1 Background

Due to rapid urbanization, large amount of domestic and industrialized wastages emitted into the atmosphere that causes air pollution. In recent years, air pollution such as smog occurs frequently and has become a global issue. It is very dangerous for atmosphere as well as for human health (Zhang et al. 2013; Chaichan et al. 2018). Air pollution has effects on human health such as nausea, skin irritation and difficulty in breathing (Barakat-Haddad et al. 2013; Künzli and Tager 2005). World Health Organization presented its estimates about air pollution disease and found that atmospheric pollution caused 3.7 million deaths worldwide in both rural and urban areas in 2012 (WHO 2014).

Air pollutants are categorized into the primary and secondary pollutants. Primary pollutants referred to substances which are emitted directly from sources such as nitrogen oxides (traffic), sulfur dioxides (coal combustion) and particulates matter (traffic, dust and combustion of coal etc.). On the other hand, secondary pollutants are generated through chemical and photochemical reactions in atmosphere (Huang et al. 2014). In the present study, the air pollution due to fine particulate matter PM2.5 is examined. A brief overview of PM2.5 is given in the following section.

1.1 Fine particulate matter (PM2.5)

Particulate matter is the suspended particles or droplets in the atmosphere usually stated as PM (Zhang and Wang 2011). From all atmospheric pollution particles, fine particulate matter PM2.5 is the main pollution source that has an aerodynamic diameter less than 2.5 µm. It has dangerous effects on health (Pope et al. 2002; Pope and Dockery 2006). It also has a long residence time in the atmosphere and can be transported for long distances. PM2.5 has a large influence on the atmospheric environment (Zhang and Wang 2011).

Asia remained under studies due to its population rate and pollution. China is a huge populated country and its industrialization and development growth has rapidly increased. These exert incredible pressure on the environment (Ji et al. 2014). Due to these reasons, China is facing air quality challenges. For energy sources, China mainly depends on coal and coal burning power plants that release large amount of air pollution. Industry and motor vehicle emission are the fastest source of outdoor pollution, especially in urban areas (Muller et al. 2015). In China, haze or smog is one of the biggest environmental worries (Chan and Yao 2008). A lot of researches on the characterization of PM2.5 pollution have been conducted in the past decades (Liu et al. 2008; Wang et al. 2014, 2013). The findings of Song et al. (2012) have exposed that the main source of smog problem is PM2.5 which has severe effects on human health, global climate and visibility. The investigation of Yu et al. (2011) suggested that main sources of pollution in Beijing are PM2.5 with percentage of secondary sulfur (26.5%), road dust (12.7%), fossil fuel combustion (16.0%), vehicle emission (17.1%), metal processing (6.0%), soil (10.4%) and burning of biomass (11.2%). The air pollution is mainly due to the particulate matter (Fayiga et al. 2018). According to Tao et al. (2013), air pollution was not only spread by transportation, but also from stationary emissions. The PM2.5 annual average concentration was found 99.5 µgm−3 for Beijing, 96.1 µgm−3 for Chengdu, 56.3 µgm−3 for Guangzhou, 61.6 µgm−3 for Shanghai and 76.3 µgm−3 for Shenyang from August 2013 to July 2014 (San Martini et al. 2015). In PM2.5, higher quantity of sulfate, nitrate and ammonium and organic matter is due to vehicle emission and burning solid fuel (Gautam et al. 2018). The assessment of PM2.5 concentration is important for environmental perception due to different human activities in day and night time. In this research paper, we evaluated the PM2.5 at urban sites of China. We analyzed diurnal, annual, weekday/weekend and seasonal variation in large populated cities of China including Beijing, Shanghai, Guangzhou, Chengdu and Shenyang. We also found and compared the highest and lowest PM2.5 concentration across cities and seasons for the study duration.

The study period of San Martini et al. (2015) was from April 2008 to July 2014; however, we presented the analysis of PM2.5 for periods including study duration from April 2008 to June 2017 for Beijing. For the overlap and extended duration, we are able to find the emerging (increasing or decreasing) trends of PM2.5 across the years, cities and seasons. We presented the seasonal analysis and also compared seasonal average for all study regions in this paper and the result of weekday/weekend and seasonal variation of PM2.5 shows a slight different behavior of variation of PM2.5 as compare to the one reported in San Martini et al. (2015).

This paper is comprised of the following sections. The data site, source and data preparation are given in Sect. 2. The analysis and results are shown in Sect. 3. Section 4 finally concludes the major findings.

2 Data description

2.1 Data site

The selected sampling sites for monitoring the air quality are five major cities of China as shown in the Fig. 1. Table 1 provides information of PM2.5 data used for Beijing, Shanghai, Guangzhou, Chengdu and Shenyang (http://www.stateair.net). All locations are selected from the urban sites. The study sites are the huge populated cities of China which are facing problems related to atmospheric pollution. These cities are main economic and transport centers. Beijing is the capital and is heavily populated industrial city of China. Poor air quality is a main problem in major cities of China. The climate varies all over the regions of China.

Fig. 1
figure 1

Location map of sites of study area: Beijing, Shanghai, Guangzhou, Chengdu and Shenyang

Full size image
Table 1 Gives information about sampling site in the form of geographic coordinate, approximate population and starting date of PM2.5 data for Beijing, Shanghai, Guangzhou, Chengdu and Shenyang
Full size table

2.2 Data source and preparation

U.S. diplomatic mission in China monitors air quality and the mission has presented and made available data of the PM2.5 for the public. The data is provided as hourly averaged PM2.5 data. Hourly PM2.5 is measured from the roofs of the U.S. Embassy using a beta attenuation monitor (BAM-1020, MetOne) for Beijing and consulates in Shanghai, Guangzhou, Chengdu and Shenyang. The details of BAM-1020 sensor and instrumentation can be found in (Mukherjee et al. 2017). We downloaded PM data from the link (http://www.stateair.net), however, other source for PM2.5 data could be found at (https://archive.ics.uci.edu/ml/datasets/PM2.5+Data+of+Five+Chinese+Cities).

Hourly PM2.5 data was recorded with respect to time. Diurnal and seasonal data is analyzed. The missing observations are removed. All the PM2.5 data are collected into one worksheet for a city. The time value is stored in a column with format of “YYYY/mm/dd 00:00:00”. The daily mean of PM2.5 concentrations is calculated by taking average of values during time: 00:00 to 23:00. Then, we get annual and seasonal mean concentrations and finally data is used to detect the pollution levels. For seasonal variation, we divided the yearly data into the four seasons. For seasons, we considered the months from December to February as winter, from March to May as spring, from June to August as summer and rest three month data is taken as autumn season.

3 Results and discussions

In the following section, we present analysis for annual, weekday/weekend and seasonal evaluation of PM2.5for the study duration. Thus, the information about possible PM2.5 emission sources and time when the air pollution level exceeds some certain standards is reported. Figure 2 shows cumulative probability for PM2.5 calculated for Beijing from April 2008 to June 2017, for Chengdu from June 2012 to June 2017, for Guangzhou from November 2011 to June 2017, for Shanghai from December 2011 through June 2017 and for Shenyang from August 2013 to June 2017. The standards for daily 24-h average PM2.5 are at 35 and 75 µgm−3 for U.S. and Chinese NAAS, respectively.

Fig. 2
figure 2

Cumulative probability for PM2.5 is calculated for Beijing from April 2008 to June 2017, for Chengdu from June 2012 to June 2017, for Guangzhou from November 2011 to June 2017, for Shanghai from December 2011 through June 2017 and for Shenyang from August 2013 to June 2017. Daily U.S. and Chinese NAAS of 24-h average are at 35 and 75 µgm−3, respectively

Full size image

Note: yearly and monthly PM2.5 concentration and standard deviation values as well as a number of observation are given in the Appendix A Tables 9, 10, 11, 12, and 13 for Beijing from April 2008 to June 2017, for Shanghai from December 2011 through June 2017, for Guangzhou from November 2011 to June 2017, for Chengdu from June 2012 to June 2017 and for Shenyang from August 2013 to June 2017.

3.1 Analysis of PM2.5 for Beijing

For Beijing, PM2.5 hourly data is analyzed from April 2008 through June 2017. Table 2 shows the values of PM2.5 as an annual average, standard deviation, median, minimum, maximum and percentile. The annual average of PM2.5 concentration varies between 72.7 and 104.1 µgm−3. PM2.5 shows decreasing trend for 2013 to 2016 as PM2.5 decreases from 101.7 to72.8 µgm−3. It is also clear from the percentile values given in the Table 2. For Beijing, all studied years show that PM2.5 values (average) are greater by two times and three times than that of the value equal to 35 µgm−3 (i.e., annual China national ambient air quality standard). Figure 2 shows that there are approximately 78% and 48% of days of Beijing that had daily average concentration greater than U.S. and China 24 h NAAQS of PM2.5, respectively. However, in the previous study (San Martini et al. 2015), it was reported daily average is above 35 µgm−3 for 81% of days for Beijing, 68% of days for Shanghai, 71% of days for Guangzhou, 94% of days for Chengdu and 80% of days for Shenyang, respectively.

Table 2 Shows PM2.5 annual average concentrations for Beijing, Shanghai, Guangzhou, Chengdu and Shenyang
Full size table

3.1.1 Weekday and weekend variation

To distinguish the sources of pollution, we evaluated the PM2.5 in two groups of days: weekdays and weekend. For weekdays, we considered days from Monday to Friday and the remaining days: Saturday and Sunday are treated as weekend. Figure 3 shows PM2.5 hourly concentration of Beijing for weekday, weekend and the average from April 2008 to June 2017. The daily concentration is found maximum around midnight and it is observed minimum between 14:00 and 16:00 h, afternoon rush hour. After 16:00 h, PM2.5 concentration starts to increase due to more motor vehicle emission. The observed diurnal variation is inconsistent with the results reported in Zhao et al. (2009), but our results of diurnal variation for Beijing show agreement with the findings of San Martini et al. (2015). Moreover, pollution by particulate matter (at night) may increase due to diesel trucks which are only allowed to run at night time. Due to restriction on the heavy vehicles during the day, the PM2.5 values may also reduce as also suggested by Sun et al. (2013). Figure 3 also shows Beijing PM2.5 hourly concentrations and linear fit trend which is increasing for weekday but decreasing for weekend as well as for average that can also be verified from the values of slopes given in Table 8. The daily maximum PM2.5 concentration is observed around midnight.

Fig. 3
figure 3

Beijing PM2.5 hourly concentrations and linear fit for weekday, weekend and average for the duration from April 2008 to June 2017. The daily maximum PM2.5 concentration is observed around midnight

Full size image

3.1.2 Seasonal variation

The linear fit trends and average seasonal variation of PM2.5 for 24hour in winter, spring, summer and autumn for Beijing from April 2008 to June 2017 are shown in the Fig. 4. PM2.5 variations have got considerably higher fluctuations in winter and autumn generally at midnights. However, concentration in spring seasons is slightly above the one observed in summer unlike previous studies (Zhang et al. 2013; San Martini et al. 2015). This suggests that spring has dry climate, and wind has generated more dust and cold winter has transported more primary pollutants by coal ash used for heating. The comparison of seasonal average between previous finding and the one reported in this study is presented in the Table 3. The linear curve fit shows that winter and autumn have increasing trend, but spring as well as summer has decreasing trend that can also be confirmed from the values of slopes given in Table 8.

Fig. 4
figure 4

Seasonal variations and linear fit for winter, spring, summer and autumn for Beijing from April 2008 to June 2017. Highest PM2.5 concentration is found in winter season

Full size image
Table 3 Seasonal averaged PM2.5 concentrations for Beijing as reported by Zhang and San Martini (Zhang et al. 2013; San Martini et al. 2015) and for this study
Full size table

3.2 Analysis of PM2.5 for Shanghai

For Shanghai, hourly PM2.5 data are evaluated from December 2011 to June 2017. The highest PM2.5 annual average is found 59.8 µgm−3 in 2013. The above city is not showing a clear trend like Beijing, but annual PM2.5 concentration in Shanghai is less than that of Beijing as shown in the Table 2. Figure 2 shows that there are approximately 36% of days that are matched with PM2.5 35 µgm−3 (24-h U.S NAAQS).

3.2.1 Weekday and weekend variation

PM2.5 hourly average concentrations for weekday, weekend and average from December 2011 to June 2017 for Shanghai and the linear fit trends are shown in Fig. 5. The diurnal variation of PM2.5 concentration for working days starts increasing in the after morning (working hour) as well as in the evening. The diurnal concentration is high at night which suggests that low temperature and stagnant weather may buildup the air pollutants overnight in Shanghai likewise Beijing. The weekend peak is above the weekday peak at most of the hours of the day unlike the findings of San Martini et al. (2015). The weekday peak shows that the contribution of traffic pollution to PM2.5which is smaller than that of other sources (coal combustion or dust). However, linear fit curve shows that all trends are increasing and are also shown from the values of slopes given in Table 8.

Fig. 5
figure 5

PM2.5 hourly concentrations and linear fit for weekday, weekend and average from December 2011to June 2017 for Shanghai. The daily maximum PM2.5 concentration is observed between 20:00 and 22:00 h

Full size image

3.2.2 Seasonal variation

The linear fit trends and average PM2.5 concentration for all seasons for 24 h are presented for Shanghai in the Fig. 6. Linear curve fit shows that winter has decreasing and the rest of the seasons have increasing trend that can also be visualize from the values of slopes given in Table 8. However, the seasonal annual concentration is found highest during the winter and autumn and is found less in spring and summer, similar to previous studies, but overall trend is decreasing. Highest concentration in winter indicates importance of coal burning in Shanghai (or in China). The monthly maximum is found in December, and monthly minimum is found in August. Our diurnal trend of results for Shanghai is similar to the finding of San Martini et al. (2015). The comparison for seasonal average for previous finding and the one reported in this study is presented in Table 4.

Fig. 6
figure 6

Diurnal variations and linear fit for winter, spring, summer and autumn for Shanghai from December 2011 to June 2017

Full size image
Table 4 Seasonal-averaged PM2.5 concentrations for Shanghai as analyzed by San Martini (San Martini et al. 2015) and by our study
Full size table

3.3 Analysis of PM2.5 for Guangzhou

For Guangzhou, the PM2.5 data are investigated from November 2011 to June 2017. In case of annual average, we consider year when data is available for the full year. The annual average concentration for 2012–2016 is found in the range from 56.8 to32.6 (µgm−3) as shown in Table 2. The annual PM2.5 concentration for Guangzhou is less than that of Beijing. Figure 2 shows that there are more than 80% of days that have better quality of air (i.e., match with the PM2.5 standard) in case of Guangzhou.

3.3.1 Weekday and weekend variation

For weekday and weekend, the daily maximum is found at 21:00 h, and minima are found at 16:00 and 17:00 h, respectively, as shown in the Fig. 7. In Guangzhou, weekday concentration of PM2.5 is found above the weekend concentration peak which suggests that major air pollution discharge is coming from vehicle. The night higher concentration may be due to heating from coal burning in winter.

Fig. 7
figure 7

Hourly concentrations and linear fit from November 2011 to June 2017 PM2.5 for weekday, weekend and average are presented for Guangzhou. The daily maximum PM2.5 concentration is observed around 9 pm

Full size image

3.3.2 Seasonal variation

The average seasonal concentration is found highest during the winter and it is approximately half of that of winter in summer as shown in Table 5 for Guangzhou. Figure 8 shows the average seasonal variation for Guangzhou for 24 h in winter, spring, summer and autumn for the duration from November 2011 to June 2017. Similar to that of Beijing, diurnal concentration of PM2.5 for Guangzhou is found maximum during winter at night. In winter, PM2.5 concentration is 59.8 and it is 30.9 µgm−3 in summer. The maximum variation for winter is due to heating activities, especially at night. Linear fits of Figs. 7 and 8 show that all trends are increasing that can also visualize from slope values given in Table 8.

Table 5 PM2.5 seasonal averaged concentration for Guangzhou in this study from November 2011 to June 2017and compare with previous seasonal research of San Martini (San Martini et al. 2015)
Full size table
Fig. 8
figure 8

Guangzhou’s seasonal variations and linear fit for winter, spring, summer and autumn from November 2011 to June 2017

Full size image

3.4 Analysis of PM2.5 for Chengdu

PM2.5 hourly data are evaluated from June 2012 to June 2017 for Chengdu. For 2013–2016, the annual average concentration of PM2.5 is 97.1–72.9 µgm−3 as shown in Table 2. Like Beijing, Chengdu also shows a decreasing trend of PM2.5. However, our annual measurements are still double to that of China annual NAAQS and more than thrice to that of U.S. NAAQS. But, PM2.5 annual concentration is found lower than that of reported by Tao (Tao et al. 2013). On the other hand, there are 89% of days that cross the U.S, air quality standards for Chengdu are shown in Fig. 2.

3.4.1 Weekday and weekend variation

For Chengdu, weekday and weekend variation is different from that of Beijing. Figure 9 shows daily maximum is just after the morning working hour that is approximately at 10:00–11:00 h and minimum is at 18:00 h. The difference between weekday and weekend peaks is less, but weekday peaks are higher. This suggests that working days show some contribution to daily PM2.5 concentrations due to traffic. The diurnal results of Chengdu in this study are showing an agreement with the finding of Tao et al. (2013).

Fig. 9
figure 9

PM2.5 hourly concentrations and linear fit for weekday, weekend and average from June 2012 to June 2017 for Chengdu. The daily maximum PM2.5 concentration is measured around 10:00–11:00am

Full size image

3.4.2 Seasonal variation

For Chengdu, we found pronounced seasonal variation in winter. Figure 10 shows the linear fit trends and average seasonal variation for 24 h. As previous study suggests that the PM2.5 concentrations are found higher during the winter than that of during the summer. The winter concentration is almost double to that of summer. Seasonal concentration in Chengdu is different from Beijing, but our results are similar to Tao et al. (2013, 2014) (San Martini et al. 2015) that show highest concentration in winter. The average seasonal PM2.5 concentration is decreasing as shown in Table 6.

Fig. 10
figure 10

Chengdu’s seasonal variations and linear fit for winter, spring, summer and autumn from June 2012 to June 2017 are shown

Full size image
Table 6 Comparison of PM2.5 seasonal averaged variation measured by Tao et al. (2013) from April 2009 to January, (Tao et al. 2014) for 2011, (San Martini et al. 2015) from June 2012 to July 2014 and in this study from June 2012 to June 2017 for Chengdu
Full size table

3.5 Analysis of PM2.5 for Shenyang

For Shenyang, PM2.5 data is examined from April 2013 through June 2017. The annual trend of PM2.5 for Shenyang is similar to that of Shanghai. The highest annual average concentration of PM2.5 is found as 78.8 µgm−3 in year 2015 as shown in Table 2.

3.5.1 Weekday and weekend variation

The weekday and weekend variation in Shenyang is similar to Chengdu. Figure 11 shows daily maxima and minima arise in Shenyang before that of Chengdu. The result suggests that wind during night clears many pollutants from the air in Shenyang as well as in Chengdu. Likewise Beijing, weekend peaks are found higher than that of weekdays in Shenyang too. The soil, dust and ash of coal were the leading sources of PM2.5 in Shenyang as also suggested by Ni et al. (2012). So, our result is consistent with Ni et al. (2012) for Shenyang.

Fig. 11
figure 11

Weekend, weekday and average variation of PM2.5 and linear fit for Shenyang from April 2013 to June 2017. Similar to Beijing weekend peaks are higher than weekdays in Shenyang

Full size image

3.5.2 Seasonal variation

The seasonal variation is more pronounced in winter than the other seasons. The average seasonal concentration is found as 79.1 µgm−3 in winter, 53.9 µgm−3 in spring, 39.4 µgm−3 in summer and 59.1 µgm−3 in autumn, respectively. The seasonal average PM2.5 concentrations for Shenyang as reported by San Martini et al. (2015) and in this study from April 2013 to June 2017 are compared and are given in the Table 7.The concentration of winter is more than double of that of summer. This seasonal variation is higher than Guangzhou and Shanghai but lower than Chengdu and Beijing. The average concentrations of PM2.5 for all seasons for 24 h for Shenyang as well as linear fit trends are shown in the Fig. 12. Linear fit curve of Figs. 9, 10, 11 and 12 show that all trends are decreasing that can also be justified from slope values given in Table 8.

Table 7 Seasonal averaged PM2.5 concentrations for Shenyang by San Martini et al. (2015) and in this study from April 2013 to June 2017 are compared
Full size table
Fig. 12
figure 12

Diurnal variations and linear fit for winter, spring, autumn and summer for Shenyang. In summer, PM2.5 concentrations are almost half of that of winter

Full size image
Table 8 The equation of linear fit for weekday, weekend, and average and seasons (winter, spring, summer and autumn) for five cities
Full size table

3.6 Linear fit curves for five cities of China

The linear fit analysis of PM2.5 with regard to air quality for five populated cities (Beijing, Chengdu, Guangzhou, Shanghai and Shenyang) is given in the Table 8. We study PM2.5 variations for annual, weekday/weekend and seasonal evaluation for the study duration. PM2.5 concentration and linear fit for weekday/weekend and average are shown in the Figs. 3, 5, 7, 9 and 11 for five cities. PM2.5 concentration and linear fit for four seasonal variations are shown in the Figs. 4, 6, 8, 10 and 12. The Table 8 shows the equation of linear fit for weekday, weekend and average as well as seasons that are winter, spring, summer and autumn for five cities, respectively. Linear fit equation is in the form of y = mx + c, where x represents the hour of day of weekday, weekend, average for seasons that are winter, spring, summer and autumn The “m” shows slope of trend line, and “y” represents the mean values of PM2.5. If the slope of trend line is positive then it shows increasing trend, whereas the negative slope is the indication of decreasing trend. The values of R2 represent goodness-of-fit which means the difference of given data and fitted values, while SSE values show the sum of square of residuals which means the expected deviation from the actual data.

4 Conclusion

Although there are decreasing trends in pollutions with respect to set standards in the past few years, but overall China is still facing increasing trend for fine particulate pollution significantly so far. In this study, we analyzed the PM2.5 concentration for four cities of China including Beijing, Shanghai, Guangzhou, Chengdu and Shenyang. For Beijing, our study duration is from April 2008 to June 2017. We, for the first time, presented the long-term diurnal and seasonal analysis of PM2.5 for China, especially for Chengdu (June 2012–2017 June), Guangzhou (November 2011–2017 June) and Shenyang from April 2013–2017 June. It is found that Beijing and Guangzhou had the highest and lowest annual average PM2.5 concentrations, respectively. The highest diurnal concentrations are found in Beijing, Shanghai and Guangzhou at night. For Chengdu and Shenyang, diurnal (weekday) concentrations are found highest after the morning working hours which suggests that night wind clears many of the pollutants from the air which leads to morning maximum. However, the diurnal concentrations for weekend and weekday are found lowest during 15:00–18:00 h for all studied cities except for Shanghai (weekday) where lowest concentration is found during (3:00 am–4:00 am).Weekend peaks are found above to that of the working day for Beijing, Shenyang and Shanghai. This suggests that in these cities traffic contribution to daily PM2.5 concentrations is relatively small than rest of the cities. We found significant seasonal concentration in winter and autumn. The summer concentrations are half the winter concentrations. Unfavorable weather condition and burning of coal may lead to high PM2.5 concentrations in winter and autumn season at night. Finally, our results show that air quality has improved for Beijing and Chengdu from 2013 to 2017 and for Guangzhou (2012–2017 June), but it is needed to perform more progressive measures to come close to that of set PM2.5 standards.