Study on Temporal and Spatial Evolution Characteristics of Air Quality in Urban Agglomerations in Northwestern China

This paper analyzes the air quality monitoring data of Qingyang, Tianshui and Pingliang in northwestern China in 2015 by time series analysis. The research results are as follows: (1) the concentrations of PM2.5, PM10 and SO2 in the three cities began to decline in March, reached a minimum in September before rising gradually; (2) concentrations of PM2.5, PM10 and SO2 in the three cities were low in summer and autumn, but high in winter and spring; (3) the coefficient of correlation among PM2.5, PM10 and SO2 in the three cities was greater than 0.589, indicating a strong correlation.


Introduction
China has experienced rapid industrialization and urbanization with the development of technology and economy since the reform and opening-up. The advancement of industrialization and urbanization is accompanied by huge consumption of resources and severe damages to the environment. The extensive development model with high energy consumption and low efficiency has further aggravated the destruction of the ecological environment and caused a series of environmental problems [1] . Air pollution will not only cause serious economic losses [2] , but also reduce atmospheric visibility [3] . The particulate matter in the air can induce respiratory diseases [4] and cardiovascular and cerebrovascular diseases [5] , having a significant impact on human health. According to statistics, about 6.1 million people died of air pollution-related diseases [6] in 2016.
The monitoring and characteristic analysis of air quality have become the current research hotspots of Chinese and foreign scholars in recent years. Mu et al. [7] studied a nationwide large-scale smog incident in China in January 2013 and estimated that the incident caused about RMB 23 billion direct economic losses to traffic and health based on the publicly reported data. Liao et al. [8] studied the variation trend of air pollutant concentration in Shenzhen from 2006 to 2016. The results showed that concentrations of SO 2 , PM 10 and NO 2 in Shenzhen presented an overall downward trend from 2006 to 2016 and that the air quality had improved significantly year by year but there was still a big gap from air quality standards of developed countries. Existing studies mostly focus on municipalities directly under the central government such as Beijing, Shanghai and Tianjin and capital cities in central and eastern China. Related studies in northwestern China also focus on provincial capital cities. The overall research on the air quality in small underdeveloped urban agglomerations has not been reported yet.
Qingyang, Pingliang and Tianshui are located in northwestern China, the climate is the warm temperate zone, continental monsoon climate. This study analyzes and evaluates the air quality status of the three cities in 2015 and analyzes the correlation of cities and pollutants with the multiple regression method with a view to provide data support for air pollution prevention and control decision-making and comprehensive crossregional management in third-and fourth-tier cities in under-developed regions in northwestern China.

Data source
The air quality data of Qingyang, Pingliang and Tianshui in 2015 came from the China Air Quality Online Monitoring and Analysis Platform (https://www.aqistudy.cn/). The air quality monitoring data included average concentrations of PM 2.5 , PM 10 , SO 2 , CO, NO 2 and O 3 , etc.

Evaluation criteria
The evaluation criteria used in this study are from the Ambient Air Quality Standard (BG3095-2012) [9] .

Research method
The analysis was mainly conducted with the time series analysis method and the multiple regression method in statistics, and the data was processed preliminarily based on Excel 2007. The variation trend of PM 2.5 , PM 10  According to the figure, the PM 2.5 concentration in all the three cities was low in the middle and high on both sides, and had a small peak in July. The PM 2.5 concentration in Pingliang was slightly higher than that in the other two cities, with a maximum of 83 μg/m 3 and a minimum of 24 μg/m 3 .   and Qingyang had small variations, but the concentration in Pingliang was much higher than that in Qingyang. The concentration of NO 2 in Qingyang was the lowest, ranging from 12 to 95 μg/m 3 . That in Tianshui had the greatest variation.  In summary, the concentrations of PM 2.5 , PM 10 and SO 2 in the three cities in eastern Gansu showed a downward trend from March, fell to the lowest level in September and then increased gradually, among which the SO 2 concentration remained at a low level from April to October and increased sharply thereafter. The NO 2 concentrations in the three cities were slightly higher in November and December. The variation of their O 3 concentrations was different from that of other pollutants, showing an inverted "U" shape.

Correlations between air pollutants in different cities
Correlation analysis refers to the analysis of two or more related variable elements to measure their correlation. Table 1 shows the results of Spearman correlation analysis of different pollutants in the same city for the three cities in 2015. .979 ** 1 SO2 .915 ** .887 ** 1 NO2 .685 * .567 .713 ** 1 O3 -.523 -.385 -.556 -.907 ** 1 **. At the level of 0.01 (two-tailed), indicating significant correlation. *. At the level of 0.05 (two-tailed), indicating significant correlation. Table 1 indicates a strong positive correlation among PM 2.5 , PM 10 and SO 2 in the three cities with a coefficient of correlation between 0.589 and 0.979, a positive correlation between NO 2 and PM 2.5 , PM 10 and SO 2 in Qingyang and Pingliang and a negative correlation between the same in Pingliang, and a strong negative correlation between O 3 and NO 2 in Qingyang and Tianshui. The reason is that O 3 in the air near the ground in urban areas is generated by a series of photochemical reactions of its precursor NOx [10] : In the real environment, the more oxidizing free radical present will react with NO, leading to the termination of the zero-cycle reaction in formulas 1 to 3 and thereby the accumulation of O 3 concentration [25] as shown in formulas (4 ) and (5).
The reactions above show that the most fundamental cause of O 3 generation is the photolysis of NO 2 . Therefore, NO 2 , as one of the precursors of O 3 , shows a good negative correlation with the concentration. However, O 3 and NO 2 had a weak correlation in Pingliang due to the high level of the NO 2 concentration throughout the year in combination with the influence of volatile organic compounds (VOCS) and other pollutants produced by the chemical industry and other industries [11] . Table 2 shows the results of further analysis of the Spearman correlation of the same pollutant among different cities. Except for the correlation coefficients of NO 2 between Pingliang and Qingyang and between Pingliang and Tianshui, the others were close to or greater than 0.8, indicating a strong correlation. According to Figure 5, the NO 2 concentration in Pingliang remained at a high level throughout the year, resulting in the correlation coefficients of NO 2 between Qingyang and Pingliang and between Pingliang and Tianshui ranging from 0.3 to 0.4. Studies have shown that the contribution rate of motor vehicles to NO 2 is about 60% [12] . Pingliang as an important coal-producing area in Gansu Province, it had an annual output of raw coal of 21.3556 million tons in 2015 . Except for mining and processing, the tail gas emission caused by the transportation of a large amount of raw coal was much higher compared to the other two cities, which is also an important reason for this phenomenon. .958 ** .869 ** 1 **.At the level of 0.01 (two-tailed), indicating significant correlation.

Time series analysis of air quality
PM 2.5 and PM 10 concentrations in all the three cities were low in summer and autumn and high in winter and spring. The concentration of SO 2 showed a typical "U" shape, remained at a low level from April to October and increased significantly in other months. That of O 3 showed an inverted "U" shape with the characteristic of being "high in spring and summer and low in autumn and winter". This agrees with the research results of Wang et al. [13] on urban air quality in the Fenwei Plain. These variation characteristics are related to the combined effect of pollutant emissions, meteorological conditions, and photochemical reactions. In summer, there are fewer combustion sources and pollution emissions, strong vertical diffusion of the atmosphere, more precipitation and strong scavenging action, which can lower the concentration of pollutants. A large amount of pollutants from central heating and the emission of fossil fuel combustion coupled with unfavorable diffusion conditions during the heating period in winter (mid-November to mid-March of the following year) in comparison with other seasons result in a high concentration of pollutants . NO 2 mainly comes from automobile exhaust, power plants and industrial activities, among which automobile exhaust is the main emission source [14] . Huating Coal Field in Pingliang is the largest coalfield in Gansu Province. The transportation and deep processing of coal produce a large amount of NO2, resulting in a high NO 2 concentration throughout the year in Pingliang.

Correlation analysis and reasons
There was a strong positive correlation among PM 2.5 , PM 10 and SO 2 in the three cities, with a correlation coefficient between 0.589 and 0.979. Their main pollutants were all particulate matter, which was the same as in other parts of China and was consistent with the research results of scholars such as Sun [15] on northern cities. As a coal resource-based city, Pingliang was faced with a large amount of NO 2 from the mining, processing and transportation of raw coal.
According to the analysis of the correlation among different cities for the same pollutant, the other coefficients were close to or greater than 0.8, indicating a strong correlation, except for the correlation coefficient of NO 2 between Pingliang and Qingyang and between Pingliang and Tianshui. The three cities, all located in Gansu Province in northwestern China, have similar geographical environment and climate, city scale and level of social and economic development which result in their strong correlation.

Advantages and limitations
This paper studies the characteristics and correlation of air quality, monthly average mass concentrations of pollutants and AQI in small urban agglomerations in northwestern China, explores their underlying variation features and regional synergic relationship and discusses the variation mechanism. In addition, it conducts a multiscale analysis on the evolution of each pollutant in time series and provides more research results. However, it only analyzes the evolution characteristics of the three cities in 2015 and lacks research on large-scale evolution rules. Moreover, urban air quality is also affected by meteorological factors, population, economic development and other factors, which will be studied in future work.

Conclusion
In summary, this paper makes the following conclusions: 1) The concentrations of PM 2.5 , PM 10 and SO 2 in the three cities showed a downward trend from March, fell to the lowest level in September and then increased gradually, among which the SO 2 concentration remained at a low level from April to October and increased sharply thereafter. The NO 2 concentration in Pingliang was 2.17 times that of Qingyang and 1.31 times that of Tianshui. The variation of their O 3 concentrations was different from that of other pollutants, showing an inverted "U" shape.
2) In the three cities, particulate matter had a higher over-limit ratio. The over-limit ratio of PM 10 in Qingyang was slightly higher than that of PM 2.5 , and the ratio of fine particulate matter in Pingliang and Tianshui was higher than that of coarse particulate matter. SO 2 and NO 2 concentrations had a low over-limit ratio and exceeded the standard only in a few months of the year. The concentration of O 3 did not exceed the standard.
3) According to the analysis of the Spearman correlation among different pollutants in the three cities, the correlation coefficient of PM 2.5 , PM 10 and SO 2 was between 0.589 and 0.979, indicating a strong correlation. According to the analysis of the Spearman correlation among different cities for the same pollutant, the other correlation coefficients except for NO 2 were all high.
These findings can help in better understanding the evolution characteristics of air quality in small urban agglomerations in northwestern China and provide a reference for the application of policies. PM 2.5 and PM 10 concentrations in Pingliang were slightly higher than those in the other two cities due to the exhaust gas produced by coal processing, production and transportation. Qingyang, which is also positioned as an energy base, should learn from Pingliang, accelerate industrial upgrading and improve scientific and technological innovation capabilities to avoid deterioration of air quality due to energy mining, transportation and production.