ABSTRACT
ABSTRACT
To identify the characteristics of air pollutants and factors attributing to the formation of haze in Wuhan, this study analyzed the hourly observations of air pollutants (PM2.5, PM10, NO2, SO2, O3, and CO) from March 1, 2013, to February 28, 2014, and used hybrid receptor models for a case study. The results showed that the annual average concentrations for PM2.5, PM10, NO2, SO2, O3, and CO during the whole period were 89.6 μg m−3, 134.9 μg m−3, 54.9 μg m−3, 32.4 μg m−3, 62.3 μg m−3, and 1.1 mg m−3, respectively. The monthly variations revealed that the peak values of PM2.5, PM10, NO2, SO2, and CO occurred in December because of increased local emissions and severe weather conditions, while the lowest values occurred in July mainly due to larger precipitation. The maximum O3 concentrations occurred in warm seasons from May to August, which may be partly due to the high temperature and solar radiation. Diurnal analysis showed that hourly PM2.5, PM10, NO2, and CO concentrations had two ascending stages accompanying by the two traffic peaks. However, the O3 concentration variations were different with the highest concentration in the afternoon. A case study utilizing hybrid receptor models showed the significant impact of regional transport on the haze formation in Wuhan and revealed that the mainly potential polluted sources were located in the north and south of Wuhan, such as Baoding and Handan in Hebei province, and Changsha in Hunan province. Implications: Wuhan city requires a 5% reduction of the annual mean of PM2.5 concentration by the end of 2017. In order to accomplish this goal, Wuhan has adopted some measures to improve its air quality. This work has determined the main pollution sources that affect the formation of haze in Wuhan by transport. We showed that apart from the local emissions, north and south of Wuhan were the potential sources contributing to the high PM2.5 concentrations in Wuhan, such as Baoding and Handan in Hebei province, Zhumadian and Jiaozuo in Henan province, and Changsha and Zhuzhou in Hunan province.
Introduction
Rapid urbanization and industrialization in China put high pressure on the environment, such as from air-quality degradation, water pollution, excess solid waste, resource depletion, and so on. Among these various environmental problems, air pollution has become a matter of great concern in the megacities of China (Chan and Yao, 2008; Hao and Wang, 2005; Shao et al., 2006). The main source of air pollution in most Chinese cities has recently shifted from coal combustion to a mixture of coal combustion and vehicle emissions because of rapidly increasing vehicle numbers (Hao and Wang, 2012; Shao et al., 2006). One of the main air pollutants is fine particulate matter (PM2.5, particles with an aerodynamic diameter less than or equal to 2.5 μm), which is mostly responsible for the regional haze (days with visibility <10 km under conditions of 80% relative humidity) formation, radiative forcing, and adverse public health (Tan et al., 2009; Bao et al., 2010; Hyslop, 2009; Pope and Dockery, 2006; Wang and Hao, 2012; Wang et al., 2014; Wang et al., 2015; Yu et al., 2014a; Zhang et al, 2010). In January 2013, extremely severe and persistent haze pollution affected ~1.3 million km2 and ~800 million people over China with the daily average PM2.5 concentrations exceeding 75 μg m−3 for 69% of days at 74 major cities and a record-breaking daily PM2.5 concentration of 772 μg m−3 (China National Environmental Monitoring Center [CNEMC], 2013). Many studies about the haze formation have been carried out in the most developed and highly populated megacities such as the Yangtze River Delta, the Pearl River Delta, and the Beijing–Tianjin–Hebei regions (Chan and Yao, 2008; Hao and Wang, 2012; Shao et al., 2006; Yu et al., 2014a). However, central China recently has also experienced a series of significant air pollution episodes and the reasons for the air pollution still remain an issue of uncertainty.
As the most populous city in Central China, Wuhan is the capital of Hubei province with a land area of 8494 km2 and a population of more than 10 million. It is a major transport hub with dozens of railways, roads, and expressways passing through the city and connecting to major cities in China and currently is in a boom of construction, with nearly 68 million m2 of the building areas under construction in 2012. In recent years, increasingly occurrence of serious haze caused by rapid urbanization and social and economical developments poses challenges to sustainable development for Wuhan city. On the basis of observations at an urban site and a suburban site in Wuhan city from August 2012 to July 2013, Zhang et al. (2015) found that the annual mean PM2.5 concentrations were 106.5 to 114.9 μg m−3 with sulfate, nitrate, ammonium, and organic matter as dominant components. They also believed that PM2.5 pollution in Wuhan was from regional instead of local sources. It has been estimated that the emissions from industrial activities accounted for 34% of secondary particulate matter, 57% of primary dust, and 45% of total SO2 emissions in Wuhan (Querol et al., 2006). The seasonal patterns of air pollution in Wuhan exhibited strong seasonal distributions, with the lowest values in summer and the highest value in winter (Zhang et al., 2015). This is because during the summer, the monsoon caused a high frequency of intense and long rain episodes, which led to air pollutant removal and low concentrations of PM, while for the rest of the year, especially in the period of November to February, dry and cold continental western winds are more prevalent in Wuhan city (Qian et al., 2010; Querol et al., 2006; Zhang et al., 2014; Zhuang et al., 2014). In addition, due to its unique topography and meteorology, Wuhan is a city sensitive to regional atmospheric aerosols transported from its surrounding provinces. In order to control the air pollution in Wuhan city, it is necessary to carry out an in-depth study of the seasonal variations in regional transport and potential sources in Wuhan. The main objectives of this study are to (1) comprehensively analyze the ambient air quality in Wuhan based on the hourly observations at nine urban sites for the period from March 1, 2013, to February 28, 2014, with temporal (annual, monthly, seasonal and diurnal variations) characteristics of air pollutants be examined together with the meteorological conditions; and (2) use hybrid receptor models to explore the possible impacts of local and regional transport sources on the formation of haze in Wuhan. The outcomes of this study are therefore very useful for the implementation of air pollution mitigation policy in Wuhan.
Observational data and methodology
Observational data
The observational data of hourly air pollutants (PM2.5, PM10, CO, NO2, SO2, and O3) at nine urban monitoring stations—Hanyangyuehu (30.55°N, 114.25°E), Hankou-huaqiao (30.61°N, 114.28°E), Wuchangziyang (30.55°N, 114.30°E), Qingshanganghua (30.61°N, 114.43°E), Zhuan-kouxinqu(30.48°N, 114.15°E), Hankoujiangtan (30.59°N, 114.30°E), Donghugaoxin(30.48°N, 114.39°E), Wujiashan (30.64°N, 114.21°E), and Chenhuqihao (30.30°N, 113.85°E), as shown in Figure 1 in Wuhan were obtained from the national air quality monitoring network operated and maintained by the Ministry of Environmental Protection (MEP) in China (http://datacenter.mep.gov.cn/). To analyze the characteristics of air pollutants in Wuhan, this study analyzed the air pollutants (PM2.5, PM10, NO2, SO2, O3, and CO) hourly observations from March 1, 2013, to February 28, 2014. The measurements for each species were carried out by a set of commercial instruments, and the operations and maintenances of instruments, data assurance and quality control were properly conducted (http://www.cnemc.cn/publish/106/0536/newList_1.html). The data used in this study were obtained from the website of the MEP in China and are publicly accessible (http://106.37.208.233:20035). The meteorological data (temperature, pressure, wind speed, relative humidity) and visibility data used in this study were obtained from http://www.wunderground.com/history/airport/ZHHH.htm
Published online:
01 June 2017![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2017/uawm20.v067.i07/10962247.2016.1240724/20170601/images/medium/uawm_a_1240724_f0001_b.gif"})
Figure 1. Locations of monitoring stations in Wuhan.
To get the spatial movement pictures of heavy haze, the aerosol optical depth (AOD) data at 550 nm obtained from the Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA’s Terra satellite were used in this study. The aerosol optical depth (AOD) is a good indicator of the aerosol mass loading in the vertical column of the atmosphere. As shown in the work of Xin et al. (2014), there were high correlations between the observed AOD and MODIS AOD when the Angstrom exponent was less than 1.5 on the basis of data over the background of North China from 2009 to 2011. They also found that the slopes decreased from 0.97 to 0.83 with the Angstrom exponent increasing (the dominant aerosol size decreasing). Although satellites cannot directly interpret haze clouds under some meteorological conditions, their observations provide a useful way to distinguish different events by their optical characteristics and variations. Owing to a relatively small amount of effective satellite daily data over the study area, the AOD data from MOD08_D3 data (Level-3 data) with a 1° spatial resolution from January 21 to February 7, 2014, were used.
Back-trajectory and receptor model descriptions
Air mass back-trajectory statistical analyses (Sprovieri and Pironne, 2008) and receptor models (Lewis et al., 2003) are often used to gain insights into the source regions and the prevailing transport pathways for airborne particles and gases. In this study, clustering (Harris and Kahl, 1990; Sirois and Bottenheim, 1995), the potential source contribution function (PSCF) (Ashbaugh et al., 1985), and the concentration-weighted trajectory (CWT) (Hsu et al., 2003a; Seibert et al., 1994; Li et al., 2015) were used to identify the different source regions of air pollution at a receptor site. The Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model (http://ready.arl.noaa.gov/HYSPLIT.php) developed by the National Ocean and Atmospheric Administration (NOAA) is capable of establishing source–receptor relationships over long distances. Trajectory cluster analysis based on the trajectory space similarity is used to group a large number of trajectories. Euclidean distance or angle distance (Sirois and Bottenheim, 1995) can be selected in the cluster model. In this work, we used the angular distance to do cluster analysis mainly because our interest was to use the trajectories to locate the direction and sources from which the air masses reaching the receptor site had been transported. Meteorological data input to the model were from the NCEP/FNL (National Centers for Environmental Prediction, Final Analyses) fields obtained from NOAA, which are available every 3 hr with a 1° × 1° spatial resolution. Based on the meteorological fields, the arrival level was set at 100 m, and the backward trajectory model was run eight times per day at starting times of 16:00, 19:00, 22:00, 01:00, 04:00, 07:00, 10:00, and 13:00 UTC (corresponding to 00:00, 03:00, 06:00, 09:00, 12:00, 15:00, 18:00, and 21:00 LT [local time], respectively). The starting locations were the nine urban monitoring stations in Wuhan, and the calculation duration was from January 21 to February 7, 2014. Forty-eight hours was chosen in back-trajectories in this study because it was sufficient to determine probable locations of regional emission sources and explain regional transport pathways. Nine back trajectories for 9 monitoring sites were calculated each time and in total 1229 back-trajectories were obtained for the study period because of some missing data in the observations.
The PSCF method is based on air masses trajectory analysis to identify the regional source areas (Hsu et al., 2003a; Hsu et al., 2003b). PSCF value is the ratio of polluted trajectories number through the grid and the total number of trajectories at the grid. The PSCF value for the cell (i, j) is defined as follows (Ashbaugh et al., 1985; Wang et al., 2009): 
(1)
where nij is designated as the total number of the trajectory segment endpoints at the grid cell (i, j), and mij denotes the number of the polluted trajectory segment endpoints at the grid cell (i, j) with PM2.5 concentrations higher than 75 μg m−3. PM2.5 criterion was set to 75 μg m−3 as the polluted concentration in this study according to hourly China National Air Quality Secondary Standard (CNAQSS). This method can be used to evaluate the potential sources that influence the air quality in the study area. The grids with the high PSCF values are the areas that have high potential contributions to the high polluted value at the receptor site. Because PSCF is a kind of conditional probability, the value of PSCF will appear with larger fluctuation when air masses in each grid retention time are short or the total numbers of trajectories are small. Another limitation of the PSCF method is that grid cells may have the same PSCF values for the concentrations with only slightly higher or very much higher than the criterion.
To overcome the limitation of the PSCF method, the CWT method was used in this study. The CWT analysis is a method to calculate trajectory-weighted concentration in potential source areas and can reflect the effects of the degree of pollution of different trajectories. The CWT method calculates the average weighted concentration (Cij) for the grid cell (i, j) as follows (Hsu et al., 2003a; Wang et al., 2009): 
(2)
where l is the current trajectory and M represents the total number of the trajectories; Cl is the concentration of trajectory l at the receptor site; and Tijl is the time for the trajectory l spent in the grid cell (i, j). A high Cij value implies that the high potential emission sources in the grid cell can result in high polluted values at the receptor site.
Results and discussion
Temporal variations
Annual average concentrations
Table 1 lists the annual average concentrations of PM2.5, PM10, NO2, SO2, O3, and CO in Wuhan and the corresponding CNAQSS. Because there are no annual average concentration standards for O3 and CO, we used 1-hr and 24-hr average concentration standards to evaluate the annual average concentrations of O3 and CO in Wuhan. As shown in Table 1, the annual average concentrations of SO2, O3 and CO were 32.4 μg m−3, 62.3 μg m−3, and 1.1 mg m−3, respectively, which did not exceed the corresponding CNAQSS. It could be found that the annual average concentrations of PM2.5, PM10, and NO2 exceeded the corresponding CNAQSS by 256%, 192%, and 137%, respectively. This reveals that the PM2.5 pollution was the most serious air pollution in Wuhan. We focus on PM2.5 pollution in the following discussions.
Table 1. Annual average concentrations of various air pollutants in Wuhan from March 1, 2013, to February 28, 2014.
Monthly variations
Figure 2 shows that the monthly average concentrations for all pollutants except O3 had the similar monthly variations with the peak values in December and the lowest values in July. For O3, the peak concentrations occurred during the period from May to August and the lowest concentrations occurred from November to February. In December, the increased local emissions from burning fossil fuels for heat made a significant contribution to the enhanced concentrations of PM2.5, PM10, SO2, NO2, and CO. In addition to the emissions, the severe meteorological conditions were also the important factors influencing the pollutant concentrations. As shown in Table 2, the low temperature, low wind speed, and high pressure in December facilitated the accumulation of air pollutants and contributed to high PM concentrations and low visibility. The maximum, minimum, and average visibility values of Wuhan during the whole study period were 27 km, 1 km, and 7.4 km, respectively, and their monthly variations were similar to PM2.5. Figure 3a shows the correlation coefficient between visibility and PM2.5. The high negative correlation coefficient (R = –0.79) indicates that visibility in Wuhan was influenced by the PM2.5 concentrations to a great extent.
Published online:
01 June 2017Figure 2. Monthly variations of ambient air pollutant concentrations in Wuhan from March 1, 2013, to February 28, 2014.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2017/uawm20.v067.i07/10962247.2016.1240724/20170601/images/medium/uawm_a_1240724_f0002_b.gif"})
Figure 2. Monthly variations of ambient air pollutant concentrations in Wuhan from March 1, 2013, to February 28, 2014.
Published online:
01 June 2017Figure 3. Correlations between (a) visibility and daily average PM2.5 concentrations, and (b) daily average O3 concentrations and temperatures in Wuhan.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2017/uawm20.v067.i07/10962247.2016.1240724/20170601/images/medium/uawm_a_1240724_f0003_b.gif"})
Figure 3. Correlations between (a) visibility and daily average PM2.5 concentrations, and (b) daily average O3 concentrations and temperatures in Wuhan.
Table 2. Monthly meteorological conditions and visibility in Wuhan during March 1, 2013, to February 28, 2014.
In warm months (from May to August), the high temperature and solar radiation could promote the photochemistry activities, which can increase the concentrations of O3 (Chou et al., 2011; Tang et al., 2006). The average temperature in these four months was 27.85°C, while that in cold months (from November to February) was just 7.43°C. A good positive correlation coefficient (R = 0.60) between O3 and temperature was found in Figure 3b.
Table 3 summarizes monthly PM2.5/PM10, PM2.5/NO2, PM2.5/SO2, PM2.5/CO, and PM2.5/O3 mean ratios in Wuhan. The values of these ratios were higher during December 2013 to February 2014 and lower during July 2013 to August 2013, which was consistent with the monthly variations of PM2.5. The highest PM2.5/PM10 value (0.90) in January 2014 indicates that PM2.5 contributed most, while the coarse particles were only minor in PM10. Wei et al. (1999) reported short-term measurements of PM10 and PM2.5 in Wuhan and found that PM2.5 accounted for about 60% of the mass of PM10. For the whole year in this study, PM2.5 contributed to 66.4% of PM10 and coarse particles only accounted for 33.6% of the PM10 in Wuhan. The higher ratios of PM2.5/NO2 (1.93–2.30), PM2.5/SO2 (2.82–4.39), and PM2.5/O3 (2.67–7.68) in cold months (December 2013 to February 2014) revealed the importance of photochemical formation of secondary aerosol production. Being a long-lived tracer of human activity, the concentration of CO is associated with sources such as combustion, industry, mobile, and oxidation of hydrocarbons (Yu et al., 2006; Chin et al., 1994). Thus, the PM2.5/CO ratio is usually used to evaluate the contribution of primary combustion emissions (Zhang and Cao, 2015). The variations of PM2.5/CO ratios in Table 3 indicated that primary emissions contributed PM2.5 production to a great extent in warm months while secondary particle formation was also very important in cold months.
Table 3. Monthly PM2.5/PM10, PM2.5/NO2, PM2.5/SO2, PM2.5/CO, and PM2.5/O3 mean ratios in Wuhan during the period of March 1, 2013, to February 28, 2014.
Seasonal variations
Figure 4 shows the mean results of the four seasons: spring (March, April, and May), summer (June, July, and August), autumn (September, October, and November), and winter (December, January, and February). As shown in Figure 4, the concentrations of PM2.5, PM10, SO2, NO2, and CO were higher in winter but lower in summer, while the O3concentrations were higher in summer but lower in winter. The severe air pollution during wintertime was associated with the high anthropogenic emissions. Furthermore, the atmosphere in Wuhan during winter season is characterized by low wind speed, low relative humidity, and low solar heating of land, which result in less dispersion of air pollutants. The more frequent occurrences of stagnant weather and intensive temperature inversion during the colder months in winter can intensify the accumulation of pollutants near the ground and lead to high PM episodes (Gong et al., 2015; He et al., 2001; Qu et al., 2010; Xia et al., 2006). In contrast, high temperatures and intense convection in summer may be more favorable for the pollutant diffusion and dilution. Abundant precipitation due to the summer monsoon, which increases wet scavenging, could also reduce the frequency of air pollution in summer.
Published online:
01 June 2017Figure 4. Box plots of seasonal variations of air pollutants in Wuhan for (a) PM2.5, (b) PM10, (c) NO2, (d) SO2, (e) O3, and (f) CO for different seasons from March 1, 2013, to February 28, 2014. The box plots depict the minimum, the 25th, 50th (median), and 75th percentiles, and the maximum for the air pollutant concentrations; the square in the box depicts the arithmetic mean concentrations.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2017/uawm20.v067.i07/10962247.2016.1240724/20170601/images/medium/uawm_a_1240724_f0004_b.gif"})
Figure 4. Box plots of seasonal variations of air pollutants in Wuhan for (a) PM2.5, (b) PM10, (c) NO2, (d) SO2, (e) O3, and (f) CO for different seasons from March 1, 2013, to February 28, 2014. The box plots depict the minimum, the 25th, 50th (median), and 75th percentiles, and the maximum for the air pollutant concentrations; the square in the box depicts the arithmetic mean concentrations.
Diurnal variations
Figure 5 illustrates the hourly variations of ambient pollutant concentrations in Wuhan. The hourly variations of air pollutants were mainly affected by the emissions and temperature/solar radiation. The maximum PM2.5 levels occurred in the morning hours (9 a.m. to 12 noon) and was accompanied by the increased NO2, SO2, and CO concentrations, indicating the significant impact of combustion of fossil fuel and car emissions (Bossioli et al., 2009; Li et al., 2008). This peak was followed by a gradual decrease through the afternoon. This might be caused by the stronger solar heating and deeper planetary boundary layer (PBL) associated with lesser traffic density during the afternoon (Yadav et al., 2014). The reason for the evening peak may be attributed to the vehicle emissions and stagnant atmospheric conditions (Badarinath et al., 2009). Similar diurnal variations were also found for PM10 concentrations, but with more pronounced variations. However, O3 concentration variations were different with the highest concentration from 2 p.m. to 5 p.m. in the afternoon, mainly caused by stronger photochemical effects (Chen et al., 2015; Zhou et al., 2015). The decrease of O3 concentrations during nighttime might be caused by chemical titration of NOx released by emissions from the urban activities (such as mobiles and heating) (Ran et al., 2009; Yu et al., 2006; Chin et al., 1994; Shan et al., 2009).
Published online:
01 June 2017Figure 5. Diurnal variations of air pollutant mean concentrations in Wuhan for different seasons from March 1, 2013, to February 28, 2014.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2017/uawm20.v067.i07/10962247.2016.1240724/20170601/images/medium/uawm_a_1240724_f0005_b.gif"})
Figure 5. Diurnal variations of air pollutant mean concentrations in Wuhan for different seasons from March 1, 2013, to February 28, 2014.
A case study
The temporal variation characteristics analyses revealed that air pollution was the most serious in winter, especially for PM2.5. The results showed that enhanced local emissions and stagnated weather were the important factors contributing to the pollution in Wuhan. In order to indentify the regional transport impact on the haze formation in Wuhan, a case study was carried for the period of January 21 to February 7, 2014. Figure 6 shows the hourly concentrations variations of PM2.5, PM10, SO2, CO, NO2, and O3 at the nine monitoring stations in Wuhan for the study period. As shown in Figure 6, the O3 concentrations all below 150 μg m−3 were lower than the hourly CNAQSS of 200 μg m−3. Meanwhile, the concentrations of SO2, NO2, and CO during this period ranged from 1 to 130 μg m−3, from 2 to 198 μg m−3, and from 0.004 to 4.7 mg m−3, respectively, which were also lower than the corresponding hourly CNAQSS (500 μg m−3 for SO2, 200 μg m−3 for NO2, and 10.0 mg m−3 for CO). Since biomass, coal combustion, steel manufacture, smelting, and vehicle emissions are the major sources of SO2, NO2, and CO, their concentrations can be affected more significantly by the local sources. Figure 6 shows that PM2.5 concentrations were higher than 75 μg m−3 most of the time from January 25 to February 3, 2014, with some days higher than 200 μg m−3. Following Yu et al. (2014b), the whole data were separated into three cases on the basis of PM2.5 concentrations: PM2.5 concentrations less than 75 μg m−3 (relatively clean air period), PM2.5 concentrations greater than 75 μg m−3 but less than 200 μg m−3 (haze period), and PM2.5 concentrations greater than 200 μg m−3 (heavy haze period). We found that during the relatively clean air, haze, and heavy haze periods, the average concentrations of PM2.5 were 42.1 ± 15.7 μg m−3, 124.5 ± 35.3 μg m−3, and 268.0 ± 45.4 μg m−3, respectively.
Published online:
01 June 2017Figure 6. Time series of hourly concentrations of (a) PM2.5, (b) PM10, (c) CO, (d) NO2, (e) SO2, (f) O3 in Wuhan, (g) PM2.5 in Baoding, and (h) PM2.5 in Handan from January 21 to February 7, 2014.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2017/uawm20.v067.i07/10962247.2016.1240724/20170601/images/medium/uawm_a_1240724_f0006_b.gif"})
Figure 6. Time series of hourly concentrations of (a) PM2.5, (b) PM10, (c) CO, (d) NO2, (e) SO2, (f) O3 in Wuhan, (g) PM2.5 in Baoding, and (h) PM2.5 in Handan from January 21 to February 7, 2014.
Trajectory cluster analysis
Trajectory clustering was used in this study to identify the relationship between atmospheric transport patterns and hourly PM2.5 levels in Wuhan from January 21 to February 7, 2014. The results of trajectory clustering indicated that different periods have different characteristics of air pollution sources. Figure 7 shows back-trajectories for different periods as well as clusters as a function of the vertical pressure profiles. In Figure 7a, four clusters for all data during the whole period were determined by the cluster analysis algorithm: S (South), N (North), NW (Northwest), and NE (Northeast). Figure 7b indicates that the 48-hr back-trajectories for the relatively clean air period were coming from the far-away regions like Mongolia, Inner Mongolia, and China Yellow Sea. As shown in Figure 7c and Figure 7d, for the haze and heavy haze periods, most of the 48-hr back-trajectories were coming from the nearby provinces like Shanxi, Jiangxi, Anhui, Henan, Hebei, and Hunan.
Published online:
01 June 2017Figure 7. Cluster analysis of the 48hr air mass back trajectories starting at 100 m in Wuhan from January 21 to February 7, 2014. (a) All back-trajectories for the whole period. The four transport pathways (clusters) are determined: S (South), N (North), NE (Northeast), and NW (Northwest). (b) All back-trajectories for the relatively clean cases (PM2.5 < 75 μg m−3), (c) All back-trajectories for the haze case (PM2.5 ≥ 75 μg m−3 but PM2.5 < 200 μg m−3). (d) All back-trajectories for the heavy haze case (PM2.5 ≥ 200 μg m−3). (e) The pressure profile of four clusters.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2017/uawm20.v067.i07/10962247.2016.1240724/20170601/images/medium/uawm_a_1240724_f0007_oc.jpg"})
Figure 7. Cluster analysis of the 48hr air mass back trajectories starting at 100 m in Wuhan from January 21 to February 7, 2014. (a) All back-trajectories for the whole period. The four transport pathways (clusters) are determined: S (South), N (North), NE (Northeast), and NW (Northwest). (b) All back-trajectories for the relatively clean cases (PM2.5 < 75 μg m−3), (c) All back-trajectories for the haze case (PM2.5 ≥ 75 μg m−3 but PM2.5 < 200 μg m−3). (d) All back-trajectories for the heavy haze case (PM2.5 ≥ 200 μg m−3). (e) The pressure profile of four clusters.
Table 4 lists the percentages of trajectories for each trajectory cluster, as well as corresponding mean concentrations of PM2.5 and mean wind speeds for four cases. Note that the mean wind speeds were calculated on the basis of meteorological fields from NCEP/FNL obtained from NOAA along the back trajectories. The results for the polluted cases (PM2.5 concentrations greater than 75 μg m−3) are also summarized in Table 4. The trajectory clusters for the whole period were dominated by N (35.8%) and NE (26.5%) clusters, with some contributions from NW (18.1%) and S (19.6%) clusters. For the relatively clean air period, the predominant cluster was NE (77.2%), followed by NW (14.1%) and N (8.6%). Mean wind speeds of NE, NW, and N clusters for this clean period were 6.7, 10.8, and 3.6 m/s, respectively. For the haze period, the predominant clusters were S (35.6%) and N (34.4%). Mean wind speeds of S, N, NW, and NE clusters were 4.8, 5.2, 6.4, and 3.3 m/sec, respectively. For the heavy haze period, clusters N mainly accounted for 57.8%, followed by NW (20.9%) and S (19.5%), while clusters from NE (1.8%) only contributed a very small part. Note that back trajectories from north and south have the relatively slow mean wind speeds with 2.5 and 3 m/sec for this period, respectively. The results indicated that the pollution sources in the north and south areas of Wuhan were mainly responsible for the heavy haze formation in Wuhan. Figure 7e shows that the long transport pathway (NW cluster) originating from low barometric altitude (817.3 hPa) with the height of 1764 m had the fast speed, while the dominant NE cluster that brought clean air masses came from relatively low altitude (689 m). The clusters S and N originated with the heights of 792 and 725 m, respectively.
Table 4. Mean concentrations of PM2.5 and percentages of trajectories for each trajectory cluster for the whole period (January 21 to February 7, 2014) and three sampling periods (relatively clean air: PM2.5 < 75 μg m−3, haze: PM2.5 ≥ 75 μg m−3, but PM2.5 < 200 μg m−3, heavy haze: PM2.5 ≥ 200 μg m−3).
Source contributions from PSCF and CWT analyses
The result of the PSCF values of PM2.5 in Wuhan for the whole period is shown in Figure 8a. We can infer that the potential source areas that most likely led to high PM2.5 concentrations in Wuhan were located primarily in Handan and Baoding in Hebei province, Zhumadian in Henan province, and Changsha and Jinggangshan in Hunan province. According to the PSCF analyses, it is worth noting that in addition to the local sources, high PSCF values were located mainly in the north and south of Wuhan.
Published online:
01 June 2017Figure 8. (a) The PSCF map for PM2.5 for the whole period, and the CWT maps for PM2.5 for (b) the whole period, (c) the haze period, and (d) the heavy haze period. The unit of CWT is μg m−3.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2017/uawm20.v067.i07/10962247.2016.1240724/20170601/images/medium/uawm_a_1240724_f0008_oc.jpg"})
Figure 8. (a) The PSCF map for PM2.5 for the whole period, and the CWT maps for PM2.5 for (b) the whole period, (c) the haze period, and (d) the heavy haze period. The unit of CWT is μg m−3.
Figure 8b–8d show the results of the CWT values of PM2.5 for the whole, haze, and heavy haze periods. Figure 8b indicates that for the whole period, the potential polluted source areas were mainly located in Handan and Baoding in Hebei province and Changsha in Hunan province. For the haze period, the sources affecting high PM2.5 concentrations in Wuhan were mainly located in Bozhou in Anhui province and Changsha in Hunan province. Distributions of the CWT values of PM2.5 in Figure 8d for the heavy haze period reveal that the potential source areas were mainly located north of Wuhan, such as Baoding and Handan in Hebei province, Xiaoyi in Shanxi province, Taian in Shandong province, and Bozhou in Anhui province.
The heavy haze observations of 1 or 2 days earlier in the upwind cities (2 days earlier for Baoding city and 1 day earlier for Handan city) than those in Wuhan, as shown in Figures 6g–6i, confirm the analysis that the air pollution transported from Baoding and Handan in Hebei province made a significant contribution to the haze formation in Wuhan.
AOD data analysis
Figure 9 shows the spatial distributions of the average AOD data at 550 nm over the eastern China for January 21 to 23, January 24, January 25 to 29, January 30 to February 3, and February 4 to 7, and for January 21 to February 7, 2014. The movements of haze (red colored areas) over China for the whole studying period can be seen from Figure 9. Figure 9f reveals that AOD values for the whole period in Wuhan was around 0.9, indicating the severe air pollution during this period. Figure 9a shows that most of the heavy haze pollutions were formed and located in the northern and western part of Wuhan during the period from January 21 to 23. Figure 9b reveals that heavy haze pollution were located in the northern and southern of Wuhan before the heavy haze outbreak in Wuhan on January 24. As shown in Figure 9c, northern, southern and eastern of China have suffered the heavy haze pollution, and the AOD values was around 1.0 in Wuhan from January 25 to 29. From January 30 to February 3, although Wuhan still suffered severe haze, the degree of pollution had relatively alleviated, and high AOD values were located in north of Wuhan as shown in Figure 9d. February 4 to 7, 2014, was the clean period in Wuhan, but due to relatively small number of effective satellite daily data, we cannot clearly see the AOD distributions in Figure 9e.
Published online:
01 June 2017Figure 9. Observations of AOD values at 550 nm from the MODIS (a) for the period of January 21 to 23, 2014, (b) on January 24, 2014, (c) from January 25 to January 29, 2014, (d) from January 30 to February 3, 2014, (e) from February 4 to February 7, 2014, and (f) from January 21 to February 7, 2014.
![]({"type":"image","src":"/na101/home/literatum/publisher/tandf/journals/content/uawm20/2017/uawm20.v067.i07/10962247.2016.1240724/20170601/images/medium/uawm_a_1240724_f0009_oc.jpg"})
Figure 9. Observations of AOD values at 550 nm from the MODIS (a) for the period of January 21 to 23, 2014, (b) on January 24, 2014, (c) from January 25 to January 29, 2014, (d) from January 30 to February 3, 2014, (e) from February 4 to February 7, 2014, and (f) from January 21 to February 7, 2014.
Conclusion
As one of the highest industrial developmental areas in China, Wuhan, the capital of Hubei province, has inevitably experienced severe haze in recent years. To identify the characteristics of air quality and factors attributed to the formation of haze in Wuhan, we analyzed the air pollutants (PM2.5, PM10, NO2, SO2, O3, and CO) observations from March 1, 2013, to February 28, 2014, at the nine monitoring stations. The temporal variations (annual, monthly, seasonal, and diurnal) study indicates that meteorological conditions and anthropogenic activities were responsible for the annual/monthly/seasonal/diurnal variations of air pollutants. The peak values of PM2.5, PM10, NO2, CO, and SO2 occurred in December because of increased local emissions, low temperature, low wind speed, and high pressure, while the lowest values occurred in July, mainly caused by scavenging via heavy precipitation. For O3, the high temperature and solar radiation could promote the photochemistry activity, which increased the concentrations of O3 from May to August. It is found that there were two ascending stages for hourly PM2.5, PM10, NO2, and CO concentrations caused by the two traffic peaks. However, the O3 concentration variations were different and the highest concentrations in the afternoon were mainly caused by strong photochemical effects.
The case study results revealed that the potential sources responsible for the haze formation in Wuhan primarily originated from the north and south of Wuhan, such as Baoding and Handan in Hebei province, Zhumadian in Henan province, Bozhou in Anhui province, Xiaoyi in Shanxi province, Taian in Shandong province, and Changsha in Hunan province.
The comprehensive analysis indicated that emissions, photochemical formation, and meteorology conditions have significant impact on air quality. In addition, the surrounding province emissions also exert a profound impact on the haze formation of Wuhan. Therefore, it is essential to implement air pollution control not only at a local level, but also for all surrounding areas, especially for the regions located in the north and south of Wuhan. Further studies will be needed to verify the modeling results and to develop environmental management strategies.
| PM2.5 | PM10 | NO2 | SO2 | O3 | CO | |
|---|---|---|---|---|---|---|
| Average | 89.6 | 134.9 | 54.9 | 32.4 | 62.3 | 1.1 |
| Second National Standard | 35* | 70* | 40* | 60* | 200** | 4.0*** |
Note. The unit for CO was mg m−3, and for other pollutants was μg m−3; mg m−3 = M/22.4 × ppmV × {273/(273 + T)} × (Ba/101325). M refers to the gas relative molecular weight, T refers to temperature, and Ba refers to pressure.
*Annual average concentration standard.
**Hourly average concentration standard.
***24-hr average concentration standard.
| Month | Temperature, ºC | Pressure, hpa | Wind speed, km/hr | Relative humidity, % | Visibility, km |
|---|---|---|---|---|---|
| March | 13.3 | 1017.9 | 10.7 | 64.5 | 7.1 |
| April | 17.3 | 1013.9 | 9.9 | 58.8 | 9.4 |
| May | 22.6 | 1008.5 | 9.2 | 68.9 | 8.0 |
| June | 26.5 | 1005.0 | 9.5 | 70.1 | 9.0 |
| July | 31.0 | 1002.6 | 13.4 | 63.3 | 13.3 |
| August | 31.3 | 1004.1 | 11.9 | 59.6 | 13.3 |
| September | 23.4 | 1013.3 | 7.0 | 71.3 | 6.2 |
| October | 18.9 | 1019.9 | 6.3 | 59.0 | 4.7 |
| November | 12.3 | 1023.2 | 7.1 | 70.6 | 5.6 |
| December | 5.7 | 1026.3 | 6.6 | 61.3 | 3.8 |
| January | 6.6 | 1025.0 | 8.6 | 62.3 | 3.9 |
| February | 5.1 | 1023.8 | 9.7 | 78.4 | 4.5 |
| Month | PM2.5/PM10 | PM2.5/NO2 | PM2.5/SO2 | PM2.5/CO | PM2.5/O3 |
|---|---|---|---|---|---|
| March | 0.63 | 1.68 | 2.46 | 0.08 | 0.98 |
| April | 0.55 | 1.38 | 2.20 | 0.07 | 0.93 |
| May | 0.62 | 1.40 | 2.50 | 0.07 | 0.80 |
| June | 0.60 | 1.22 | 2.35 | 0.05 | 0.58 |
| July | 0.48 | 1.03 | 1.74 | 0.05 | 0.40 |
| August | 0.46 | 1.16 | 1.87 | 0.05 | 0.44 |
| September | 0.58 | 1.29 | 2.47 | 0.06 | 0.85 |
| October | 0.65 | 1.64 | 3.33 | 0.09 | 1.65 |
| November | 0.59 | 1.44 | 2.66 | 0.08 | 3.23 |
| December | 0.79 | 1.93 | 2.82 | 0.10 | 7.68 |
| January | 0.90 | 2.30 | 3.37 | 0.10 | 7.12 |
| February | 0.71 | 2.04 | 4.39 | 0.08 | 2.67 |
| Percent (%) | Mean PM2.5 (μg m−3) | Polluted percent (%) | Polluted mean PM2.5 (μg m−3) | WS (m s−1) | ||
|---|---|---|---|---|---|---|
| All | N | 35.8 (440) | 205.5 | 46.5 (410) | 216.7 | 3.3 |
| NE | 26.5 (326) | 54.5 | 6.6 (58) | 128.5 | 6.3 | |
| NW | 18.1 (222) | 177.4 | 19.6 (173) | 214.5 | 7.6 | |
| S | 19.6 (241) | 173.3 | 27.3 (241) | 173.3 | 3.2 | |
| Clean | N | 8.6 (30) | 52.0 | 0 (0) | 0 | 3.6 |
| NE | 77.2 (268) | 38.6 | 0 (0) | 0 | 6.7 | |
| NW | 14.1 (49) | 46.41 | 0 (0) | 0 | 10.8 | |
| Haze | N | 34.4 (147) | 119.1 | 34.4 (147) | 119.1 | 4.8 |
| NE | 11.7 (50) | 106.1 | 11.7 (50) | 106.1 | 5.2 | |
| NW | 18.3 (78) | 126.2 | 18.3 (78) | 126.2 | 6.4 | |
| S | 35.6 (152) | 132.4 | 35.6 (152) | 132.4 | 3.3 | |
| Heavy Haze | N | 57.8 (263) | 271.2 | 57.8 (263) | 271.2 | 2.5 |
| NE | 1.8 (8) | 268.1 | 1.8 (8) | 268.1 | 6.9 | |
| NW | 20.9 (95) | 286.9 | 20.9 (95) | 286.9 | 6.9 | |
| S | 19.5 (89) | 243.3 | 19.5 (89) | 243.3 | 3.0 | |
Note. Values in the parentheses are number of back-trajectories.
Funding
Part of this work is supported by the “Zhejiang 1000 Talent Plan” and Research Center for Air Pollution and Health in Zhejiang University. Part of this work is also partially supported by the National Natural Science Foundation of China (number 21577126) and Department of Science and Technology of China (number 2014BAC22B06). This work was supported by the Joint NSFC-ISF Research Program (No. 41561144004), jointly funded by the National Natural Science Foundation of China and the Israel Science Foundation.