Increases of nitrogen oxides (NO_x) in the troposphere due to rapid economic development and industrialization have been observed by satellite and ground-based observations in recent decades over eastern China, exacerbating the risk to human health and ecosystems (Richter et al., 2005; Larssen et al., 2006; van der A et al., 2006, 2008; Fu et al., 2007; Chan and Yao, 2008). Eastern China has become one of the most polluted areas in the world with respect to NO_x pollution. Among China’s major anthropogenic NO_x sources, such as the coal-burning thermal power industry, heavy industry, and road transport, the thermal power industry is the largest emitting source, consuming more than half of national yearly coal and emitting about 49.9% of total anthropogenic NO_x emissions, according to China’s annual statistics report on the environment in 2008 (Luo, 2009; Li and Leung, 2012; Lin, 2012). Based on figures in the China statistical yearbook in 2009, about 85% of coal-burning thermal power plants have been built in eastern and central parts of China to meet local electricity demand, which has arisen from rapid industrialization and urbanization, and this region has the greatest concentration of NO_x emissions in the country. Road transport is the second-largest NO_x source in China. However, its NO_x emissions only accounted for slightly more than one-third of thermal power emissions in 2008; specifically, 17.4% of total NO_x emissions (Luo, 2009). Power plants, industry, and transport were major sources of NO_x emissions in China in 2010, accounting for 28.4%, 34.0%, and 25.4% of the total, respectively (Zhao et al., 2013). As a result of the increase in NO_x emissions from energy consumption and transport, acid rain in China has changed from sulfuric acid to a mixture of sulfuric and nitric acids, with the nitrate portion having increased from one-tenth in the 1980s to one-third in the 2000s (Hao et al., 2002; Streets et al., 2003; Luo, 2009). The adverse effect of the rapidly increasing NO_x emissions cannot be fully offset by the slightly decreasing sulfur dioxide emissions in China, leading to exacerbation of the acid rain problem. This has prompted the Chinese government to develop a scheme to control NO_x emissions and pollution during the period of the 12th Five-Year Plan (2011–15) (Wu X. Q., 2009). NO_x emissions have been listed by the Chinese government among the pollutants that need to be controlled, with strict controls on NO_x emissions by the thermal power industry having been set. Therefore, it is necessary to quantitatively assess the NO_x pollution from this industrial sector in China.

Satellite-based observations of tropospheric NO₂ have proven useful in estimating NO_x anthropogenic emissions, making trend analyses, setting control measures, and evaluating the skill of air quality forecast models. Using data from the Global Ozone Monitoring Experiment (GOME), Jaeglé et al. (2004) identified spatial and seasonal variations in NO_x that were mainly caused by biomass burning and soil emissions over the Sahel. Satelliteretrieved summertime NO₂ column densities show large decreases in the Ohio River valley, where power plants dominate NO_x emissions (Kim et al., 2006). Tropospheric NO₂ measurements from GOME and SCIAMACHY (Scanning Imaging Absorption Spectrometer for Atmospheric Chartography) indicate a large growth of NO₂ densities over eastern China—one of the most polluted areas of tropospheric NO₂ in the world. These areas showed rapid growth in tropospheric NO₂ densities (about 20% per year) in the last two decades (Richter et al., 2005; van der A et al., 2006, 2008; He et al., 2007; Xing et al., 2011; Zhang et al., 2012). The unique capabilities of the ozone monitoring instrument (OMI) for air-quality monitoring are evident in that, with its near-real-time, full global mapping and long-term continuous observations, it can directly observe the so-called weekend effect in tropospheric NO₂ variations. Indeed, it has been applied to examine the instantaneous effectiveness of air-quality measures (Beirle et al., 2003; Boersma et al., 2009). For example, following large-scale restrictions in vehicular traffic in Beijing during the Sino–African Summit in 2006, reductions in associated emissions of NO₂ of about 40% were detected by the OMI aboard the Aura satellite (Wang et al., 2007). As a result of the effect of the air-quality control measures for the Beijing 2008 Olympic Games, a reduction in tropospheric NO₂ of about 40%–60% over the Beijing area and 20%–30% in surrounding cities was observed from July to August 2008, using OMI tropospheric NO₂ column densities combined with a regional chemistry transport model (Mijling et al., 2009; Yu et al., 2010). The accuracy of the air-quality forecast model was evaluated by comparing the OMI NO₂ data with that in a study by Herron-Thorpe et al. (2010). Of course, satellite observations of tropospheric NO₂ are quite uncertain, especially in the winter.

The Yangtze River Delta (YRD), one of the main cultural and economic centers of China, has undergone rapid industrialization and urbanization, and is heavily dependent on coal to sustain its rapid economic growth. Its coal consumption, mainly from 275 local thermal power plants with 5.24×10⁹ kW power-generation capacity, accounted for 14.89% of total national coal consumption in 2008, according to the China city statistical yearbook and annual statistic report on the environment in China. The YRD’s NO_x emissions and densities have undergone rapid growth since 1996, as evident from emissions inventory and satellite observations, resulting in severe local NO_x pollution (Richter et al., 2005; Zhang Q. et al., 2007; Zhang X. Y. et al., 2007; van der A et al., 2008; Wu X. L., 2009). It has been established that NO₂ in the troposphere has a relatively short lifetime, as well as a correspondingly high spatial and temporal variability; its densities exhibit a large spatial inhomogeneous distribution in the troposphere, with high densities being close to high local emissions (Spichtinger et al., 2001; Fishman et al., 2008; Zhou et al., 2009; Kar et al., 2010). When NO_x emissions from local industrial sources fluctuate, tropospheric NO₂ densities may respond (Stavrakou et al., 2008). The 12-month-averaged NO₂ densities in northern East China increased by 27%–33% prior to the economic downturn, compared with the 49% increase in thermal power generation. Emissions decreased by 20% from January 2008 to January 2009, consistent with the 18% decrease in thermal power generation. The decline in emissions of NO_x from January 2008 to January 2009 was found to be a consequence of both the economic downturn and a decrease in industrial activity (Lin and McElroy, 2011; Lin et al., 2014). Large amounts of emissions were embodied in the imports of eastern regions from northern and central regions, and these were determined by differences in regional economic status and environmental policy (Zhao et al., 2015). The YRD experienced a significant industrial and economic downturn in the first quarter of 2009 as a result of the global financial crisis: industrial output decreased by 5.7% compared with the first quarter of 2008 (National Bureau of Statistics of People’s Republic of China, 2009). Local thermal generating capacity correspondingly decreased by 11.98% over the same period. Some key cities, e.g., Nanjing, Shanghai, Hangzhou, and Ningbo, which have high NO_x emissions and NO₂ densities, suffered a greater decline in industrial output of more than 10%. The global financial crisis provides an excellent opportunity to assess the effects of industrial downturn on NO_x pollution in the YRD, especially fluctuations in the thermal power industry. In addition, the merged aerosol optical depth over the YRD showed growth until 2008, followed by a sharp decline in 2009 (Cheng et al., 2013; Boys et al., 2014; Gong et al., 2014; Lin and Li, 2016).

In this study, we began by addressing whether any possible changes in NO_x pollution in the YRD due to the financial crisis can be detected in existing OMI NO₂ products. Then, taking advantage of the OMI’s high spatial resolution, high-quality data products, and instantaneous observations, its monthly NO₂ column products were used to analyze the spatiotemporal distribution of NO₂ pollution before and after the financial crisis, and quantitatively assess the effects of thermal power-generation capacity fluctuations on local NO₂ pollution in the YRD. Finally, the need for change in energy consumption, electricity generation, and industrial structure was discussed.

OMI tropospheric monthly averaged NO₂ column density data, with a spatial resolution of 0.125°×0.125°, were obtained from , provided by the Royal Netherlands Meteorological Institute (KNMI).

Monthly statistical data on thermal generation and other major industrial products were obtained from the statis-tical yearbooks of major cities in the YRD.

Ground-based NO₂ densities, observed in Shanghai, Cixi, and Linan from environmental protection departments, were used to assess temporal variations in ground NO₂ densities.

In KNMI’s DOMINO ( Derivation of Ozone Monitoring Instrument tropospheric NO₂ in near-real time) NO₂ products, differential optical absorption spectroscopy (DOAS-type) retrievals are very sensitive to errors in cloud parameters. The uncertainty due to cloud fraction has been estimated to be as much as 30% for strongly polluted regions (Boersma et al., 2007). To ensure retrieval accuracy, the monthly mean tropospheric NO₂ column data from OMI at KNMI were obtained by the arithmetic average on cloud-free days: pixels with a cloud radiance fraction that does not exceed 50%. As is evident in Fig. 1b, in some months over the YRD, less than 50% of the total number of days were cloud-free; but, this was not the case in the referencing region (34°–40°N, 112°–121°E). The monthly mean NO₂ densi-ties over these two regions showed high correlation: their correlation coefficient reached 0.95. Therefore, the uncertainty in the NO₂ concentration over the YRD for months when less than 50% of the total number of days were cloud-free could be reduced by using the data series in the referencing region.

Figure 1 Time series of (a) monthly averaged NO2 column densities (1014 molec. cm–2) in the YRD and (b) monthly cloud cover (%) in the YRD and referencing region.

Figure options

The results are shown in Fig. 1a. A fluctuation in tropospheric NO₂ column densities was clearly evident from OMI’s DOMINO products during the financial crisis period of January 2008 to January 2010. In addition, evident from Fig. 1a, tropospheric NO₂ column densities had a stronger seasonal cycle along with the seasonal changes in air temperature. Anomaly time series were obtained from the differences between monthly NO₂ density and the corresponding monthly average over several years to remove the annual cycle.

Under the influence of economic fluctuations, structural changes resulting from the financial crisis could occur in the time series of industrial output, industrial products, power generation, and pollutant densities in the YRD. Break points can be detected by the Chow test. The Chow test is an established statistical approach for objectively detecting variations or structural changes in a time series. To detect all break points, an optimal piecewise linear modeling approach with the Chow test, presented by Gao et al. (1997), was applied to find all time series with structural changes. The extension of the Chow test with the standard linear regression model is given in the following equation:

where n is the length of the time series, k₁, k₂, …, k_{m –1} are structural change points in the intervals (k₁ –k₂, …, k_{m –2} –k_{m -1}), m–1 is the number of structural change points, and a₁₁, a₂₁, …, a_{m –1} are the linear rates in the respective interval periods.

Generally, an anomaly time series of over 1–3 times the standard deviation is defined as a statistically significant fluctuation. In this study, statistically significant fluctuations in the time series of industrial output, industrial products, power generation, and NO₂ densities in the YRD, presented an anomaly that was over 1.5 times the standard deviation. This method was used to identify the period of the financial crisis for all time series with structural changes.

Table 1 shows that the thermal power industry had the largest NO_x emissions in the YRD in 2008, accounting for 47.82% of total emissions; the road-transport sector had the second-largest NO_x emissions, accounting for 21.37%. The largest quantity of NO_x emissions from the thermal power industry was more than twice that from the road-transport sector, showing that the thermal power industry was the main source of NO_x emissions in the YRD. The central part of the YRD, consisting of Shanghai, Suzhou, Wuxi, Ningbo, and Nanjing, had the highest NO_x emissions; annual NO_x emissions were over 10 million tons for four of these cities, and the total NO_x emissions amounted to 132.93 million tons, accounting for 72.13% of the whole regional NO_x emissions in 2008. NO_x emissions from the thermal power industry accounted for 35.06% of total regional emissions. Large thermal power plants with an installed capacity of more than 50 MW, which are mainly concentrated in this part of the YRD, accounted for 25% of the total amount of thermal power plants in the YRD (Cheng et al., 2009), which emit most of the local NO_x emissions. It can be seen that only about 1.09×10⁵ tons of NO_x were removed in the YRD, amounting to 5.91% of total emissions. Since the power industry produces elevated levels of NO_x, these emissions extend over a larger area than those of lower NO_x emission sources, such as dense traffic and urban living. When fluctuations in thermal generating capacity occur, NO_x pollution may fluctuate correspondingly.

Analysis of OMI’s NO₂ density variations showed that the monthly NO₂ density increased by 1.62×10¹⁵ molec. cm^–2 from October 2004 to September 2010 in the YRD. The largest increase of about 6.68×10¹⁵ molec. cm^–2 occurred in winter. The high NO₂-polluted areas as observed by OMI generally corresponded to large NO_x emissions around the big cities and industrial zones in the YRD. The Shanghai metropolitan area, consisting of Wuxi, Suzhou, and Shanghai along the Yangtze River and situated in the central YRD, had the highest averaged NO₂ tropospheric densities of more than about 2.0×10¹⁵ molec. cm^–2 from October 2004 to September 2010 (Fig. 2a).

Figure 2 (a) Spatial distribution of the mean NO2 density from October 2004 to September 2010 and (b) variation of monthly NO2 density over 10 equal area parts with ascending concentrations (shaded; 1014 molec. cm–2) and the regional monthly density (black line) in the YRD from OMI.

Figure options

The regional monthly mean tropospheric NO₂ density (black line in Fig. 2b) underwent three periods of variation. The first period, from October 2004 to October 2008, showed a rapid increase in tropospheric NO₂ densi-ty; the largest increment of about 5.09×10¹⁵ molec. cm^–2 occurred in winter from 2004 to 2008. The second period (from November 2008 to February 2009) showed a significant decrease in tropospheric NO₂ density of about 3.71×10¹⁵ molec. cm^–2 than the averaged density from November 2007 to February 2008. However, NO₂ densities rapidly increased again in the third period, from March 2009 to September 2010; the NO₂ density increased to 2.12×10¹⁶ molec. cm^–2 in winter 2009, which was over 5.30×10¹⁵ molec. cm^–2 greater than in winter 2008. Furthermore, with the highest tropospheric NO₂ densities, which were less than 30% in Fig. 2b, the same three periods were also evident, the most significant fluctuations also occurred in winter 2008, which was at the height of the financial crisis.

Figures 3a and 3b compare the NO₂ densities from ground-based and OMI observations in cities (Cixi), industrial zones (Shanghai), and rural areas (Linan). Shanghai, the largest central industrial city in the YRD, showed a reduced trend in ground monthly averaged NO₂ density from October 2004 to September 2010, decreasing by 14.13 μg m^–3. However, no significant fluctuations were evident in the ground-based NO₂ density in Shanghai during the financial crisis (Fig. 3a). In Cixi, a smaller city in the YRD that underwent rapid industrialization, the ground monthly mean NO₂ density increased by 11.41 μg m^–3 from January 2005 to May 2010. In line with the variations in ground NO₂ densities in Shanghai, no significant fluctuations occurred during the period of the financial crisis. The trend of ground monthly mean NO₂ densities at the rural station was the same as in Cixi, but the rate of increase was lower. It can be concluded that there was no significant fluctuation in ground-based NO₂ densities in the YRD, where lower emissions sources dominate, e.g., dense traffic and urban living, during the period of the financial crisis. In contrast, from OMI’s observations, significant fluctuations occurred in tropospheric NO₂ densities over Cixi, Shanghai, and Linan, which have different levels of industrial development, during the height of the financial crisis in winter 2008 (Fig. 3b). In rural Linan, the tropospheric NO₂ density in winter 2008 was 43.88% and 33.7% lower than in winters 2007 and 2009, respectively. It can be concluded that fluctuations in NO₂ densities during the financial crisis mainly occurred in the troposphere over all the cities, but not at ground level.

Figure 3 Temporal variation of standardized NO2 densities in the YRD from (a) ground-based observations ( μg m–3) and (b) OMI (×1015 molec. cm–2).

Figure options

The detection of local NO_x pollution fluctuations due to the financial crisis is linked to several factors, such as the determination of the fluctuating period, the magnitude of the NO₂ densities, consistency with the period of the financial crisis, and local NO₂ density fluctuation. Three major industrial sectors, consisting of thermal power, smelting and pressing of ferrous metals, and petroleum processing, have the largest NO_x emissions among all industrial sectors in the YRD (Cheng et al., 2009; Wu X. L., 2009), and they were used to represent the fluctuation in industrial production. The results showed that monthly industrial products of these three industrial sectors experienced significant fluctuations from June 2007 to December 2009, according to the Chow test (Fig. 4a). Thermal power growth showed two structural breakpoints in July 2007 and November 2008. With smelting metal, two structural breakpoints occurred in September 2007 and August 2008; and a structural breakpoint occurred in January 2006 for petroleum processing. Correspondingly, the structural breakpoints appeared from May to November 2008 for tropospheric NO₂ densities, industrial power consumption, and industrial added value (Fig. 4b).

Figure 4 (a) Variations of industrial output in three major industrial sectors and (b) regional averaged NO2 density, power consumption, and industrial added value in the YRD. The large spots in the time series represent where structural changes occurred. All time series are standardized.

Figure options

Parts of the structural breakpoints of each time series among mainly industrial products, industrial added value, thermal power, power consumption, and tropospheric NO₂ densities showed a distinct V-shaped fluctuation from May 2007 to December 2009, indicating that the financial crisis had a tremendous impact in terms of industrial production, power generation, energy consumption, and tropospheric NO_x pollution. Deviations in all time series were more than 1.5 times the standard deviation from November 2008 to February 2009, suggesting that the fluctuation was statistically significant. On the basis of fluctuations for all time series, November 2008 to February 2009 was defined as the worst period in the financial crisis. Major industrial products in this period decreased by 3.27%–12.10% over the same period of the previous year. When thermal power showed its maximum decrease of 12.10%, it led to tropospheric NO₂ density decreasing by 16.97% over the same period of the previous year. When thermal power capacity increased by 29.63% in the same period of the following year, when the effect of the financial crisis was weaker, tropospheric NO₂ density correspondingly increased by 30.07%. Therefore, the financial crisis from November 2008 to February 2009, which caused the industrial slowdown and a decrease in thermal power, significantly affected regional NO₂ pollution in the YRD.

In the YRD, 77.25% of the electricity was produced by local coal-burning thermal plants in 2008. In five cities whose NO_x emissions amounted to more than 0.1 million tons in 2008, consumption of coal was 161.95 million tons, accounting for 58.69% of the total coal consumption in the YRD. Their coal consumption was concentrated in the thermal power sector, accounting for 65.96% of that in all industrial sectors in 2008, as seen in Table 2. A comparison of industrial output, coal consumption, and electricity consumption for each industrial sector in the cities’ statistical yearbooks shows that the thermal power sector has the greatest coal consumption; the smelting and pressing of ferrous metal sector and the sector of manufacturing communication equipment, computers, and other electronic equipment have the greatest industrial output of about 8.84% and 20.98% and the largest electricity consumption of about 19.54% and 12.5%, respectively. In the YRD, electricity is mainly obtained from local coal-burning thermal plants without NO_x-processing equipment, and the coal-based energy structure will not change in the short term. At present, installing NO_x-removal devices in thermal power plants is an important, feasible route in controlling regional NO_x emissions and pollution. In the long run, it will be necessary to change the regional structure of electricity production and the industrial sector and develop clean alternative energy sources to provide a fundamental solution to local NO_x pollution.

The industrial sector accounted for 63.09% of total NO_x emissions in 2008, in which the thermal power industry had the largest NO_x emissions of about 47.82% (Table 1). The financial crisis had a tremendous impact in terms of industrial production, power generation, and energy consumption, and induced a distinct V-shaped fluctuation from May 2007 to December 2009 (Fig. 3a). When there were fluctuations in industrial production, NO_x pollution fluctuated in a corresponding manner (Fig. 3a). Based on the yearly outputs of industrial sectors and thermal power and their NO_x emissions during the period 2005–10, we obtained the NO_x emission intensity of industrial and thermal power unit output for each year. Finally, monthly NO_x emissions could be generated from the monthly industrial and thermal power outputs and their unit emissions. The monthly variation of NO_x emissions is demonstrated in Figs. 5a, c, in which we can see notably different variations of NO_x emissions and monthly NO₂ densities. Furthermore, removing the seasonal cycle, the variation of tropospheric NO₂ densities was similar to the NO_x emissions of industrial sectors and thermal power plants (Figs. 5b, d), with correlation coefficients of about 0.54 and 0.52 (95% confidence level). In particular, distinct V-shaped fluctuations of NO_x emissions and densities occurred from May 2007 to December 2009. When NO_x emissions from local industrial sources fluctuated from November 2008 to February 2009, tropospheric NO₂ densities responded to the economic downturn, decreasing by 21.31% compared with that from November 2007 to February 2008, and by 31.91%, compared with that from November 2009 to February 2010. These declines were greater than the approximate 20% decrement of NO_x emissions due to the economic downturn and weakened industrial activity in eastern China calculated in a previous study (Lin and McElroy, 2011). The tropospheric NO₂ concentration decreased by about 36.71% than the density before the financial crisis and 16.55%–35.28% than the density after the financial crisis. The YRD experienced a significant downturn in industrial production in the first quarter of 2009 as a result of the global financial crisis: industrial output decreased by 5.7% compared with the first quarter of 2008 (National Bureau of Statistics of People’s Republic of China, 2009). In short, the decline in industrial output and thermal power by about 10% induced a decrease of about 30% in NO_x emissions and NO₂ densities, indicating that a relatively small fluctuation in industrial production can cause a large decrease in tropospheric NO₂ densities over industrially developed areas like the YRD region. Since electricity is mainly generated by local coal-burning thermal plants, which emit the largest NO_x emissions and have no NO_x-processing equipment, installing NO_x-removal devices for all thermal power plants is an important and feasible way of controlling local NO_x pollution.

Figure 5 Standardized NOx emission variation of (a) industrial sectors and (c) thermal power plants and regional mean tropospheric NO2 concentration variations in the YRD. (b) and (d) As in (a) and (c), but without the seasonal cycle.

Figure options

(1) The financial crisis exerted a tremendous impact on all industrial production, power generation, and energy consumption. During the financial crisis, fluctuations in NO₂ pollution were also evident using the NO₂ column products derived from OMI. Furthermore, parts of the structural breakpoints of each time series among major industrial products, industrial added value, thermal power, power consumption, and tropospheric NO₂ densities presented a distinct V-shaped fluctuation from May 2007 to December 2009. All time series showed their greatest fluctuation of more than 1.5 times the standard deviation at the height of the financial crisis, from November 2008 to February 2009.

(2) The YRD experienced significant fluctuations in tropospheric NO₂ densities in response to the industrial downturn in the financial crisis period. Thermal power fluctuations were mainly responsible for the NO₂ fluctuations. From November 2008 to February 2009, thermal power showed a maximum decrease of 12.10%, leading to tropospheric NO₂ density decreasing by 16.97% over the same period of the previous year. When thermal power capacity increased by 29.63% in the same period of the following year, when the effect of the financial crisis had declined, tropospheric NO₂ density correspondingly increased by 30.07%. A decline in industrial output and thermal power of about 10% can induce a decrease of about 30% in NO_x emissions and NO₂ densities, meaning that a relatively small fluctuation in industrial production can cause a large decrease in tropospheric NO₂ densities over industrially developed areas like the YRD region.

(3) The thermal power industrial sector accounts for the greatest coal consumption and highest NO_x emissions in the YRD. The smelting and pressing of ferrous metal sector and the sector of manufacturing communications and computer equipment have the greatest industrial output and electricity consumption. Since electricity is mainly obtained from local coal-burning thermal plants without NO_x-processing equipment, and the coal-based energy structure will not change in the short term, installing NO_x-removal devices in thermal power plants is an important and feasible way of controlling regional NO_x emissions and pollution in the YRD. In the long run, it will be necessary to improve electricity production and the industrial structure, and develop clean alternative energy sources to provide a fundamental solution to local NO_x pollution.