Space-based measurements of air quality during the World Expo 2010 in Shanghai

During the World Exposition 2010 (Expo, from May to October), emission control measures were implemented in Shanghai and surrounding areas to improve the air quality. To evaluate the effect of these measures, we use the tropospheric NO2 column, aerosol optical thickness (AOT) and CO concentration observations from the satellite instruments GOME-2, MODIS and MOPITT, respectively. The analysis shows about 8% and 14% reductions of tropospheric NO2 columns and AOT respectively over Shanghai during the Expo period, compared to the past three years. A 12% reduction of CO concentration at 700 hPa over Shanghai and surrounding areas is found during the Expo period. On the other hand, the satellite measurements show increases of NO2 by 20% and AOT by 23% over Shanghai urban areas after the Expo (November 2010–April 2011), when the short-term emission control measures were lifted. Our study indicates that the air quality measures were effective in Shanghai and surrounding provinces during the Expo period.


Introduction
.5 • E), one of the largest cities in China, held the World Exposition 2010 (Expo) from 1 May to 31 October 2010. Its fast growing economy, long history of industrial development and rapid industrialization in neighboring provinces pose large challenges to the air quality in the Shanghai area. Chen et al (2006) reported that in Shanghai the power plants, industries and transportation contributed about 40%, 24% and 22% of NO x emissions and about 34%, 46% and 7% of PM 10 emissions in 2000, respectively.
They predicted that the transportation sector, which contributed about 30% and 18% of NO x and PM 10 emissions respectively in 2010, would become increasingly important in the future. Using GOME and SCIAMACHY satellite measurements, van der A et al (2006) showed a yearly increase of 20%±6% in the tropospheric NO 2 columns over Shanghai during the period 1996-2005, compared to a yearly increase of 10% ± 4% in Beijing. Based on ground-based measurements, Chan and Yao (2008) found that the local soot, SO 2 and NO 2 emissions in Shanghai were 2-3 times than those in Beijing during 2003-5. The air quality during the Expo in Shanghai was a major concern for the local government, similar to that during the Olympic Games in Beijing. To improve the air quality during the Expo, the Shanghai municipal government pursued many emission control measures (ECMs) focusing on energy, industry, transport and construction (UNEP 2010). During the Expo period, short-term ECMs were brought forward for the main pollution sources that greatly affect the air quality in Shanghai. Similarly to during the Beijing Olympic Games (Wang et al 2009), several high polluting enterprises such as coking and cement industries were ordered to reduce their operating capacities and all construction activities in the city were suspended during the Expo. Meanwhile, agricultural burning was fully banned in Shanghai and the surrounding provinces because agricultural fires have a significant influence on the air quality in Shanghai (Li et al 2010). Shanghai enforced some traffic control measures including implementation of new vehicle emission standards, replacement of older polluting cars, and access restriction of trucks and old cars entering into Shanghai (CAI-Asia 2010). Although no very strict traffic control measures (such as removing half of the cars from the road as implemented during the 2008 Beijing Olympic Games) were implemented during the Expo period because of its long duration, the NO x emissions from vehicles would decrease by about 10%, as estimated by Shanghai Environmental Protection Bureau (2010). In addition, joint pollution prevention and control measures were carried out in the surrounding provinces (Zhejiang and Jiangsu Provinces) because air pollutants from the surrounding areas could also affect the air quality in Shanghai under certain meteorological conditions. A number of papers (e.g., Cheng et al 2008, Mijling et al 2009 have been published on the air quality changes during the Sino-African Summit and Olympic Games in Beijing measured by ground-based or satellite measurements, indicating the decrease of the measured pollutants because of ECM implementation. Satellite measurements showed significant decreases in tropospheric NO 2 (43%), SO 2 (13%) and CO (12%) during the Olympic Games (Witte et al 2009) and the emissions of NO x over Beijing were reduced by 40% during the Sino-African summit (Wang et al 2007). However, these ECMs were only enforced for a short time period (from a few days to 1-2 months).
In this study, we use satellite measurements to monitor the evolution of major air pollutants in Shanghai and the surrounding areas: NO 2 , aerosol loading and CO. Their changes during and after the Expo are tracked using data from three satellite instruments: the Global Ozone Monitoring Experiment 2 (GOME-2), Terra's Measurements of Pollutants in the Troposphere (MOPITT) and Aqua's Moderate Resolution Imaging Spectroradiometer (MODIS). Meanwhile, comparisons between GOME-2 tropospheric NO 2 columns and in situ NO 2 measurements are also shown. The six-month duration of the Expo provides a unique opportunity to examine the effectiveness of ECMs on air quality over Shanghai and the surrounding provinces and to demonstrate the ability of satellite observation to monitor and quantify these changes.

Data and methods
To evaluate the effect of ECMs on the nitrogen oxides (NO x = NO + NO 2 ) pollution during the Expo period, tropospheric NO 2 observations from the GOME-2 sensor are used. GOME-2 is a nadir-viewing scanning spectrometer on EUMETSAT's MetOp-A satellite, launched in 2006. It has a spectral coverage of 240-790 nm and a spectral resolution between 0.26 and 0.51 nm. GOME-2 observes four times smaller ground pixels (80 × 40 km 2 ) than its predecessor GOME on ERS-2, and provides a global coverage within about 1.5 days (Munro et al 2006). The MetOp-A satellite is flying on a sunsynchronous orbit and has an equator crossing time of 9:30 local time. Operational total and tropospheric NO 2 products are routinely retrieved at the German Aerospace Center (DLR) in the framework of the EUMETSAT Satellite Application Facility on Ozone and Atmospheric Chemistry Monitoring (O3M-SAF) with the GOME Data Processor (GDP) version 4.4 (Valks et al , 2011. In this study, monthly average GOME-2 tropospheric NO 2 data (horizontal resolution 0.1 • × 0.1 • ) for the period of January 2007-April 2011 are used. The total uncertainty in the GOME-2 tropospheric NO 2 column is about 40% for polluted regions (Valks et al 2011). A first validation of the GOME-2 NO 2 product using MAXDOAS in Beijing indicated that the GOME-2 tropospheric NO 2 columns correlated well (correlation coefficient up to 0.81) with the ground-based data (Pinardi et al 2010).
Daily mean NO 2 concentrations in Shanghai are derived from the NO 2 air pollution index (API), which is the average of in situ measured NO 2 concentrations at nine monitoring stations from the Shanghai Environmental Monitoring Center (SEMC). The NO 2 API records for previous years  have been used to compare in situ and satellite NO 2 measurements (Zhang et al 2007).
To analyze the aerosol loading during the Expo, the aerosol optical thickness (AOT) products from MODIS data have been used. The MODIS instrument onboard NASA's Aqua satellite (launched in 2002) has 36 spectral channels ranging from 0.4 to 14.4 μm representing three spatial resolutions: 250 m (2 channels), 500 m (5 channels) and 1 km (29 channels). Seven of these channels (0.47-2.13 μm) are used to retrieve aerosol characteristics (Remer et al 2005). The uncertainty of the MODIS AOT is within 20-30% (Levy et al 2007). Wang et al (2010) concluded that the MODIS aerosol products have good accuracy over China. Monthly averaged AOT values on a horizontal resolution of 1 • × 1 • (MOY08 collection 5.1 (Levy et al 2007)) from January 2007 to April 2011 are analyzed in this work.
Atmospheric profiles of CO are measured by MOPITT (launched on 1999 onboard NASA's Terra spacecraft) using thermal radiation at 4.7 μm with a spatial resolution of 22 × 22 km 2 . Long-term MOPITT data have been used for monitoring the evolution of CO levels in Northeast Asia induced by both industrial activities and biomass burning (Choi and Chang 2006). In the present study, MOPITT V4 level 3 data (Deeter et al 2009) from January 2007 to April 2010 are analyzed. The monthly average CO mixing ratios on a horizontal resolution of 1 • × 1 • at the 700 hPa level (instead of 900 hPa) are used because MOPITT has little sensitivity to CO within the boundary layer (BL, Deeter et al (2003), Emmons et al (2004Emmons et al ( , 2007). Emmons et al (2004) showed that the MOPITT CO measurement precision above 700 hPa is about 10%.
To analyze the effect of the meteorological conditions on the tropospheric NO 2 and CO distribution as well as the aerosol loading for the Shanghai area, precipitation, temperature and wind data recorded by automatic weather stations situated at Hongqiao airport, Shanghai (31.2 • N, 121.3 • E), Pudong airport, Shanghai (31.1 • N, 121.8 • E) and Shanghai (31.4 • N, 121.5 • E) are used. Furthermore, satellite retrieved convective and non-convective precipitation rates (mm h −1 ) for every 3 h in 0.25 • spatial resolution are also utilized to calculate the  The details about precipitation rate retrieval using passive microwave satellite can be found in Ferraro et al (1997,2000) and Joyce et al (2004).

Results and discussion
In order to analyze the effect of the ECMs on the air quality during the Expo 2010, two periods are defined: the Expo period (May-October 2010) when all the ECMs were in effect and the post-Expo period (November 2010-April 2011) when shortterm ECMs were lifted. May to October 2007-9 served as a reference for the Expo period. Similarly, November to April 2007-10 served as a reference for the post-Expo period.
The GOME-2 tropospheric NO 2 distribution for the Expo period shows high tropospheric NO 2 columns above large urban and industrial areas such as Bohai Economic Rim including Beijing and Tianjin, the Yangtze River Delta including Shanghai, Zhejiang and Jiangsu Provinces and the Pearl River Delta including Guangzhou and Hongkong   (c) and (f). A ratio of less than 1 indicates more NO 2 in the reference period. During the Expo period, significant NO 2 reduction (c) is found around Shanghai (SH). recession on air pollutants over Northeast China is discussed in Lin et al (2010)). The strong increase of tropospheric NO 2 over the urban areas of China from 1994 to 2006 has been discussed by Richter et al (2005) and van der A et al (2006).
The GOME-2 tropospheric NO 2 columns for the Expo and the post-Expo periods are compared with those for the reference periods over Shanghai (see figure 3). The ratios of NO 2 columns between 2010 and the reference period show a reduction of NO 2 during the Expo period over Shanghai. Compared to previous years, tropospheric NO 2 columns over Shanghai decreased by about 8% during the Expo period. Figure 3 also shows that the tropospheric NO 2 columns increased by about 20% during the post-Expo period, compared to the reference period. As mentioned before, the yearly increase of the tropospheric NO 2 columns over Shanghai was about 20% ± 6% from 1996 to 2005 (van der A et al 2006) and from 2007 to 11 the tropospheric NO 2 columns also increased by about 20% over Northeast China. The NO 2 increase during the post-Expo period is consistent with the NO 2 trend over Shanghai during 1996-2005and over Northeast China during 2007. It is interesting to compare the reductions of NO 2 for the Expo to the NO 2 reductions during the Beijing Olympic Games (Witte et al 2009). It was concluded that during the Olympic Games short-term ECMs, especially removal of half of the cars from the road, contributed to a decrease of 43% in NO 2 columns over Beijing. For the Expo, we did not find such a large NO 2 decrease, which is possibly related to the absence of very strict short-term traffic control measures over Shanghai and the surrounding areas due to the long duration of the Expo (6 months). Moreover, the meteorological conditions during the 2008 Beijing Olympic Games were atypical and were favorable to reductions of tropospheric NO 2 columns (Mijling et al 2009), while the meteorological conditions had no clear influence on the changes of air pollutants during the Expo (see below).
In situ NO 2 measurements derived from NO 2 API for the period May 2007 to April 2011 have been used for the comparison with the GOME-2 measurements. Figure 4 shows a scatter plot of the monthly average in situ NO 2 Table 1. Tropospheric NO 2 columns (10 15 molec cm −2 ) and 1-σ standard deviation over Shanghai (30.7 • -31.5 • N, 120.9 • -121.9 • E, see the box in figure 3) as measured by GOME-2. In situ NO 2 concentrations (μg m −3 ) and 1-σ standard deviation over Shanghai as measured by the Shanghai Environmental Monitoring Center (SEMC). Tropospheric NO 2 columns and in situ NO 2 concentrations are given for the periods May-October 2010 (Expo), November 2010-April 2011 (post-Expo) as well as for the reference period (2007-9).

GOME-2
In situ measurements   concentrations and GOME-2 tropospheric NO 2 columns. A high correlation (correlation coefficient up to 0.83) between satellite and in situ measurements is found. Table 1 lists the average NO 2 concentrations and columns observed by in situ and satellite measurements for the Expo, the post-Expo and the corresponding reference periods. The in situ measurements indicate that NO 2 concentrations decreased by about 10% during the Expo and increased by about 15% after the Expo. GOME-2 observed an 8% reduction in the tropospheric NO 2 columns during the Expo and a 20% increase after the Expo, compared to the reference periods, respectively. Similar results were reported by a post-Expo workshop report (CAI-Asia 2011) based on ground-based measurements. Both the satellite and in situ measurements indicate that the short-term ECMs implemented in Shanghai were effective in reducing the NO x concentrations during the Expo. The lifting of short-term ECMs after the Expo resulted in a rebound in NO x pollution. The reduction of NO 2 over a larger region including Shanghai and the surrounding provinces is not so clear because the traffic control measures might not be applied in the surrounding provinces.
To estimate the effect of ECMs on the aerosol loading over Shanghai, average Aqua/MODIS AOT during the Expo period and the corresponding reference period are compared in figure 5. During the Expo period, the AOT decreased by about 14% over Shanghai (circle in figure 5) compared to the reference period. Similar reductions in the PM10 concentrations in Shanghai during the Expo period were reported by a Mid-Expo Air Quality Report (CAI-Asia 2010). The results showed that PM 10 concentrations decreased by about 12% during the Expo, compared to the same months (May-July) from 2001 to 2009. Figure 5 and table 2 illustrate that after the Expo, there was an increase of about 23% in the AOT over Shanghai, compared to the reference period. The monthly mean time series of AOT over Shanghai from May 2010 to April 2011 and the reference period are plotted in It is likely that the double increase in visitors and relatively loose ECM implementation in October (the last month of the Expo) contributed to the larger AOT compared to the reference period. The observed reduction in aerosol loading is most likely related to the short-term ECMs implemented over Shanghai during the Expo period.
Shanghai's neighboring provinces Zhejiang and Jiangsu, which are becoming rapidly industrialized and are highly urbanized, have a large impact on the air quality in Shanghai (UNEP 2010). Therefore, joint pollution prevention and ECMs based on the cooperation platform of Shanghai and the surrounding provinces were in effect during the Expo period. To check the effect of these joint ECMs, an analysis domain (29 • -32 • N, 120 • -122 • E) including Shanghai and the neighboring provinces was defined (box in figure 5). AOT values over this domain for the Expo and post-Expo periods are compared with those during the same months in the reference period, as listed in table 2. AOT reduction during the Expo period is about 21% and AOT increase after the Expo is about 18%, relative to the reference period. This may be a consequence of joint ECMs implemented in Shanghai and the surrounding provinces during the Expo period. Figure 7 shows monthly averages of MOPITT CO concentrations at 700 hPa over Northeast China. In contrast to NO 2 (figure 2), there is no significant increase of CO concentrations from 2007 to 2011. Similarly to NO 2 , the effect of economic recession on CO concentrations is visible from late 2008 to the first half of 2009. For the Expo analysis, a larger region (29 • -32 • N, 120 • -122 • E) including Shanghai and the neighboring areas is used because CO has longer chemical lifetimes in the troposphere and enhanced concentrations may be found further away from the emission sources (Witte et al 2009). The relative changes in CO concentrations between the Expo period and the reference period are shown in figure 8. CO concentrations over Shanghai and the neighboring areas (box in figure 8) decreased by about 12% during the Expo period, compared to the reference years. It is interesting to mention that in contrast to Shanghai, the large urban areas located in the north of China (i.e. Beijing and vicinity) where no ECMs were implemented showed an increase in the CO concentrations.
The monthly mean time series of CO from May 2010 to April 2011 and the reference period over Shanghai and neighboring areas are plotted in figure 9. MOPITT CO exhibits a seasonal variation (minimum in summer) in Shanghai, which is similar to that of CO ground-based measurements (Ran et al 2009). The monthly mean CO in figure 9 shows low CO concentrations during the Expo period especially from July to September, relative to the past three years. After the Expo, the CO concentrations are similar to those during the reference period . The reductions of CO reflect the efficiency of the ECMs imposed on industry and agricultural burning in Shanghai and the surrounding regions during the Expo period.
It should be noted that we have used average data for a longer time period in our study, which reduces the error. We do not expect that the error in the satellite measurements will have a large impact on the analysis of the effectiveness of the ECMs. Table 3. Meteorological data at Hongqiao airport, Shanghai (31.2 • N, 121.3 • E). Mean temperature, relative humidity, mean precipitation amount per month, prevailing wind direction, wind speed (including 1-σ standard deviation) and mean number of rainy days per year are given for the Expo period (May-October 2010) as well as for the reference period . The 30 years  climatological data are also listed.  To investigate the effect of meteorological conditions on the changes in NO 2 , aerosol loading and CO as described above, precipitation, temperature and wind data for Hongqiao airport, Shanghai during the Expo period and the reference period have been used. In addition, 30 years (1971In addition, 30 years ( -2000 of climatological data for the Shanghai area are also listed in table 3. It is shown that that the meteorological conditions, including prevailing wind direction, mean precipitation amount per month and mean number of rainy days per year, during the Expo and the reference period in Shanghai were typical, compared to the climatological data in the 30 years. In addition, we checked the meteorological data from another two weather stations situated at Pudong airport, Shanghai and Shanghai. The meteorological conditions during the Expo and the reference period in these two weather stations are also typical, compared to the climatological data. A number of studies (Ding et al 2008, Zhang et al 2010a have found that in May-August the Asian summer monsoon affects the air quality over eastern China by influencing transport, chemical reactions and deposition of air pollutants. To investigate the potential effect of the East Asian summer monsoon (EASM) strength on interannual variations of air pollutants over Shanghai, EASM indices from 2007 to 2010 are used (Li and Zeng 2003). Based on the results of Zhang et al (2010a), the high aerosol pollution over East China is expected due to a relative weak summer monsoon in 2010. In addition, satellite retrieved precipitation data show that the total precipitation amount (about 720 mm) during the Expo period is smaller than for the same months in 2007 (about 1200 mm). Improved air quality in Shanghai during the Expo period is found, although a lower relative humidity and rainfall, which are favorable for air pollutant accumulation, occurred. Therefore we do not expect that the meteorological conditions had a significant impact on the improved air quality in Shanghai during the Expo period.
It is important to assess the consistency between the observed improvements in air quality and the emission reductions due to the control measures during the Expo. However, quantitative results on the emission reductions are not publicly available. The latest emission inventory over Shanghai was developed by Huang et al (2011) for the year 2007. It is difficult to compare the emission and satellite observations due to the lack of recent emission data. Zhang et al (2010b) are developing a fast method for estimating China's monthly NO x emissions by integration of the bottomup method and OMI observed NO 2 trends. In the future, we plan to make comparisons between these up-to-date emission inventory data and GOME-2 NO 2 measurements.

Conclusion
We present a space-based assessment of the atmospheric composition changes in major air pollutants in Shanghai during and after the Expo in 2010. The GOME-2 measurements show 8% reduction of tropospheric NO 2 during the Expo period but 20% increase in Shanghai during the post-Expo period, compared to the same months in previous years. A high correlation (correlation coefficient up to 0.83) between the GOME-2 tropospheric NO 2 columns and in situ NO 2 concentrations is found. The analysis of the MODIS aerosol optical thickness in Shanghai shows an AOT reduction of 14% during the Expo period and a significant AOT increase of 23% after the Expo. A 12% reduction of the CO concentration at 700 hPa over Shanghai and the surrounding areas is found by MOPITT during the Expo period. The meteorological conditions do not have a significant impact on the air quality in Shanghai during the Expo. These results indicate that the air quality measures enforced by the Shanghai government were effective in Shanghai during the Expo period. Moreover, we find an AOT reduction of 21% over a wider region (including part of Jiangsu and Zhejiang Provinces) during the Expo period, which reflects the efficiency of the joint pollution control measures imposed by Shanghai and the surrounding provinces. The lifting of short-term ECMs such as access restriction of trucks and old cars into Shanghai and suspending work at factories and construction sites most likely contributed to the NO 2 and AOT increase after the Expo. Therefore, more stringent and consistent controls are needed to further improve the air quality in Shanghai and the surrounding regions. Our study demonstrates that satellite observations can be used to monitor and quantify air quality changes in megacities.