Variation Trends of Fine Particulate Matter Concentration in Wuhan City from 2013 to 2017

Fine particulate matter (PM2.5) is directly associated with smog and has become the primary factor that threatens air quality in China. In order to investigate the variation patterns of PM2.5 concentrations in various regions of Wuhan city across different time spans, we analyzed continuous monitoring data from six monitoring sites in Wuhan city from 2013 to 2017. The results showed that the PM2.5 concentration from the various monitoring sites in the five-year period showed a decreasing trend. January, October, and December are the three months with relatively high mean monthly PM2.5 concentrations in the year, while June, July, and August are the three months with relatively low mean monthly PM2.5 concentrations in the year. The number of days with a daily mean concentration of 35–75 μg/m3 was the highest, while the number of days with a daily mean concentration of more than 250 μg/m3 was the lowest. PM2.5 accounted for a large proportion of the major pollutants and is the main source of air pollution in Wuhan city, with an average proportion of over 46%.


Introduction
With rapid industrialization and urbanization, the problem of air pollution, particularly the problem of smog, has become increasing severe. PM 2.5 is the particulate matter with an aerodynamic diameter less than or equal to 2.5 µm, which is directly associated with smog and has become the primary factor that threatens air quality in China [1]. PM 2.5 pollution in China has a relatively clear spatial and temporal distribution, with severe pollution mainly occurring during the transition from autumn to winter and winter to spring [2]. PM 2.5 contains several harmful substances, and the particles of PM 2.5 enter the lungs through the respiratory system and threaten human health, particularly that of children and the elderly. In recent years, China has experienced multiple outbreaks of respiratory diseases, which have resulted in school closures and factory shutdowns, thus causing social panic and economic losses. At the same time, the problem of smog is showing a trend of spreading from north regions to south regions in China. The air quality of Wuhan has been deteriorating, and PM 2.5 has become the city's primary air pollutant [2].
In recent years, Chinese researchers have conducted PM 2.5 studies that have been focused on the spatial distribution, composition, source, and transmission laws of PM 2.5 [3][4][5][6][7][8]. Jiao et al. employed mathematical statistics and geographic information system (GIS) spatial analysis to analyze the temporal variation and spatial distribution characteristics of PM 2.5 . Their results showed that spatial quantization of transboundary PM 2.5 and the tracking of transaction paths, identified the relationship between the consumption and production responsibilities of China, Japan, and Korea, and reviewed existing environmental cooperation mechanisms and policies. The results showed that Japan and South Korea are strongly dependent on China economically and various countries should actively participate in the governance of environmental problems such that all the participating countries benefit [32]. Aldabe et al. collected the data of PM 2.5 and PM 10 from three different areas (rural, urban, and urban-traffic) in Navarra, Spain. They performed PMF model to identify five principle sources for PM 10 and PM 2.5 in Iturrama and Plaza de la Cruz, and 4 sources for PM 10 in Bertiz [33]. Siskos et al. employed a Harvard impactor system to simultaneously collect atmospheric aerosols and PM 2.5 data in Greece. Ion chromatography and a semi-micro electrodes were used to measure the chemical composition and aerosol acidity in samples, respectively. The analysis results showed that the majority of aerosol ions near the sea were acidic, but that their overall concentration was low. Aerosol ions in city centers were neutral but their overall concentrations were high. At the same time, ion concentrations were higher in spring and winter throughout the year [34]. Saliba et al. conducted a long-term assessment of PM 2.5 and PM 10 concentration in Eastern Mediterranean region. The results showed that the semi-enclosed structure and vehicle emissions lead to higher PM 2.5 and PM 10 concentrations in the region [35]. Singh et al. employed an Andersen cascade impactor sampler to collect atmospheric organic aerosols from Delhi during January 2006 to December 2007 in order to study the particle distribution in aerosols and their seasonal characteristics. The obtained results showed that the burning of biofuels and biomass as well as other human activities significantly increase PM 2.5 concentrations. Autumn and winter are the two seasons in which PM 2.5 concentrations are higher in the year. The over-limit status of PM 2.5 concentrations throughout the year is an issue [36]. Rao et al. performed a bootstrap simulation analysis to estimate the uncertainty of the emission of wood consumption and emission factors of different appliance types [37]. Pipal et al. collected PM 2.5 and PM 10 samples from various monitoring sites in Agra in order to study the yearly variation trends of mineral aerosols. The results of the study showed that there are diverse sources of local PM 2.5 , of which crustal activity and local traffic and transportation are the primary sources [38]. Foreign studies on PM 2.5 tend to focus on composition, source, and economics, and there are few studies on the long-term variation patterns of PM 2.5 . PM 2.5 governance has become a social consensus, but the prerequisite for governance requires an understanding of PM 2.5 , and its causes, composition, and temporal and spatial distribution patterns require research. The majority of PM 2.5 studies in Wuhan have a one-year timespan, and there is a lack of studies on long-term concentration variation trends, which is necessary for PM 2.5 governance. In this study, we analyzed the PM 2.5 data from various regions in Wuhan city in the period of 2013-2017 and its concentration variation trends for different timescales. This will provide a reference for PM 2.5 governance for different periods in various regions in Wuhan city.

Sample Collection, Analysis and Calculation
Currently, Wuhan City has a total of 32 automatic ambient air quality monitoring sites, including 10 state-controlled monitoring sites, 11 city-controlled monitoring sites, 4 region-controlled monitoring sites, 1 combined atmospheric pollution automatic monitoring laboratory, 1 roadside site, 1 control site, and 4 regional sites. The state-controlled monitoring sites (Hanyang Yuehu, Hankou Huaqiao, Hankou Jiangtan, Wuchang Ziyang, Donghu Liyuan, and Qingshan Ganghua) were used for the analysis. The data in this study was obtained from the Wuhan Environmental Protection Bureau. Figure 1 shows the distribution of the monitoring sites. Sample collection and analysis were performed according to standards [39]. Table 1 shows the device range and precision information. Glass fiber filters were used because their retention efficiency with 0.3 µm-standard particles is higher than 99%. The sensitivity of the analytical balance was 0.01 mg. The thermostatic incubator temperature was set to 15-30 • C, and the temperature precision was ±1 • C. The relative humidity was adjusted to 50 ± 5%.  The sampler entrance is 1.5 m above the ground, and the sampling points are kept away from direct pollution sources and obstacles. An intermittent sampling at intervals of 5 h was used to measure the daily mean concentration. After sampling, the filters were weighed according to standards. Samples that could not be weighed immediately were labeled and stored at 4 °C. The PM2.5 concentration was calculated using Equation (1). During the data processing, if more than two data points were missing in a day, the day's data was considered to be invalid. If data from more than 6 days were missing in a month, that month's data was considered to be invalid. If the data for more than 3 months were missing, that year's data was considered to be invalid [2,16,40].

= −
(1) In the equation, ρ is the PM2.5 concentration in mg/m 3 (converted to μg/m 3 for statistical analysis); w1 = weight of blank filter in g; w2 = weight of filter after sampling in g; V = sample volume after conversion to standard conditions (101.325 kPa and 273 K) in m 3 .  The sampler entrance is 1.5 m above the ground, and the sampling points are kept away from direct pollution sources and obstacles. An intermittent sampling at intervals of 5 h was used to measure the daily mean concentration. After sampling, the filters were weighed according to standards. Samples that could not be weighed immediately were labeled and stored at 4 • C. The PM 2.5 concentration was calculated using Equation (1). During the data processing, if more than two data points were missing in a day, the day's data was considered to be invalid. If data from more than 6 days were missing in a month, that month's data was considered to be invalid. If the data for more than 3 months were missing, that year's data was considered to be invalid [2,16,40].
In the equation, ρ is the PM 2.5 concentration in mg/m 3 (converted to µg/m 3 for statistical analysis); w 1 = weight of blank filter in g; w 2 = weight of filter after sampling in g; V = sample volume after conversion to standard conditions (101.325 kPa and 273 K) in m 3 . Table 2 shows the valid hours and validity rate for PM 2.5 concentration monitoring data for the period of 2013-2017. As seen from the table, the minimum value for the validity rate for the monitoring data at all the sites in the 5-year period was 93.4%, and the maximum value was 99.7%. The data reliability was high and could be used in this paper. Figure 2 shows the annual mean concentration, standard deviation, and linear trendlines for PM 2.    From the overall variation trends, it can be observed that the PM2.5 concentration peaks showed a yearly declining trend, with the number of days of mild pollution gradually decreasing, while the number of days of severe and extreme pollution did not decrease, and the number of days with good air quality showed a slight increase. In the entire year, the PM2.5 concentration was between 35 μg/m 3 and 150 μg/m 3 . This indicates that the PM2.5 concentration on most days was at a fair or moderate pollution level.  Overall, the daily mean PM 2.5 concentration of every monitoring site shifts with a natural day and shows a saddle-like distribution.

Annual Variation Trend of PM 2.5 Concentrations
From the overall variation trends, it can be observed that the PM 2.5 concentration peaks showed a yearly declining trend, with the number of days of mild pollution gradually decreasing, while the number of days of severe and extreme pollution did not decrease, and the number of days with good air quality showed a slight increase. In the entire year, the PM 2.5 concentration was between 35 µg/m 3 and 150 µg/m 3 . This indicates that the PM 2.5 concentration on most days was at a fair or moderate pollution level. A natural day was used as the reference subject in order to analyze the PM2.5 concentration variation trends in the year. The "good air quality" points showed a scattered distribution in natural A natural day was used as the reference subject in order to analyze the PM 2.5 concentration variation trends in the year. The "good air quality" points showed a scattered distribution in natural days throughout the year, were relatively concentrated at 175-300 days, and were less distributed at other natural day intervals. The "fair air quality" points showed a scattered distribution in natural days throughout the year, were relatively concentrated at 100-330 days, and were less distributed at 0-25 days and 325-365 days. The "mild air pollution" points showed a relatively uniform distribution in the natural day throughout the year, and were seldom at 175-225 days. The "moderate air pollution" points were concentrated at 0-175 days and 250-365 days, and seldom appeared in natural days, as can be seen in the middle of the graph. The "severe air pollution" points were concentrated at 0-80 days and 275-365 days. The PM 2.5 concentration and degree of pollution showed clear patterns in natural day distribution throughout the year. This provides guidance for the governance of PM 2.5 at various timings. Figure 4 shows the variation trends of the mean monthly PM 2.5 concentrations at various monitoring sites in 2013-2017. This result is consistent with the analysis results at Section 3.1. Using the mean monthly concentration as a reference standard, the PM 2.5 concentrations showed a clear saddle-like distribution over the entire year. The binomial fitting standard formula y = ax 2 + bx + c (where y is the dependent variable, x is the independent variable, and a, b and c are the constants determined by the specific data) was chosen to fit the data. Various fitting results showed that the mean monthly PM 2.5 concentration showed a second-order function with an opening facing up. The symmetrical axis of the second-order function was concentrated in June, July, and August. days throughout the year, were relatively concentrated at 175-300 days, and were less distributed at other natural day intervals. The "fair air quality" points showed a scattered distribution in natural days throughout the year, were relatively concentrated at 100-330 days, and were less distributed at 0-25 days and 325-365 days. The "mild air pollution" points showed a relatively uniform distribution in the natural day throughout the year, and were seldom at 175-225 days. The "moderate air pollution" points were concentrated at 0-175 days and 250-365 days, and seldom appeared in natural days, as can be seen in the middle of the graph. The "severe air pollution" points were concentrated at 0-80 days and 275-365 days. The PM2.5 concentration and degree of pollution showed clear patterns in natural day distribution throughout the year. This provides guidance for the governance of PM2.5 at various timings. Figure 4 shows the variation trends of the mean monthly PM2.5 concentrations at various monitoring sites in 2013-2017. This result is consistent with the analysis results at Section 2.1. Using the mean monthly concentration as a reference standard, the PM2.5 concentrations showed a clear saddle-like distribution over the entire year. The binomial fitting standard formula y = ax 2 + bx + c (where y is the dependent variable, x is the independent variable, and a, b and c are the constants determined by the specific data) was chosen to fit the data. Various fitting results showed that the mean monthly PM2.5 concentration showed a second-order function with an opening facing up. The symmetrical axis of the second-order function was concentrated in June, July, and August.     Table 3 shows the months in which the highest and lowest values of the monthly mean PM2.5 concentrations appeared from 2013 to 2017. It can be observed that the months with the highest monthly mean PM2.5 concentration were January (23 times) and December (7 times). The months with the lowest monthly mean PM2.5 concentration were July (23 times) and August (7 times). As compared with the other years, the month with the highest mean monthly PM2.5 concentration shifted from January to December. On combining the results of Figures 4 and 5, it can be observed that January, October, and December are the three months with a relatively high monthly mean PM2.5 concentration in the entire year and the difference between the monthly mean PM2.5 concentration of October and the highest monthly mean concentration is generally less than 30 μg/m 3 . June, July, and August are the three months with a relatively low monthly mean PM2.5 concentration in the entire year, and the difference between the monthly mean PM2.5 concentration of June and the lowest monthly mean concentration is less than 15 μg/m 3 .   Table 3 shows the months in which the highest and lowest values of the monthly mean PM 2.5 concentrations appeared from 2013 to 2017. It can be observed that the months with the highest monthly mean PM 2.5 concentration were January (23 times) and December (7 times). The months with the lowest monthly mean PM 2.5 concentration were July (23 times) and August (7 times). As compared with the other years, the month with the highest mean monthly PM 2.5 concentration shifted from January to December. On combining the results of Figures 4 and 5, it can be observed that January, October, and December are the three months with a relatively high monthly mean PM 2.5 concentration in the entire year and the difference between the monthly mean PM 2.5 concentration of October and the highest monthly mean concentration is generally less than 30 µg/m 3 . June, July, and August are the three months with a relatively low monthly mean PM 2.5 concentration in the entire year, and the difference between the monthly mean PM 2.5 concentration of June and the lowest monthly mean concentration is less than 15 µg/m 3 . Table 3. Months with highest and lowest mean monthly PM 2.5 concentration at various stations. Figure 6 shows the number of days with various PM 2.5 evaluation intervals at the various monitoring sites in the five-year period. Overall, the number of days with the PM 2.5 concentration interval of 35-75 µg/m 3 was the highest, followed by the number of days with the concentration interval of 75-115 µg/m 3 . The number of days with the concentration interval of ≥250 µg/m 3 was the least and the number of days with the concentration interval of 0-35 µg/m 3 was also relatively low. At the same time, after 2013, the number of days with the concentration intervals of 115-150 µg/m 3 , 150-115 µg/m 3 , and ≥250 µg/m 3 showed an annual decreasing trend. This data distribution shows that under the majority of circumstances, the daily mean PM 2.5 concentration was at a "fair" (35-75 µg/m 3 ) and "mild" air pollution (75-115 µg/m 3 ) status in this region. In addition, the PM 2.5 concentrations of these regions showed a decreasing trend. This is similar to the analysis results of the daily mean PM 2.5 concentrations in Section 3.1. However, the over-limit status of the PM 2.5 concentration was still serious as the number of days with a "good" (0-35 µg/m 3 ) and "fair" air quality (35-75 µg/m 3 ) accounted for 40% of the entire year, and the overall PM 2.5 concentration is still not ideal. Although this does not result in smog or affect traffic, long-term exposure to these PM 2.5 concentrations can threaten human health, particularly that of children and the elderly. Overall, PM 2.5 accounted for the greatest proportion of all the primary pollutants at all the stations, with an average proportion of more than 46%. The station with the lowest PM 2.5 concentration value at 40% was the Wuchang Ziyang monitoring site, and the one having the highest value of at 58% was the Qingshan Ganghua monitoring site. This shows that PM 2.5 is the primary air pollutant in Wuhan city, followed by O 3 as the secondary pollutant with an average proportion of more than 19%. NO 2 and PM 10 have almost identical contributions to air pollution, with average proportions of 9.67% and 10%, respectively. The other pollutants are diverse and have a collective average proportion of 14.83%, which also contributes to the air pollution.  at 40% was the Wuchang Ziyang monitoring site, and the one having the highest value of at 58% was the Qingshan Ganghua monitoring site. This shows that PM2.5 is the primary air pollutant in Wuhan city, followed by O3 as the secondary pollutant with an average proportion of more than 19%. NO2 and PM10 have almost identical contributions to air pollution, with average proportions of 9.67% and 10%, respectively. The other pollutants are diverse and have a collective average proportion of 14.83%, which also contributes to the air pollution.
(2) The various air quality points showed a scattered distribution on natural days throughout the year. The "good" air quality points were concentrated at 175-300 days and were less distributed in other natural day intervals. The "fair" air quality points were concentrated at 100-330 days and were less distributed at 0-25 days and 325-365 days. The "mild" air pollution points showed a relatively uniform distribution on natural days in the entire year and were few at 175-225 days. The "moderate" air pollution points were concentrated at 0-175 days and 250-365 days. The "severe" air pollution points were concentrated at 0-80 days and 275-365 days.
(3) The months with the highest monthly mean PM2.5 concentration were January and December. The months with the lowest monthly mean PM2.5 concentration were July and August (seven times). As compared with other years, the month with the highest mean monthly PM2.5 concentration in 2017 shifted from January to December. January, October, and December are the three months with a
(2) The various air quality points showed a scattered distribution on natural days throughout the year. The "good" air quality points were concentrated at 175-300 days and were less distributed in other natural day intervals. The "fair" air quality points were concentrated at 100-330 days and were less distributed at 0-25 days and 325-365 days. The "mild" air pollution points showed a relatively uniform distribution on natural days in the entire year and were few at 175-225 days. The "moderate" air pollution points were concentrated at 0-175 days and 250-365 days. The "severe" air pollution points were concentrated at 0-80 days and 275-365 days.
(3) The months with the highest monthly mean PM 2.5 concentration were January and December. The months with the lowest monthly mean PM 2.5 concentration were July and August (seven times). As compared with other years, the month with the highest mean monthly PM 2.5 concentration in 2017 shifted from January to December. January, October, and December are the three months with a relatively high monthly mean PM 2.5 concentration in the entire year, and the difference between the monthly mean PM 2.5 concentration in October and the highest monthly mean concentration is generally less than 30 µg/m 3 . June, July, and August are the three months with a relatively low monthly mean PM 2.5 concentration in the entire year, and the difference between the monthly mean PM 2.5 concentration in June and the lowest monthly mean concentration is less than 15 µg/m 3 .
(4) The number of days with a daily mean PM 2.5 concentration interval of 35-75 µg/m 3 was the highest while the number of days with a daily mean PM 2.5 concentration of ≥250 µg/m 3 was the least. After 2013, the number of days with daily mean PM 2.5 concentration intervals of 115-150 µg/m 3 , 150-115 µg/m 3 , and ≥250 µg/m 3 showed an annual declining trend. However, the over-limit status of the PM 2.5 concentration was still serious as the number of days with a "good" (0-35 µg/m 3 ) and "fair" air quality (35-75 µg/m 3 ) accounted for 40% of the entire year, and the overall PM 2.5 concentration is still not ideal.
(5) PM 2.5 accounted for a large proportion of major pollutants and is the main source of air pollution in Wuhan city with an average proportion of over 46%. The station with the lowest value of PM 2.5 concentration of 40% was the Wuchang Ziyang monitoring site, the station with the highest value of 58% was the Qingshan Ganghua monitoring site, and the mean value of which was over 46%.