The Spatial–Temporal Variation of Tropospheric NO 2 over China during 2005 to 2018

: In recent years, new and strict air quality regulations have been implemented in China. Therefore, it is of great signiﬁcance to evaluate the current air pollution situation and e ﬀ ectiveness of actions. In this study, Ozone Monitoring Instrument (OMI) satellite data were used to detect the spatiotemporal characteristics of tropospheric NO 2 columns over China from 2005 to 2018, including spatial distribution, seasonal cycles and long-term trends. The averaged NO 2 pollution is higher in southeastern China and lower in the northwest, which are well delineated by the Heihe–Tengchong line. Furthermore, the NO 2 loadings are highest in the North China Plain, with vertical column density (VCD) exceeding 13 × 10 15 molec cm − 2 . Regarding the seasonal cycle, the NO 2 loadings in eastern China is highest in winter and lowest in summer, while the western region shows the opposite feature. The amplitude of annual range increase gradually from the south to the north. If the entire period of 2005–2018 is taken into account, China has experienced little change in NO 2 . In fact, however, there appears to be signiﬁcant trends of an increase followed by a downward tendency, with the turning point in the year 2012. In the former episode of 2005–2012, increasing trends overwhelm nearly the whole nation, especially in the Jing–Jin–Tang region, Shandong Province, and Northern Henan and Southern Hebei combined regions, where the rising rates were as high as 1.0–1.8 × 10 15 molec cm − 2 year − 1 . In contrast, the latter episode of 2013–2018 features remarkable declines in NO 2 columns over China. Particularly, the regions where the decreased degree was remarkable in 2013–2018 were consistent with the regions where the upward trend was obvious in 2005–2012. Overall, this upward–downward pattern is true for most parts of China. However, some of the largest metropolises, such as Beijing, Shanghai and Guangzhou, witnessed a continuous decrease in the NO 2 amounts, indicating earlier and more stringent measures adopted in these areas. Finally, it can be concluded that China’s recent e ﬀ orts to cut NO 2 pollution are successful, especially in mega cities.


Introduction
Nitrogen oxides (NO x = NO 2 + NO) are important trace gases in the atmosphere that have devastating impacts on the atmospheric environment and human health. In the lower troposphere, nitrogen dioxide (NO 2 ) plays a key role in air quality regulation and aerosols and ozone (O 3 ) formation [1]. NO 2 has a significant effect on atmospheric chemistry. O 3 is a secondary pollutant in the troposphere that is largely determined by the photochemical reactions of NO x and volatile organic compounds (VOC S ). On the other hand, a high level of NO x may likely combine with VOC S and other ultrafine particles and then produce photochemical smog and acid rain, respectively, further providing observational results of various pollutant gases and greenhouse gases around the world and has been widely used to monitor and analyze NO 2 as well as other pollutants over China since 2004 [36][37][38][39]. In particular, the tropospheric NO 2 columns and total NO 2 columns over China during 2004 and 2010 were analyzed by Xiao et al. [40] and it was found that the two parameters presented adverse variations during the same season. The total NO 2 had the largest value in summer, while the tropospheric NO 2 had the smallest value. Wintertime showed the opposite results. Van der A et al. [41] noted that a distinct decreasing trend was visible in 2012. Ai [42], Cai [43] and Li [44] have been comprehensively analyzing the NO 2 concentrations in different regions of China over the last few years.
Recently the Chinese government has developed policies and plans to mitigate the adverse impacts of NO 2 and other greenhouse gases [45]. For example, the 12th five-year plan (2011-2015) set a goal to reduce NO x by 10% from 2011 to 2015 [46]. The 13th five-year plan (2016-2020) continues insisting upon the goal of energy conservation and emission reduction [47]. The Clean Air Alliance of China (CAAC) also released the Air Pollution Prevention and Control Action Plan in 2013 with the aim to improve the national air quality, especially in Beijing-Tianjin-Hebei, the Yangtze River Delta and the Pearl River Delta [48]. However, there are few studies addressing the variations in NO 2 over China after these control measures were implemented. In this study, we examine the changes in NO 2 at the tropospheric level over China between 2005 and 2018 obtained from OMI, including the spatial variations, long-term trends and monthly and seasonal cycles.

OMI NO 2 Vertical Column Density (VCD) Product
The Aura is a sun-synchronous near-polar orbit satellite from the National Aeronautics and Space Administration's (NASA) EOS. OMI is an important sensor onboard the EOS Aura satellite, which was launched on 15 July 2004. OMI is also inherited from GOME and SCIAMACHY and is combined with their advantages. OMI obtains daily measurements, which means the solar irradiance spectrum could be observed once per 24 h on a global scale, and now has provided more than 14 years of NO 2 data products, including total vertical column densities of NO 2 and stratospheric and tropospheric vertical column densities of NO 2 [49,50]. Furthermore, the high spatial resolution (13 km × 24 km) allows for more finer details of atmospheric parameters.
A differential optical absorption spectroscopy (DOAS) method has been validated to effectively retrieve the NO 2 slant column density (SCD) by OMI in the wavelength of 405-465 nm [51][52][53]. The accurate calculation of tropospheric NO 2 vertical column densities (VCD) depends on the SCD and the tropospheric air mass factor (AMF) [54,55]. Many studies have also proved the creditability of OMI NO 2 data products, and its accuracy can be accepted and used in NO 2 researches. For example, Celarier et al. [5] used ground and aircraft-based measurements to validate the OMI tropospheric NO 2 data and other principal quantities. The experimental results showed that the correlations between different instruments was 0.8-0.9. Lamsal et al. [56] compared OMI tropospheric NO 2 data with in situ aircraft, MAX-DOAS and ground-based direct sun Pandora measurements, respectively. OMI NO 2 data showed lower in urban areas and higher in remote regions. However, OMI retrievals generally were consistent with other methods, and the error was less than 20%. Ialongo et al. [57] present the comparison of NASA's OMI standard product (SP) and the Royal Netherlands Meteorological Institute (KNMI) DOMINO product with ground-based observations by the Pandora spectrometer in Helsinki. Results indicated that both satellite-and ground-based showed a similar behavior in weekly and seasonal cycles. These researches all add credibility in using OMI NO 2 data for air quality monitoring and analyzing. In this study, we focus on the study region of 4 • N-60 • N and 70 • E-140 • E, which covers nearly the entire area of China (see Figure 1), and we use the OMI Level-3 Global Gridded Tropospheric NO 2 Data Product (QA4ECV) from 2005 to 2018, compiled by the KNMI [58]. The datasets are processed with a spatial resolution of 0.125 • × 0.125 • , and the cloud radiance is less than 50%.

Meteorological Data Set
Meteorological data were taken from Japanese 55-year Reanalysis (JRA-55) data, archived by the Japan Meteorological Agency. The data included total column precipitable water and temperature at pressure levels during the 14-year period. In addition, monthly means of wind speed data from the European Centre for Medium Range Weather Forecasts (ECMWF) were also used.

Trends Analysis Methods
Two independent statistical methods, namely, the least squares regression and the F-test, were used to analyze the trends over China during different periods. According to the least squares regression, the NO2 trend for each grid cell is assumed a Gaussian data distribution before calculating, and then it obtains the appropriate linear coefficient by linear fitting. In order to increase the robustness of the regression result, the F-test was performed after the calculation. An F-test is a homogeneity of variance test which is suitable for identifying the significance level of the linear trends. More details are given in Section 3.1. A trend is considered to be significant when the confidence level is above 95% for both least squares regression and the F-test [59]. Figure 1 shows the 14-year average distribution of tropospheric NO2 VCD over all of China from 2005 to 2018. Obviously, the NO2 pollution in China is significant, and the distribution of tropospheric NO2 is inhomogeneous. Previous studies have proven that NO2 production is largely influenced by human activities, which means that areas with large population densities and high levels of industrialization may accumulate high concentrations of NO2 [16]. The process of economic development and urbanization in the eastern region is much faster than in the western region in China, which explains why more serious NO2 pollution prevails in the eastern part than in the western area of China. This characteristic can also be divided by the Heihe-Tengchong line (the line begins in the northeast city of Heihe and ends in Tengchong in the southwest). The Heihe-Tengchong line is a boundary that demonstrates the different population densities, as well as geographic and climatic characteristics of the areas [60,61].The southeast area of the line is the most densely populated area, and the area northwest of the line is sparsely populated. The North China Plain is the region

Meteorological Data Set
Meteorological data were taken from Japanese 55-year Reanalysis (JRA-55) data, archived by the Japan Meteorological Agency. The data included total column precipitable water and temperature at pressure levels during the 14-year period. In addition, monthly means of wind speed data from the European Centre for Medium Range Weather Forecasts (ECMWF) were also used.

Trends Analysis Methods
Two independent statistical methods, namely, the least squares regression and the F-test, were used to analyze the trends over China during different periods. According to the least squares regression, the NO 2 trend for each grid cell is assumed a Gaussian data distribution before calculating, and then it obtains the appropriate linear coefficient by linear fitting. In order to increase the robustness of the regression result, the F-test was performed after the calculation. An F-test is a homogeneity of variance test which is suitable for identifying the significance level of the linear trends. More details are given in Section 3.1. A trend is considered to be significant when the confidence level is above 95% for both least squares regression and the F-test [59].

Results and Discussion
3.1. Spatial Distribution and Long-Term Trends of Tropospheric NO 2 at the National Scale Figure 1 shows the 14-year average distribution of tropospheric NO 2 VCD over all of China from 2005 to 2018. Obviously, the NO 2 pollution in China is significant, and the distribution of tropospheric NO 2 is inhomogeneous. Previous studies have proven that NO 2 production is largely influenced by human activities, which means that areas with large population densities and high levels of industrialization may accumulate high concentrations of NO 2 [16]. The process of economic development and urbanization in the eastern region is much faster than in the western region in China, which explains why more serious NO 2 pollution prevails in the eastern part than in the western area of China. This characteristic can also be divided by the Heihe-Tengchong line (the line begins in the northeast city of Heihe and ends in Tengchong in the southwest). The Heihe-Tengchong line is a boundary that demonstrates the different population densities, as well as geographic and climatic characteristics of the areas [60,61]. The southeast area of the line is the most densely populated area, and the area northwest of the line is sparsely populated. The North China Plain is the region over eastern China with the highest NO 2 columns. In addition, the Yangtze River Delta, Pearl River Delta, Sichuan Basin and Urumqi also have high levels of NO 2 columns located on both sides of the Heihe-Tengchong line. In addition, the capital city and nearly all the provincial capital cities, except Lhasa, have stronger NO 2 concentrations compared to the rest areas in each province.
In this study, as Figure 1 indicates, to better understand the changes and characteristics of the 14-year NO 2 concentrations over China in detail, we selected five areas with high NO 2 columns, namely Jing-Jin-Tang (38. As mentioned before, many environmental regulations have been implemented over China during 2005-2018. Since the 11th five-year plan put forward new proposals of emission reduction, more details of measures for air pollution prevention and control have been discussed. Some major regulations and policies during this period are listed in Table 1. As we can see, there are several essential policies promulgated in 2011, and then specifically implemented in 2012. Therefore, to access the effect of several measures and better explain the differences in the spatial variations of tropospheric NO 2 VCD, the year of 2012 was chosen as a turning point to divide the whole period into two stages. The tropospheric NO 2 VCD linear trends during the different stages over China are presented in Figure 2. Based on the OMI NO 2 data, a linear regression line was fit in each grid cell. Two stages were plotted, as shown in Figure 2a,b. Their significance levels were also shown in Figure 3a,b. Notably, an obvious and widespread increasing trend occurs in nearly the whole nation during 2005-2012, especially in the Jing-Jin-Tang region, Shandong Province, and Northern Henan and Southern Hebei combined regions, where the increased values were as high as 1.0-1.8 (×10 15 molec cm −2 year −1 ) (p < 0.05). In addition, the Yangtze River Delta, Chengdu-Chongqing zone, and Urumqi in the Xinjiang Province also showed a slight increase, with values ranging from 0.2 to 1.0 (×10 15 molec cm −2 year −1 ) (p < 0.05). In contrast, megacities such as Beijing, Shanghai and Guangzhou dropped significantly during this period, although their basic amounts of NO 2 column densities were still large. The change in Guangzhou was more prominent than those in the other two cities, with sharp declines of approximately −0.6 to −1.4 (×10 15 molec cm −2 year −1 ) (p < 0.005). This result is analogous to the previous study on the distribution analysis of NO 2 columns by Wang et al. [37,62]. Chinese National Government [46] 2012 Ambient air quality standards (GB3095-2012) The concentration limitation of atmosphere NO x in China has been issued.
Van der A et al. [41] 2013 The Air Pollution Prevention and Control Action Plan Aimed to improve the national air quality, especially in Beijing-Tianjin-Hebei, Yangtze River Delta and Pearl River Delta.
CAAC [48] 2013-2017 APPC-AP PM 2.5 pollution has been paid more attention to. Ma et al. [45] 2016-2020 The 13th five-year plan It insists upon the goal of energy conservation and emission reduction.    Subsequently, the overall linear trend and their significance levels over China during the 14-year period are shown in Figures 2c and 3c. Overall, the NO2 concentrations in megacities such as Beijing, Shanghai and Guangzhou have decreased substantially, especially the Guangzhou in the Pearl River Delta. Although the significance levels are not significant in many areas, confidence levels in these cities are all above 95% (p < 0.05). It means that environmental protection measures have been effectively implemented and have had some effects in these regions. Simultaneously, there was also a significant reduction in the combined Northern Henan and Southern Hebei region. In contrast to these areas, a few cities and regions such as Tangshan and the main cities of Chongqing had an increased trend to a small extent. Most of the remaining regions were maintained in a stable state without remarkable fluctuations during the 14-year study period.

Spatial Distribution and Long-Term Trends of Tropospheric NO2 VCD at the Five Hotspots
The annual variations in the five hotspots present different change patterns in , respectively, as mentioned in Section 3.1. The S. Hebei and N. Henan combined region has a maximum value in the 14-year average of NO2 VCD, followed by Jinan, and the minimum is in the Pearl River Delta, which is consistent with the overall change in NO2 columns for the five main areas over the 14 years (see Figure 4). More specifically, S. Hebei and N. Henan were the regions with the highest NO2 columns until 2010, after which the highest value region swung back and forth between S. Hebei and N. Henan combined and Jinan, but both areas were higher than any other regions. Moreover, the annual changes in the S. Hebei and N. Henan combined region, Jinan and Jing-Jin-Tang are similar during the 14-year study period. The tropospheric NO2 VCD in the three regions increased significantly from 2005 to 2007, decreasing by 7.7%, 3.6% and 18.2%, respectively, in 2008. The reduction is closely related to a series of policies issued by the government during the 2008 Olympic Games, for instance, restrictions were imposed on traffic and emission reductions were expected for industrial activities, especially in Beijing and the surrounding cities [34,63]. Increasing trends with different levels appeared in the three regions during 2008-2011. However, the three regions have all shown a decreasing trend in recent years, Subsequently, the overall linear trend and their significance levels over China during the 14-year period are shown in Figures 2c and 3c. Overall, the NO 2 concentrations in megacities such as Beijing, Shanghai and Guangzhou have decreased substantially, especially the Guangzhou in the Pearl River Delta. Although the significance levels are not significant in many areas, confidence levels in these cities are all above 95% (p < 0.05). It means that environmental protection measures have been effectively implemented and have had some effects in these regions. Simultaneously, there was also a significant reduction in the combined Northern Henan and Southern Hebei region. In contrast to these areas, a few cities and regions such as Tangshan and the main cities of Chongqing had an increased trend to a small extent. Most of the remaining regions were maintained in a stable state without remarkable fluctuations during the 14-year study period.

Spatial Distribution and Long-Term Trends of Tropospheric NO 2 VCD at the Five Hotspots
The annual variations in the five hotspots present different change patterns in Henan combined region has a maximum value in the 14-year average of NO 2 VCD, followed by Jinan, and the minimum is in the Pearl River Delta, which is consistent with the overall change in NO 2 columns for the five main areas over the 14 years (see Figure 4). More specifically, S. Hebei and N. Henan were the regions with the highest NO 2 columns until 2010, after which the highest value region swung back and forth between S. Hebei and N. Henan combined and Jinan, but both areas were higher than any other regions. Moreover, the annual changes in the S. Hebei and N. Henan combined region, Jinan and Jing-Jin-Tang are similar during the 14-year study period. The tropospheric NO 2 VCD in the three regions increased significantly from 2005 to 2007, decreasing by 7.7%, 3.6% and 18.2%, respectively, in 2008. The reduction is closely related to a series of policies issued by the government during the 2008 Olympic Games, for instance, restrictions were imposed on traffic and emission reductions were expected for industrial activities, especially in Beijing and the surrounding cities [34,63]. Increasing trends with different levels appeared in the three regions during 2008-2011.
However, the three regions have all shown a decreasing trend in recent years, where the NO 2 VCD in the S. Hebei and N. Henan combined region, Jinan and Jing-Jin-Tang has decreased by 42.6%, 51.1% and 33.4%, respectively, since 2011. 8 of 18 where the NO2 VCD in the S. Hebei and N. Henan combined region, Jinan and Jing-Jin-Tang has decreased by 42.6%, 51.1% and 33.4%, respectively, since 2011. By comparison, annual variations in the other two regions are relatively moderate, and the NO2 column concentrations are also comparatively smaller. The tropospheric NO2 VCD fluctuated over a small range in the Yangtze River Delta during 2005-2011 (increased by 26.3%), with a peak at 16.87 × 10 15 molec cm −2 in 2011 and a subsequent, steady decline (decreased by 25.1% since 2011). The mean NO2 VCD dropped slowly in the Pearl River Delta during the 14-year period and decreased by 30.1% during this period. In brief, the tropospheric NO2 VCD in the five hotspots, except the Pearl River Delta, experienced a process of first increasing and then decreasing, and this process is not simply monotonous. Table 2 shows the variations in tropospheric NO2 columns over some leading cities in the five areas we selected. The locations of these cities can be found in Figure 2c  In fact, except for the policies enforced in 2008, the government has also proclaimed some measures and plans to prevent or control air pollution. As mentioned before, the 12th five-year plan The mean NO 2 VCD dropped slowly in the Pearl River Delta during the 14-year period and decreased by 30.1% during this period. In brief, the tropospheric NO 2 VCD in the five hotspots, except the Pearl River Delta, experienced a process of first increasing and then decreasing, and this process is not simply monotonous. Table 2 shows the variations in tropospheric NO 2 columns over some leading cities in the five areas we selected. The locations of these cities can be found in Figure 2c . Guangzhou was a more special situation that decreased in these two periods. In addition, Beijing had the lowest concentration of NO 2 among these cities. In fact, except for the policies enforced in 2008, the government has also proclaimed some measures and plans to prevent or control air pollution. As mentioned before, the 12th five-year plan implemented in 2012 and the Air Pollution Prevention and Control Action Plan implemented in 2013 are important Atmosphere 2019, 10, 444 9 of 18 policies for environmental governance in China. The changes in NO 2 amount over China can also reflect the effectiveness of these policies to a certain extent.
Beijing, as the capital of China, is the political and cultural center of China. Beijing is also an area where various environmental protection policies are strictly implemented. Therefore, it is necessary to explore the effects of environmental governance. The detailed variation characteristics of the tropospheric NO 2 VCD in Beijing and Jing-Jin-Tang were analyzed as follows. Figure 5a shows the 14-year averaged NO 2 VCD in Beijing and the Jing-Jin-Tang region (12.58 and 13.46 × 10 15 molec cm −2 , respectively). It is evident that Beijing, Tianjin and Tangshan are major cities in the Jing-Jin-Tang region, all of which are characterized by high concentrations of NO 2 . Furthermore, the linear trend is reflected in Figure 5b. Compared with other areas in the Jing-Jin-Tang region, Beijing showed a significant downward trend during the 14-year study period, approximately −0.042 (×10 15 molec cm −2 year −1 ) after the regional average. However, the Jing-Jin-Tang region had an opposite tendency, approximately 0.060 (×10 15 molec cm −2 year −1 ) after the regional average. This result was mainly ascribed to the other areas in this region, especially Tangshan, which exhibited relatively strong growth trends. Beijing, as the capital of China, is the political and cultural center of China. Beijing is also an area where various environmental protection policies are strictly implemented. Therefore, it is necessary to explore the effects of environmental governance. The detailed variation characteristics of the tropospheric NO2 VCD in Beijing and Jing-Jin-Tang were analyzed as follows. Figure 5a shows the 14-year averaged NO2 VCD in Beijing and the Jing-Jin-Tang region (12.58 and 13.46 × 10 15 molec cm −2 , respectively). It is evident that Beijing, Tianjin and Tangshan are major cities in the Jing-Jin-Tang region, all of which are characterized by high concentrations of NO2. Furthermore, the linear trend is reflected in Figure 5b. Compared with other areas in the Jing-Jin-Tang region, Beijing showed a significant downward trend during the 14-year study period, approximately −0.042 (×10 15 molec cm −2 year −1 ) after the regional average. However, the Jing-Jin-Tang region had an opposite tendency, approximately 0.060 (×10 15 molec cm −2 year −1 ) after the regional average. This result was mainly ascribed to the other areas in this region, especially Tangshan, which exhibited relatively strong growth trends. Monthly variations in NO2 column densities in Beijing and the Jing-Jin-Tang region during the 14-year study period are presented in Figure 6a. First, an outstanding feature is revealed, where the variation in NO2 VCD shows a periodic seasonal cycle, with the highest values in winter and lowest in summer. This feature will be discussed in depth in Section 3.3. Second, the long-term trends in the two regions are described. The NO2 amount in Beijing is close to that of the Jing-Jin-Tang region every month, but the downward trend of the former is more obvious than the latter over the past 14 years. This characteristic is more evident in Figure 6b.
As  Monthly variations in NO 2 column densities in Beijing and the Jing-Jin-Tang region during the 14-year study period are presented in Figure 6a. First, an outstanding feature is revealed, where the variation in NO 2 VCD shows a periodic seasonal cycle, with the highest values in winter and lowest in summer. This feature will be discussed in depth in Section 3.3. Second, the long-term trends in the two regions are described. The NO 2 amount in Beijing is close to that of the Jing-Jin-Tang region every month, but the downward trend of the former is more obvious than the latter over the past 14 years. This characteristic is more evident in Figure 6b.
As   Beijing and its surrounding areas differ greatly in their variations in NO2 columns, which may be due to the following reasons: On the one hand, Beijing has implemented stricter environmental control policies, including motor vehicle management and factory relocations since 2008, which has fundamentally curbed the emissions of NOx. This inference is affirmed by Diao et al. [64]. However, NO2 has a short life, especially in summer, so pollutants cannot be transported over long distances. Accordingly, the distributions of NO2 are associated with emission sources and meteorological conditions [63].

Monthly and Seasonal Patterns of Tropospheric NO2 VCD
The seasonal distributions of NO2 VCD over China over 14 years are shown in Figure 7. The seasonal variation features are noticeable throughout the country, where east China is the highest in winter and lowest in summer. In contrast, west China is the highest in summer and lowest in winter (except Urumqi). This difference is probably due to the differences in the generation of NOx in different regions. Van der A et al. [32] concluded that the main sources of NOx in the western part of China were natural emissions, which were dependent on temperature, soil and precipitation, while the NOx in eastern China was mainly related to human activities such as industrial emissions. In particular, coal-fired heating is a prime reason for large NOx emissions during wintertime. The NOx lifetime is longer in winter than in other seasons, causing severe air pollution at this time. In addition, the areas with high concentrations of NO2 are consistent with the areas of focus in Figure 1. Beijing and its surrounding areas differ greatly in their variations in NO 2 columns, which may be due to the following reasons: On the one hand, Beijing has implemented stricter environmental control policies, including motor vehicle management and factory relocations since 2008, which has fundamentally curbed the emissions of NO x . This inference is affirmed by Diao et al. [64]. However, NO 2 has a short life, especially in summer, so pollutants cannot be transported over long distances. Accordingly, the distributions of NO 2 are associated with emission sources and meteorological conditions [63].

Monthly and Seasonal Patterns of Tropospheric NO 2 VCD
The seasonal distributions of NO 2 VCD over China over 14 years are shown in Figure 7. The seasonal variation features are noticeable throughout the country, where east China is the highest in winter and lowest in summer. In contrast, west China is the highest in summer and lowest in winter (except Urumqi). This difference is probably due to the differences in the generation of NO x in different regions. Van der A et al. [32] concluded that the main sources of NO x in the western part of China were natural emissions, which were dependent on temperature, soil and precipitation, while the NO x in eastern China was mainly related to human activities such as industrial emissions. In particular, coal-fired heating is a prime reason for large NO x emissions during wintertime. The NO x lifetime is longer in winter than in other seasons, causing severe air pollution at this time. In addition, the areas with high concentrations of NO 2 are consistent with the areas of focus in Figure 1. The analyses of time series are based on monthly averages and seasonal averages, which are shown in Figure 8. Monthly averages of NO2 VCD were calculated in five areas during the 14-year period and drawn in Figure 8a. Regardless of the regions we considered, winter (January, February and December) showed higher values than summer (June, July and August), forming a "V" shape in each chart. Moreover, annual range is the most evident in the S. Hebei and N. Henan combined region as well as in Jinan, where maximum minus minimum values exceed 25 (×10 15 molec cm −2 ) in different months. Then, the Jing-Jin-Tang and Yangtze River Delta values are more than 15 (×10 15 molec cm −2 ) regarding the difference between winter and summer. In contrast, the Pearl River Delta has the smallest amplitude in both numerical value and annual cycle, with the highest value in January and lowest value in August along with difference values of less than 5 (×10 15 molec cm −2 ).
In Figure 8b, each chart depicts seasonal comparisons of the five regions during the whole period. The characteristics of long-term seasonal variations were remarkable in winter and autumn but were not obvious in summer and spring among the five specified regions for the 2005-2018 period. In winter, the S. Hebei and N. Henan combined region and Jinan had the maximum values of tropospheric NO2 VCD of the entire period. These regions also showed the most dramatic variability. The temporal patterns in the two regions were similar to those of the analysis we completed in Figure 4. For the Jing-Jin-Tang region, the changes in the NO2 amount during winter showed an increasing trend before 2009 and then a decreasing trend until 2018, except for the abnormal peak in 2010. The reduction rate of tropospheric NO2 VCD between 2011 and 2014 was larger than that between 2014 and 2018. In contrast, the changes in tropospheric NO2 columns in the Yangtze River Delta and Pearl River Delta were slight in winter during the 14-year period, and the NO2 amounts in the former region were all higher than those in the latter period of each year. The variations in tropospheric NO2 columns in autumn during the 14-year period were somewhat similar to those in winter. However, the NO2 levels were smaller than those in winter. The analyses of time series are based on monthly averages and seasonal averages, which are shown in Figure 8. Monthly averages of NO 2 VCD were calculated in five areas during the 14-year period and drawn in Figure 8a. Regardless of the regions we considered, winter (January, February and December) showed higher values than summer (June, July and August), forming a "V" shape in each chart. Moreover, annual range is the most evident in the S. Hebei and N. Henan combined region as well as in Jinan, where maximum minus minimum values exceed 25 (×10 15 molec cm −2 ) in different months. Then, the Jing-Jin-Tang and Yangtze River Delta values are more than 15 (×10 15 molec cm −2 ) regarding the difference between winter and summer. In contrast, the Pearl River Delta has the smallest amplitude in both numerical value and annual cycle, with the highest value in January and lowest value in August along with difference values of less than 5 (×10 15 molec cm −2 ).
In Figure 8b, each chart depicts seasonal comparisons of the five regions during the whole period. The characteristics of long-term seasonal variations were remarkable in winter and autumn but were not obvious in summer and spring among the five specified regions for the 2005-2018 period. In winter, the S. Hebei and N. Henan combined region and Jinan had the maximum values of tropospheric NO 2 VCD of the entire period. These regions also showed the most dramatic variability. The temporal patterns in the two regions were similar to those of the analysis we completed in Figure 4. For the Jing-Jin-Tang region, the changes in the NO 2 amount during winter showed an increasing trend before 2009 and then a decreasing trend until 2018, except for the abnormal peak in 2010. The reduction rate of tropospheric NO 2 VCD between 2011 and 2014 was larger than that between 2014 and 2018. In contrast, the changes in tropospheric NO 2 columns in the Yangtze River Delta and Pearl River Delta were slight in winter during the 14-year period, and the NO 2 amounts in the former region were all higher than those in the latter period of each year. The variations in tropospheric NO 2 columns in autumn during the 14-year period were somewhat similar to those in winter. However, the NO 2 levels were smaller than those in winter. Similarly, a more detailed comparison and analysis of the seasonal variations between Beijing and Jing-Jin-Tang was carried out. As clearly shown in Figure 9a, conspicuous differences in the NO2 amount between Beijing and Jing-Jin-Tang occurred in January, February and December, and the difference values were approximately 3.06, 6.94 and 2.56 (×10 15 molec cm −2 ), respectively. During other months, there were only subtle differences in NO2 amounts between the two regions. This point is more evident in Figure 9b. Compared with other seasons, the long-term changes in the NO2 columns in winter were the most significant; an intense contrast between the two regions was also apparent. In the past 14 years, the values of the NO2 columns in Beijing were basically smaller than those in Jing-Jin-Tang. Furthermore, the NO2 concentrations in both places have steadily decreased during winter since 2012. With the implementation of various measures, NO2 pollution has been more effectively controlled in Beijing than in the surrounding areas. In recent years, Beijing has taken more stringent measures to control air pollution during the winter heating period. The government has focused on key emission sources during autumn and winter, and several measures were proposed, including promoting the adjustment of transportation structure and encouraging investments in new high-technology industries such as clean energy. Similarly, a more detailed comparison and analysis of the seasonal variations between Beijing and Jing-Jin-Tang was carried out. As clearly shown in Figure 9a, conspicuous differences in the NO 2 amount between Beijing and Jing-Jin-Tang occurred in January, February and December, and the difference values were approximately 3.06, 6.94 and 2.56 (×10 15 molec cm −2 ), respectively. During other months, there were only subtle differences in NO 2 amounts between the two regions. This point is more evident in Figure 9b. Compared with other seasons, the long-term changes in the NO 2 columns in winter were the most significant; an intense contrast between the two regions was also apparent. In the past 14 years, the values of the NO 2 columns in Beijing were basically smaller than those in Jing-Jin-Tang. Furthermore, the NO 2 concentrations in both places have steadily decreased during winter since 2012. With the implementation of various measures, NO 2 pollution has been more effectively controlled in Beijing than in the surrounding areas. In recent years, Beijing has taken more stringent measures to control air pollution during the winter heating period. The government has focused on key emission sources during autumn and winter, and several measures were proposed, including promoting the adjustment of transportation structure and encouraging investments in new high-technology industries such as clean energy.

Impacts of Meteorological Conditions
Generally, three main factors need to be considered to explain the NO2 changes in a specific region, including the amount of NOx emission, NOx lifetime and the NO2 transportation among diffident regions [62]. As discussed in Section 3.3, NO2 variation has a significant seasonal characteristic, which is larger in winter and lower in summer. First, from the perspective of NOx emission, large increase in fossil fuel consumption is a prime reason due to large NOx emissions during wintertime. Then in winter, the lifetime of NOx is also larger than other seasons because of the weaker solar radiation and lower atmospheric temperature. All these factors delay the process of atmospheric chemical reactions. In contrast, the solar radiation is stronger in summer and chemical reaction is active, which are beneficial to NO2 removal. Finally, NO2 transportation is related to meteorological conditions. In winter, although the average wind speed is higher, the NO2 emissions are also larger. In addition, the atmospheric boundary layer is low in winter. It is easy to form an inversion layer, which leads NO2 remaining in the lower troposphere.
To further analyze the different monthly variations in different regions, we decided to investigate the meteorological conditions in these five areas. Figure 10 illustrates the annual cycle of precipitable water, wind speed and temperature at the five hotspots. Combined with the monthly average variations of NO2 in Figure 8a, it is obvious that NO2 monthly variation is associated with the seasonal change of meteorological conditions. Temperature and precipitable water are colder and less in winter which could weaken the rate of oxidation and wet deposition. In addition, compared with Jing-Jin-Tang, the S. Hebei and N. Henan combined region and Jinan, the Yangtze River Delta and the Pearl River Delta show distinct meteorological conditions. Both two regions are located at low latitudes in the Southern China and near the ocean, with average temperatures higher than other three areas. Moreover, with monsoon influence, there are more precipitation in the two regions, which are all beneficial to NO2 eliminated. This result is also consistent with the seasonal variations of NO2 concentration in the five hotspots.

Impacts of Meteorological Conditions
Generally, three main factors need to be considered to explain the NO 2 changes in a specific region, including the amount of NO x emission, NO x lifetime and the NO 2 transportation among diffident regions [62]. As discussed in Section 3.3, NO 2 variation has a significant seasonal characteristic, which is larger in winter and lower in summer. First, from the perspective of NO x emission, large increase in fossil fuel consumption is a prime reason due to large NO x emissions during wintertime. Then in winter, the lifetime of NO x is also larger than other seasons because of the weaker solar radiation and lower atmospheric temperature. All these factors delay the process of atmospheric chemical reactions. In contrast, the solar radiation is stronger in summer and chemical reaction is active, which are beneficial to NO 2 removal. Finally, NO 2 transportation is related to meteorological conditions. In winter, although the average wind speed is higher, the NO 2 emissions are also larger. In addition, the atmospheric boundary layer is low in winter. It is easy to form an inversion layer, which leads NO 2 remaining in the lower troposphere.
To further analyze the different monthly variations in different regions, we decided to investigate the meteorological conditions in these five areas. Figure 10 illustrates the annual cycle of precipitable water, wind speed and temperature at the five hotspots. Combined with the monthly average variations of NO 2 in Figure 8a, it is obvious that NO 2 monthly variation is associated with the seasonal change of meteorological conditions. Temperature and precipitable water are colder and less in winter which could weaken the rate of oxidation and wet deposition. In addition, compared with Jing-Jin-Tang, the S. Hebei and N. Henan combined region and Jinan, the Yangtze River Delta and the Pearl River Delta show distinct meteorological conditions. Both two regions are located at low latitudes in the Southern China and near the ocean, with average temperatures higher than other three areas. Moreover, with monsoon influence, there are more precipitation in the two regions, which are all beneficial to NO 2 eliminated. This result is also consistent with the seasonal variations of NO 2 concentration in the five hotspots.

Conclusions
In recent years, some new measures that can improve and prevent air contamination have been implemented in China. It is of great significance to understand the current air pollution situation by exploring the changes in NO2 concentrations. Furthermore, these new measures can also evaluate the beneficial effects of pollution policies and provide some reference for the next steps in air pollution control. Therefore, the results of this study describe the characteristics of tropospheric NO2 in terms of spatial distributions and temporal variations from 2005 to 2018 over China using OMI satellite data. The main conclusions are summarized as follows: (1) The NO2 pollution in China is significant, and the distribution of tropospheric NO2 VCD is uneven. This pollution is higher in southeastern China and lower in the northwest, which is well divided by the Heihe-Tengchong line. In addition, the North China Plain is the region over eastern China with the highest NO2 columns. Then, five areas with high NO2 columns in southeast China were selected for detailed discussion, including Jing-Jin-Tang, the combined regions of Northern Henan and Southern Hebei, Jinan in the Shandong Province, Yangtze River Delta and Pearl River Delta.
(2) NO2 concentrations show long-term variability that varies regionally. On the one hand, the concentration of NO2 in the Pearl River Delta has been in a slow but steady decline apart from slight increases in 2007 and 2010. On the other hand, the annual variation in the NO2 columns displays two different stages in the other four areas. First, the NO2 amounts in these four regions increased significantly until 2011 or 2012, except in 2008, due to the strict measures taken for the Olympic Games. Subsequently, the growth of NO2 VCD has slowed down in recent years. Moreover, some of the largest metropolises in China, such as Beijing, Shanghai and Guangzhou, show a downward trend in the NO2 amounts during the 14-year study period.

Conclusions
In recent years, some new measures that can improve and prevent air contamination have been implemented in China. It is of great significance to understand the current air pollution situation by exploring the changes in NO 2 concentrations. Furthermore, these new measures can also evaluate the beneficial effects of pollution policies and provide some reference for the next steps in air pollution control. Therefore, the results of this study describe the characteristics of tropospheric NO 2 in terms of spatial distributions and temporal variations from 2005 to 2018 over China using OMI satellite data. The main conclusions are summarized as follows: (1) The NO 2 pollution in China is significant, and the distribution of tropospheric NO 2 VCD is uneven. This pollution is higher in southeastern China and lower in the northwest, which is well divided by the Heihe-Tengchong line. In addition, the North China Plain is the region over eastern China with the highest NO 2 columns. Then, five areas with high NO 2 columns in southeast China were selected for detailed discussion, including Jing-Jin-Tang, the combined regions of Northern Henan and Southern Hebei, Jinan in the Shandong Province, Yangtze River Delta and Pearl River Delta. (2) NO 2 concentrations show long-term variability that varies regionally. On the one hand, the concentration of NO 2 in the Pearl River Delta has been in a slow but steady decline apart from slight increases in 2007 and 2010. On the other hand, the annual variation in the NO 2 columns displays two different stages in the other four areas. First, the NO 2 amounts in these four regions increased significantly until 2011 or 2012, except in 2008, due to the strict measures taken for the Olympic Games. Subsequently, the growth of NO 2 VCD has slowed down in recent years.
Moreover, some of the largest metropolises in China, such as Beijing, Shanghai and Guangzhou, show a downward trend in the NO 2 amounts during the 14-year study period. (3) The characteristics of the seasonal cycle are obvious in China. The NO 2 amount in eastern China is the highest in winter and lowest in summer, while the western region (except Urumqi) shows the opposite feature. The features of monthly and quarterly variations also vary depending on the region under consideration. The monthly and quarterly variations in NO 2 in the Pearl River Delta are relatively moderate. In contrast, the change in the S. Hebei and N. Henan combined region and Jinan are clearly primarily due to excessive coal heating during wintertime. In addition, meteorological conditions have significant impacts influenced in different regions, especially in the Yangtze River Delta and Pearl River Delta. The higher temperature, larger amount of precipitable water and faster wind speed are all beneficial to NO 2 removal. (4) In particular, Beijing and its surrounding areas show different characteristics of NO 2 variations.
Beijing showed a significant downward trend during the 14-year study period. However, the Jing-Jin-Tang region showed the opposite tendency. For seasonal variations, conspicuous differences in NO 2 amount between Beijing and Jing-Jin-Tang occur in winter (January, February and December), and the values of the NO 2 columns in Beijing are basically smaller than those in Jing-Jin-Tang. The reasons for the different results are mainly because Beijing has implemented stricter environmental control policies.
Through the analysis of the spatial temporal variations in NO 2 VCD during the 14 studied years, the effectiveness of several environmental protection policies implemented in China are seen. In the future, the factors affecting the changes in NO 2 in different areas will be investigated in greater depth. Furthermore, specific control measures should be put forward according to different regions and different factors.
Author Contributions: C.W., T.W. and P.W. conceived and designed the experiments; C.W. collected and processed data; C.W. and T.W. analyzed and wrote the paper; and T.W. and P.W. contributed revisions to the paper.