Spatiotemporal Characteristics of Air Quality across Weifang from 2014–2018

Air pollution has become a severe threat and challenge in China. Focusing on air quality in a heavily polluted city (Weifang Cty), this study aims to investigate spatial and temporal distribution characteristics of air pollution and identify the influence of weather factors on primary pollutants in Weifang over a long period from 2014–2018. The results indicate the annual Air quality Index (AQI) in Weifang has decreased since 2014 but is still far from the standard for excellent air quality. The primary pollutants are O3 (Ozone), PM10 (Particles with aerodynamic diameter ≤10 µm), and PM2.5 (Particles with aerodynamic diameter ≤10 µm); the annual concentrations of PM10 and PM2.5 show a significant reduction but that of O3 is basically unchanged. Seasonally, PM10 and PM2.5 show a U-shaped pattern, while O3 exhibits inverted U-shaped variations, and different pollutants also present different characteristics daily. Spatially, O3 exhibits a high level in the central region and a low level in the rural areas, while PM10 and PM2.5 are high in the northwest and low in the southeast. Additionally, the concentration of pollutants is greatly affected by meteorological factors, with PM2.5 being negatively correlated with temperature and wind speed, while O3 is positively correlated with the temperature. This research investigated the spatiotemporal characteristics of the air pollution and provided important policy advice based on the findings, which can be used to mitigate air pollution.


Introduction
Since the Chinese economic reform, with the rapid industrial development and growing energy consumption in China, air pollutant emissions have increased continuously, thus exacerbating air pollution. For example, in January 2013, a wide range and long duration of hazy weather occurred in Central and eastern China, raising public concern about urban air quality [1][2][3]. To address more serious air pollution, the Chinese government announced the new National Ambient Air Quality Standards in 2012 and completed all-round monitoring of air quality in cities at the prefecture level in 2015. By 2016, air quality monitoring was implemented in 338 cities in China. Air pollution can cause respiratory, cardiovascular, and cerebrovascular diseases through gas exchange in the lungs [4,5]. Air pollution also reduces atmospheric visibility and causes smog [6,7] and inconvenience to travelers. Therefore, it is critical to accurately identify the temporal and spatial characteristics of air quality and its main influential factors and to correspondingly control and alleviate air pollution.
At present, existing research on the temporal and spatial characteristics of air quality can be classified into three types: Studies based on remote sensing inversion of the aerosol optical depth (AOD), studies based on ground monitoring stations and studies based on modelling which is complementary to satellite and ground-based observations; Studies based on remote sensing inversion usually focus on large-scale areas to provide a regional overview [8]. Due to the high frequency monitoring, research based on real-time ground-monitored data is significant to better explore the detailed patterns (seasonal, 3 of 15 between primary pollutants and meteorological factors quantitatively. This paper helps to improve the understanding of the temporal and spatial characteristics and the mechanisms of air pollution in Weifang City, and can provide scientific and technical support for subsequent air pollution research. Furthermore, the study helps to formulate effective prevention and control measures and offers policy advice specifically for environmental management in Weifang and other similar areas.
This paper includes four sections. Section 2 describes data sources and methodologies. The temporal and spatial characteristics and the influential factors of the air quality in Weifang are explained in Section 3, followed by the conclusions in Section 4.

Study Area
Weifang City is located in the middle of the Shandong Peninsula ( Figure 1), with Zibo City to the west, Linyi City to the south, Qingdao to the east and Laizhou Bay, Bohai Sea, to the north, between 118.17 • E-120.01 • E and 35.68 • N-37.43 • N. The study area includes four districts, six cities and two counties, with a total area of 16,000 km 2 . Weifang City is high in the south and low in the north. The southern part is dominated by low hills, while the northeastern part consists of mainly plains, bays and rivers. Weifang is a semihumid area with a temperate continental monsoon climate. In 2018, Weifang was one of the most rapidly developing cities in Shandong Province. There have been rapid advancements in various fields, such as industry, transportation, and economy. However, the air pollution that comes with these developments has become increasingly serious. The air quality of the city is slightly below average compared to both those of other cities in Shandong and those of key cities in China. Nevertheless, there are few studies and reports on different air pollution indicators in Weifang. Investigation and analysis of the temporal and spatial characteristics of air pollutants are therefore important for improving the air quality in Weifang.

Data Sources
The monitoring station data used in this study were obtained from 5 national stations, 4 provincial stations, and 29 urban stations in Weifang and included six air pollutants, namely, coarse particulate matter (PM 10 ), fine particulate matter (PM 2.5 ), ozone (O 3 ), nitrogen dioxide (NO 2 ), sulfur dioxide (SO 2 ), and carbon monoxide (CO). Data are retrieved by the monitors every five minutes. All the data used in this paper are in units of hours, which are calculated as the average of the measurements taken every five minutes. The meteorological data used in this paper are from the Weifang Statistical Bulletin, including the wind speed and air temperature.
The AQI is a dimensionless index that quantitatively describes the air quality. According to the new National Ambient Air Quality Standards (GB 3095-2012) implemented by the Ministry of Environmental Protection (now known as the Ministry of Ecology and Environment) of the People's Republic of China in 2012, different individual air quality indices (IAQIs) of various factors (PM 10 , PM 2.5 , O 3 , NO 2 , SO 2 , and CO) are first computed, and then the highest IAQI value is taken as the value of the AQI. The equation is given as follows.

Data Processing
Up to 10 million air quality data points are used in this paper. Traditional Geographical Information System (GIS)-based statistical analysis methods cannot satisfy the processing of a large amount of data. To achieve efficient calculations and analyses and to allow interactive visualization of the temporal and spatial characteristics of the air pollutants, the frequent pattern (FP)-growth algorithm is used in this paper [22].
The FP-growth algorithm is an FP mining algorithm based on the FP-tree structure. FP-tree structure is an extended prefix-tree structure for storing compressed, crucial information about frequent patterns. The basic idea of this algorithm is as follows. First, for each item, its conditional projection database and projection FP-tree are constructed. Then, this process is repeated for each newly built FP-tree until the constructed new FP-tree is empty or contains only one path. Finally, when the constructed FP-tree is empty, its prefix is frequent pattern.
The FP-growth algorithm can improve efficiency of mining with techniques as follows. By preserving key information of each set, a large database in the transaction database is compressed into an FP-tree, which can avoid costly, repeated database scanning. Furthermore, a partitioning-based method is used to decompose the mining task into a set of smaller tasks for mining confined patterns in conditional databases, which dramatically reduces the search space [22]. The FP-growth algorithm utilizes tree structures to compress transactions, which greatly enhances the efficiency for a large amount of data processing [23]. Han et al. reported that the FP-growth method is efficient and scalable for mining both long and short frequent patterns, and is about an order of magnitude faster than the Apriori algorithm [22]. This paper employs the FP-growth algorithm based on the Spark platform.
A valid check on hourly data was conducted to eliminate problematic and missing data points before data processing tasks. Statistical analysis and ArcGIS are used to explore the temporal and spatial patterns of air pollution. For temporal statistical analysis, C++ language is applied in this study to obtain annually, seasonal, monthly, and daily variations of air pollutants; these variations are calculated by the arithmetic means of the hourly renewed data from 38 ground monitoring stations from 2014-2018. For spatial analysis, spatial characterization is conducted through the use of spatial analysis methods with the support of ArcGIS software (published by the Environmental Systems Research Institute (ESRI), Redlands, California, USA).

Urban Air Quality in Weifang
3.1.1. AQI Characteristics Figure 2 shows the variation characteristics of the annual average AQI in Weifang from 2014-2018. The figure indicates that the AQI in Weifang is high. From 2014-2017, the AQI is higher than 100, which is slightly polluted. Specifically, the air pollution in Weifang is serious in 2014, with an average AQI of 124. In subsequent years, the value decreases, with a decrease of 20.16%. However, the value is still far from the standard for excellent air quality, where AQI < 50. The decreasing trend may be closely related to the environmental protection actions taken under the eight major initiatives, including controls on pollution from industry, fossil fuel consumption, and motor oil and gas consumption in 2017.

Urban Air Quality in Weifang
3.1.1. AQI Characteristics Figure 2 shows the variation characteristics of the annual average AQI in Weifang from 2014-2018. The figure indicates that the AQI in Weifang is high. From 2014-2017, the AQI is higher than 100, which is slightly polluted. Specifically, the air pollution in Weifang is serious in 2014, with an average AQI of 124. In subsequent years, the value decreases, with a decrease of 20.16%. However, the value is still far from the standard for excellent air quality, where AQI < 50. The decreasing trend may be closely related to the environmental protection actions taken under the eight major initiatives, including controls on pollution from industry, fossil fuel consumption, and motor oil and gas consumption in 2017.   Figure 3 illustrates the proportions of days with different AQI categories in Weifang from 2014-2018. In the past five years, the proportion of days with excellent and good air quality to the total number of days has gradually increased. The proportion reaches 60% in 2018, but there is still a gap between this value and the expected value of 80% of the Chinese government. The number of days when the city is slightly polluted decreases gradually from 2014-2017, but the number of slightly polluted days rises slightly in 2018. The number of days when the city is heavily or severely polluted significantly decreased by 20% in 2018 compared to 2014. However, there are still days when the city is severely polluted.
The variations in the daily average AQI in Weifang from 2014-2018 are shown in Figure 4. There are certain periodic characteristics, with five U-shaped distributions. From 2014-2018, the number of days when the AQI falls in the second category (good air quality) is the greatest, accounting for approximately 49% of the total days. Days with good air quality (51 < AQI < 100) are often found in July, August, and September. This phenomenon is because pollutant diffusion and dilution are more favorable when precipitation and vegetation coverage increase, and surface convection becomes stronger in Summer. The number of days when the AQI falls in the slightly polluted level ranks second (29%). The number of days with moderate pollution accounts for approximately 12%, while those with AQI > 200 (heavy and severe pollution) account for approximately 5% of the total days. The days with heavy and severe pollution occur mostly in January, November, and December, which is consistent with many other studies [9,24]. This finding may be because the meteorological conditions in Winter, such as frequent temperature inversion, low air temperatures, weak convection, little precipitation, and low humidity, do not facilitate air movement in Weifang. Furthermore, heating during Winter leads to a surge in fossil fuel consumption and thus increased emissions of pollutants [25]. In summary, the AQI in Weifang is high in Winter and low in Summer.

Primary Pollutants Affecting the AQI in Weifang
Statistical analyses are conducted to calculate the proportions of days when different pollutants become the major pollutants in Weifang ( Figure 5) in 2018. The results demonstrate that the number of days when O 3 is the major pollutant in the city is the greatest (148 days) and accounts for approximately 40.55% of the total days, followed by PM 10 and PM 2.5 , with percentages of 31.25% and 20.82%, respectively. However, the days when NO 2 , SO 2 , and CO are the major pollutants account for only 7.67% of the total days. This finding is similar to the conclusions of Zhan et al. [9] and Song et al. [26], who performed AQI analyses across China and discovered that PM 2.5 , PM 10 , and O 3 are the major air pollutants. However, the proportion of days when PM 2.5 is the major pollutant is the highest for mainland China and that of O 3 is the highest in Weifang [26]. To more specifically analyze the mechanisms affecting the air quality in Weifang, this paper evaluates the temporal and spatial characteristics of the top 3 primary pollutant (O 3 , PM 10 , and PM 2.5 ) that affect the AQI separately.

Seasonal and Monthly Characteristics
The statistical seasonal and monthly characteristics of the primary pollutants in Weifang are illustrated in Figures 7 and 8, respectively. In Winter, PM 2.5 and PM 10 are the primary pollutants. In Summer, O 3 becomes the major pollutant, followed by PM 10 . Specifically, the annual peak of the daily maximum 8-hour concentration of O 3 is found in summer, and the maximum concentration occurs in June. An annual trough appears in Winter, and the minimum concentration is reached in January and December. This phenomenon results in a distinct inverted U-shaped variation curve with seasonal O 3 concentrations in the ascending order of winter > spring > autumn > summer. In contrast to the monthly O 3 variations, the variations in the PM 2.5 concentrations follow a distinct U-shaped curve. A trough appears in summer, which agrees well with previous studies [28,29]. The minimum concentration normally occurs in August, whereas the maximum is reached in January and December. The seasonal PM 2.5 concentrations follow the ascending order of summer > autumn > spring > winter. The PM 10 variations are similar to those in PM 2.5 , but the PM 10 trough occurs in autumn, and the seasonal PM 10 concentrations can be arranged in the ascending order of autumn > summer > spring > winter.  The O 3 concentration in Weifang increases in summer, probably because intense sunshine, prolonged solar radiation, and high temperatures during summer are conducive to more intense photosynthesis, thereby promoting the conversion of nitrogen oxides and volatile organic pollutants into O 3 [26,30]. The O 3 concentration is significantly and positively correlated with the air temperature. In contrast, the high regional temperatures, stronger solar radiation, rapid surface convection, and increased precipitation and vegetation coverage in Summer facilitate air movement and thus promote diffusion and dilution of atmospheric PM, such as PM 2.5 and PM 10 [31]. In addition, weak regional convection and a relatively stable atmosphere in Winter easily result in inversion layers, and this situation is not conducive to the diffusion of atmospheric PM. Moreover, both central heating and greater energy consumption in Winter give rise to a higher PM 2.5 concentration.

Daily Characteristics
The Fourier fitting method is used to conduct Fourier approximation analysis of the characteristics of the daily average concentrations of the primary pollutants in Weifang from 2014-2018. The Fourier approximation functions of the AQI and the primary pollutants, O 3 , PM 10 , and PM 2.5 , are shown in Figure 9 (multiple determination coefficient R2 > 0.99). As illustrated in Figure 9, the concentrations of both PM 10 and PM 2.5 peak at approximately 09:00. The peak concentrations are 153 µg/m 3 and 82 µg/m 3 , respectively. The troughs are observed at approximately 16:00, probably because human activities become less intense at this time, leading to a reduction in atmospheric PM emissions. Furthermore, there is sufficient time for most pollutants to gradually diffuse. This finding is partly in agreement with previous studies of Wang et al. [32], who showed that the daily average value of PM 2.5 concentration peaked at 10:00 and the valley occurred at 16:00. In contrast, the characteristics of O 3 are opposite to those of the AQI, PM 10 , and PM 2.5 . The minimum O 3 concentration of 40 µg/m 3 is observed from 06:00-07:00; then, the concentration increases rapidly and reaches its maximum of 123 µg/m 3 from 14:00-16:00, suggesting that the O 3 concentration is positively correlated with the solar radiation intensity. In general, the daily characteristics of PM 2.5 , PM 10 , and AQI are greatly similar, but the AQI shows some delay compared to PM 2.5 and PM 10 .

Spatial Characteristics of the Primary Pollutants that Affect the AQI
The spatial distribution of the annual average concentrations of the primary pollutants is shown in Figure 10. One should note that the county/district air pollution level is represented by the average of points where concentrations are monitored within the area, which may lead to extrapolation errors/bias. However, generally speaking, these statistical results provide a way to understand the spatial distribution characteristics of regional air pollution, which support the local government decision-making for environmental governance.
normally approximately 110 μg/m³. In particular, the highest concentration of 135.14 μg/m 3 is observed in the Weicheng District. The concentrations in the Hanting District and Linqu are 131.02 μg/m 3 and 129.33 μg/m 3 , respectively. Gaomi and Changyi have the lowest concentrations of 103.30 μg/m 3 and 106.75 μg/m 3 , respectively.
The PM2.5 concentration in Weifang shares similar spatial characteristics with the PM10 concentration, with higher values in the west compared to the southwest. The regional average concentration falls within the range of 55-85 μg/m 3 . Specifically, the highest PM2.5 concentration of 83.38 μg/m 3 is measured in the Weicheng District, followed by that in Qingzhou (74.22 μg/m 3 ). Anqiu and Changyi have the lowest concentrations of 58.24 μg/m3 and 56.13 μg/m 3 , respectively.
There may be two reasons for the spatial characteristics of PM2.5 and PM10. One reason is the topography of the region: the southern part of Weifang consists mostly of low hills with extensive vegetation, which can effectively adsorb atmospheric particulates and reduce the amount of ground aerosols reaching the atmosphere, which thereby lowers the PM concentrations [20]; in contrast, the western Weifang area includes Shouguang and Qingzhou counties, and is adjacent to Zibo City, a heavy industrial city with more severe air pollution. Substantial air pollutants and toxic particles from Zibo City can be transported to the western areas, resulting in an intense air pollution effect. The other reason is related to the rapid population growth in the western cities, such as Linqu and Qingzhou. The population densities in these cities are higher, resulting in increased residential emissions of atmospheric pollutants and therefore higher regional PM concentrations [34].

Correlation Between the Primary Pollutants and Meteorological Factors
The above analysis shows that PM10 and PM2.5 have similar characteristics, so in this section, the correlations among PM2.5, which represents PM10, and O3 and meteorological factors are analyzed. Figure 11 shows the correlations of the PM2.5 concentration with meteorological factors. As illustrated in panel A, the PM2.5 concentration has a negative relationship with the temperature. The PM2.5 concentration is higher from −6-(−5 ℃), when PM pollution occurs frequently. It can be deduced that this temperature range is the most conducive to PM2.5 generation. In addition, this temperature range is not conducive to pollutant diffusion, easily leading to accumulation. When the temperature exceeds 15℃, the PM2.5 concentration decreases, possibly because high temperatures and abundant precipitation in Summer help to effectively wash away pollutants. As shown in panel B, the PM2.5 concentration also has a negative relationship with the wind speed. More specifically, when the wind There may be two reasons for the spatial characteristics of PM 2.5 and PM 10 . One reason is the topography of the region: the southern part of Weifang consists mostly of low hills with extensive vegetation, which can effectively adsorb atmospheric particulates and reduce the amount of ground aerosols reaching the atmosphere, which thereby lowers the PM concentrations [20]; in contrast, the western Weifang area includes Shouguang and Qingzhou counties, and is adjacent to Zibo City, a heavy industrial city with more severe air pollution. Substantial air pollutants and toxic particles from Zibo City can be transported to the western areas, resulting in an intense air pollution effect. The other reason is related to the rapid population growth in the western cities, such as Linqu and Qingzhou. The population densities in these cities are higher, resulting in increased residential emissions of atmospheric pollutants and therefore higher regional PM concentrations [34].

Correlation Between the Primary Pollutants and Meteorological Factors
The above analysis shows that PM 10 and PM 2.5 have similar characteristics, so in this section, the correlations among PM 2.5 , which represents PM 10, and O 3 and meteorological factors are analyzed. Figure 11 shows the correlations of the PM 2.5 concentration with meteorological factors. As illustrated in panel A, the PM 2.5 concentration has a negative relationship with the temperature. The PM 2.5 concentration is higher from −6-(−5°C), when PM pollution occurs frequently. It can be deduced that this temperature range is the most conducive to PM 2.5 generation. In addition, this temperature range is not conducive to pollutant diffusion, easily leading to accumulation. When the temperature exceeds 15°C, the PM 2.5 concentration decreases, possibly because high temperatures and abundant precipitation in Summer help to effectively wash away pollutants. As shown in panel B, the PM 2.5 concentration also has a negative relationship with the wind speed. More specifically, when the wind speed is 0-4 m/s, the frequency of serious PM 2.5 pollution is the highest. However, when the wind speed is higher than 4 m/s, the severe pollution cases decrease, suggesting that at such wind speeds, winds facilitate diffusion and dilution of atmospheric pollutants more effectively, while winds slower than 4 m/s can weaken these effects. In addition, winds slower than 4 m/s will entrain dust, leading to increased pollutant concentrations in certain localities. Hence, at wind speeds of 0-2 m/s, effective measures should be taken in advance to avoid exacerbated air pollution due to PM diffusion. These findings echo many previous studies which showed a negative correlation between the temperature, wind and the air pollution [2]. However, the same finding was not observed in the study of Zhan et al. [9], who presented a positive relationship between AQI and wind speed and reported that both downwind areas and dust storm-prone areas are likely to suffer serious air pollution by increasing wind speed.

Policy Implications
Some effective policy advice based on the findings of this study is presented to reduce air pollution. First, the main primary pollutants influencing AQI in Weifang are O 3 , PM 10 and PM 2.5 , which should receive more attention by the government. According to their spatio-temporal characteristics, more emphasis should be placed on mitigating PM 10 and PM 2.5 in winter and in western regions, and O 3 pollution in Summer, which is mainly distributed in central areas.
Second, O 3 is the primary pollutant influencing AQI in Weifang, which is not a special case. Studies in Shenzhen [33] and Beijing [27] also showed similar conclusions. With more efforts were devoted to control the pollution of particulate matter, such as PM 2.5 , near-surface ozone will be another key factor affecting China's air quality after particulate matter. However, the attention to ozone in China is still in its infancy. The Chinese government should pay more attention to ozone pollution. For instance, the O 3 concentration is closely related to intense human activities and temperature, which are usually high in Summer and at approximately 15:00 daily, and initiatives such as strictly controlling volatile organic pollutants and vehicle emissions should be implemented by the government in the central region during the period when O 3 concentration peaks (15:00 pm).
Third, serious air pollution (such as PM 10 and PM 2.5 ) occurred in western Weifang (such as Linqu and Qingzhou) due to two reasons. One is related to the local high population density, the other is correlated with the air pollution transportation from adjacent areas. Zibo City is located on the western side of Weifang City, a heavy industrial city with more severe air pollution. The air pollutants from Zibo City can be transported to the western areas in Weifang. The government should promote the improvement of industrial structure, implement and set pollutant emission standards and use clean energy to decrease air pollution. Additionally, this result also implies that it is a global problem for air pollution mitigation, and implying that is significant for regional cooperation [35]. When formulating measures to prevent and control air pollution, attention should be paid to the interaction between cities and surrounding areas, and the joint prevention and cooperation between regions should be strengthened.
Finally, vegetation plays an important role in air quality. Government can plant more trees in heavily polluted areas, which can significantly increase the surface leaf area and relative humidity, achieve pollutant adsorption and a dilution effect, and effectively reduce air pollution [24].

Conclusions
Using hourly ground-monitoring data from 2014-2018 in Weifang City, this study examined the temporal and spatial characteristics of the AQI in Weifang along with its primary pollutants (O 3 , PM 10 and PM 2.5 ) and analyzed the relationships between the major pollutants of the city and meteorological factors. The major conclusions are as follows.
(a) The results showed that the AQI in Weifang was higher than 100 from 2014-2017, and decreased significantly since 2017. In the last five years, the annual average proportion of days with excellent and good air quality has increased gradually, reaching 60% in 2018, while the number of days with heavy or severe pollution has decreased substantially (declining by 20%). This may be because environmental protection actions were implemented under the eight major initiatives in 2017, which are beneficial for air pollution mitigation. This paper is significant in terms of the in-depth understanding of the spatiotemporal characteristics of air quality and the primary pollutants that affect the AQI in Weifang in recent years. The study provides critical insights into future urban development and pollution control. However, this study has limitations. For instance, the county/district air pollution level is represented by the average of point-monitored concentrations within the area, which may lead to extrapolation errors/bias. Furthermore, air quality is affected by various factors, such as weather and human activities, and there is a lack of quantitative analysis on air quality and other influential factors in this paper. There is a need for future work to combine remote-sensed data and ground-monitored data to control extrapolation errors/bias and to explore relationships between other driving factors and air pollutants for more urban sites.

Conflicts of Interest:
The authors declare no conflict of interest.

AQI
Air Quality Index PM 10 Particles with aerodynamic diameter ≤ 10 µm PM 2. 5 Particles with aerodynamic diameter ≤ 2. Index breakpoint corresponding to C Hi CAAQS Chinese Ambient Air Quality Standards