutants were usually cleared out by a strong wind and/or precipitation.

CONCLUSION
We characterized PM pollution in China during January 2013 using data from the national monitoring network.Five pollution episodes occurred during this month.Monthly average PM 2.5 and PM 10 concentrations in 74 cities were 128.7 and 184.4 µg/m 3 , respectively.On January 12 th , the most polluted day in this month, 13 cites were severely polluted and 18 cities were heavily polluted.PM 2.5 pollution in China showed significant regional characteristics with the North China Plain, include Jing-Jin-Ji area (Beijing,

Estimation of daily PM 10 from the API value
For the days when PM 10 was reported as the primary pollutant, daily PM 10 concentrations were derived from the API using the following equation: where C is the concentration of PM 10 , I is the reported API values.I low and I high represent API grading limits that are lower and larger than I, respectively; C high and C low denote the PM 10 concentrations corresponding to I high and I low , respectively.

Fig. 1 .
Fig. 1.Statistical results for daily average (a) PM 2.5 and (b) PM 10 concentrations in 74 cities.The short dash lines and long dash lines are the class II values in the new China air quality standard (CNAAQS 2012) and the WHO guideline(AQG  2005), respectively.Class II values in CNAAQS apply for residential, commercial, cultural, industrial, and rural areas.In a single box plot, the central rectangle spans the first quartile to the third quartile.The segment inside the rectangle shows the median.The whiskers above and below the box show the locations of the 10 th and 90 th percentiles.The points above and below the whiskers show the 5 th and 95 th percentiles.

Fig. 2 .
Fig. 2. PM 2.5 concentrations in 74 Chinese cities: (a) daily average on 12 th January 2013 and (b) monthly average in January 2013.Locations of 74 cities and their monthly average concentrations of six pollutants are given in Table S-1.

Fig. 4 .
Fig. 4. (a) PM 2.5 concentrations in Beijing from February 2012 to December 2013.Before October 2012, only PM 2.5 concentrations at Chegongzhuang site are available.After that, PM 2.5 concentrations were averaged from all 35 sites in Beijing.(b) PM 10 concentrations in Beijing from July 2000 to December 2013.Before 2013, the PM 10 levels were calculated using the API data.

Fig. 5 .
Fig. 5.The correlations between (a) PM 2.5 and wind speed, (b) PM 2.5 and relative humidity, (c) PM 10 and wind speed, (d) PM 10 and relative.Daily average data in Beijing for different months were shown here as an example.

Table 1 .
Table S-1.Monthly average concentrations of six pollutants at different regions in China.
* The Jing-Jin-Ji Area is included in North China Plain.

Table 2 .
Correlation coefficients (r) for five pollutants based on hourly concentrations in January 2013.

Table 3 .
Correlation coefficients (r) between PM 2.5 and gaseous pollutants at different relative humidities (RH).
Tianjin, and Heibei province), being the most polluted area.PM 2.5 and PM 10 are strongly correlated with NO 2 , SO 2 , and CO, indicating that both vehicle emissions and coal combustion contribute to these pollution episodes.This is consistent with the high nitrate and sulfate observed in fine particles.High PM 2.5 and PM 10 concentrations often occurred at low wind speed and high relative humidity.PM pollution levels in Chinese cities were lower than that during the 1952 London Great Smog, but were much higher than London and LA during January 2013.PM 10 concentrations in Xingtai, the most polluted city in January 2013, were even higher than that in London in December 1951, one year prior to the London Great Smog.Though average PM 10 concentration in China showed a decreasing trend during the last decade, the rate of decrease was slowing down.PM 2.5 pollution did not improve in cities like Beijing and Shanghai during the last decade.In Beijing, PM 2.5 even showed a tendency to increase.With the release of the Ten Air Pollution Prevention and Control Measures by the State Council of China in 2013, stricter air pollution control measures will be implemented and lower PM 2.5 level might be anticipated in the future.

Table S -
1. List of 74 national cities in this study and their monthly average concentrations of six pollutants in January 2013.NCP(JJJ) means the Jing-Jin-Ji (Beijing, Tianjin, and Heibei provience) area.It is included in the NCP (North China Plain).YRD and PRD are the Yangtze River Delta and the Pearl River Delta, respectively.