Interactive comment on “ How to improve the air quality over megacities in China ? – Pollution characterization and source analysis in Shanghai before , during , and after the 2010 World Expo ”

This manuscript addressed a very important issue. The 2010 World Expo held in Shanghai provided a unique opportunity to analyze the effectiveness of humanperturbed emission reduction on air quality. The conclusions of this manuscript have important implications for the future improvement of the air quality in Shanghai and other mega-cities around the world. The manuscript was well organized and presented. Therefore, the manuscript is suggested to be accepted by ACP with minor revisions documented below: 1. I gave a comment that the author did not provide enough proves for the probable formation mechanism of some pollution episodes when reviewing the


Introduction
Shanghai hosted the 2010 World Expo from 1 May to 31 October, which was the most attractive mega-event in China after the 2008 Beijing Olympic Games.The six months long exhibition under the theme of "Better City Better Life" aimed to promote an environment-friendly and resourcesaving "Green Expo".A number of records were achieved during the expo.The number of countries and organizations that participated was the largest with more than 240 countries and international organizations.The number of volunteers was also the largest and the visitor numbers exceeded the target of 70 million, making the Shanghai Expo the biggest event in the World Expo history.
Air quality issue has always been a big concern and attracted tremendous attention in megacities of China, especially during some international mega-events.Lessons and experiences from the 2008 Beijing Olympic Games demonstrated the effect of unprecedented human perturbation on the large reductions of air pollution emissions and improvement of air quality (Wang et al., 2009(Wang et al., , 2010)).Different from the short-term Olympic Games, the Shanghai Expo extended a much longer time and hosted many more visitors, bringing more difficulties and challenges for ensuring the good air quality.Various long-term control measures were implemented during the 11th Five Year Plan (2006Plan ( -2010)), including implementation of stricter coal-fired boiler emission standards and clean fuel adoption, upgrade of the motor vehicles to National Standard IV, clampdowns on heavily polluting trucks, control of dust produced by construction operations, etc. (CAI-Asia, 2009).During the expo, additional measures were implemented.For instance, public transportation was encouraged as the primary means for travel.According to the circular of Shanghai Municipal People's Government on restriction on transport of high-pollution vehicles, the vehicles within the inner-ring roads should have the environmental protection label.The exhaust emission of new cars must reach the national Class IV standard (equivalent to European IV standard) (cf.http://en.expo2010.cn/a/20090605/000002.htm).In addition, joint pollution controls over the Yangtze River delta (YRD) region (i.e,Shanghai, Jiangsu and Zhejiang provinces) were carried out to minimize the effect of regional transport on the air pollutants in Shanghai.Especially, the burning of straw was strictly prohibited (SEPB, 2010), which is a major source of air pollutants in the harvest season (Huang et al., 2012a).
The 2010 World Expo held in Shanghai provided a unique opportunity to analyze the impact of human-perturbed emission on air quality and advocate policy that will improve future air quality in Shanghai and in other Chinese cities.A space-based study showed a preliminary view of air quality in Shanghai and neighboring provinces, finding reductions of AOT (aerosol optical thickness), NO 2 , and CO in Shanghai during the expo period compared to the past three years.However, significant increases of NO 2 by 20 % and AOT by  23 % over Shanghai urban areas were observed after the expo (Hao et al., 2011).A high-resolution time-of-flight aerosol mass spectrometer (HR-ToF-AMS) characterized the chemical composition of PM 1 (particulate matter) in Shanghai during a less than 1 month study period of the expo (Huang et al., 2012c).However, a clear picture of potential impacts on air quality from human regulations during the expo remains ambiguous and is rarely interpreted.
The primary goal of this study is to obtain detailed information of the atmospheric chemical composition under the varied emissions during the expo.Three intensive field campaigns were performed during the three seasons that the expo spanned; namely 2 April-14 May in spring, 25 July-24 August in summer, and 20 October-29 November in autumn (Fig. 1).We specifically included some periods before and after the expo for comparison between the expo and non-expo periods.The atmospheric processing, source identification and formation mechanisms of selected pollution episodes are discussed on the basis of each season.Insights into the role of meteorology on air quality and response of atmospheric chemistry to anticipated emission variations are illustrated.In addition to the discussion of air quality during the expo, we also include results from 2009 for a detailed comparison to 2010.

Observational site
Figure 1 shows the map of the Shanghai metropolitan area with district borders.The area of the expo campus is shown by the red polygon, partly built on the Pudong District and partly on the Yangpu District, which was separated by the Huangpu River.It occupied a total area of 5.28 km 2 .The Fudan observational site (31.3 • N, 121.5 • E) was located in the Yangpu District as shown by the star in Fig. 1.The Fudan monitoring site was located to the northwest of the expo campus and the distance from the Fudan site to the expo campus was around 8-11 km.All the instruments were located on the roof (∼20 m high) of a teaching building on the campus of Fudan University.Almost no high buildings are around this sampling site.This site could be regarded as representative of the megacity of Shanghai, standing for the mix of residential, traffic, construction, and industrial sources (Huang et al., 2012b).To verify whether the Fudan site can reflect the influences from the expo, we compare the PM 10 concentration of Fudan and that of the Weifang site (data source: SEMC (Shanghai Environment Monitoring Center)) inside the expo campus.As shown in Fig. S1a, the time series of PM 10 concentrations at these two sites co-varied very consistently with each other.Figure S1b shows the linear correlation of the hourly PM 10 concentrations between the two sites, with a significant correlation coefficient of 0.83.The slope of the regression equation reaches 0.97, very close to 1.00.Thus, this indicated the Fudan site could represent the air quality of the expo campus.Additionally, Fig. S2 shows the monthly wind rose at Fudan from May to October in 2010.Southeast winds dominated during the whole study period.Since the Fudan site was located to the northwest of the expo campus as shown in Fig. 1, it was evident that our monitoring site would be significantly impacted by the upstream emissions from the expo campus.

Automatic aerosol and gases monitoring
The Thermo Scientific TEOM 1405-D monitor simultaneously measured PM 2.5 , PM-coarse (PM 10−2.5 ) and PM 10 mass concentration upon an oscillating balance.PM accumulated on a mounted filter.The accumulation of the PM mass caused the changes in the frequency of oscillation.And this frequency was detected in quasi-real time and then converted by a microprocessor into an equivalent PM mass concentration every few seconds and recorded in the running average of 10 min.The sampler split a PM 10 sample stream into its fine (PM 2.5 ) and coarse (PM 10−2.5 ) fractions using a USEPA (US Environmental Protection Agency) designed virtual impactor for the additional 2.5 µm cutpoint.The total flow rate operated at 16.67 L min −1 , and two sepa-rate flow controllers maintained the coarse particle stream at 1.67 L min −1 and the fine particle stream at 3.0 L min −1 .The instrument was operated at a temperature of 50 • C to avoid the interference of moisture on the calculation of aerosol concentrations.PM concentrations were averaged and used at intervals of 1hr in this study.Trace gases instruments included a TECO 43i SO 2 analyzer, 49i O 3 analyzer, 48i CO analyzer, and a 42i NO-NO 2 -NO x analyzer.Some measures were implemented to eliminate the potential interference of NOy species on NO 2 measurements as much as possible, for example, using filter and subtracting the concentration of HONO.A Teflon filter was placed in front of the Mo catalytic converter that connected to the gas sampler.The Teflon filter had a retention rate of over 99.7 % for particles (e.g., particulate HNO 3 ) larger than 0.3 µm.The routine QA/QC (quality assurance/quality control) procedures included the daily zero/standard calibration, span and range check, station environmental control, staff certification, etc. according to the Technical Guideline of Automatic Stations of Ambient Air Quality in Shanghai based on the national specification HJ/T193-2005, which was developed following the technical guidance established by the U.S. Environmental Protection Agency (USEPA, 1998).The multi-point calibrations were weekly applied upon initial installation of the instruments and the two-point calibrations were applied on a daily basis.

Manual sampling
Aerosol samples of TSP (total suspended particles) and PM 2.5 were collected on Whatman 41 filters (Whatman Inc., Maidstone, UK) using medium-volume samplers manufactured by Beijing Geological Instrument-Dickel Co., Ltd.(model: TSP/PM 10 /PM 2.5 ; flow rate: 77.59 l min −1 ).All the samplers were co-located with the online sampler.The duration time of sampling was generally 24 h.More samples with shorter duration times were collected during the heavily polluted days.The filters before and after sampling were weighed using an analytical balance (model: Sartorius 2004MP) with a reading precision 10 mg after stabilizing in constant temperature (20 ± 1 • C) and humidity (40 ± 1 %).All the procedures were strictly quality controlled to avoid the possible contamination of samples.

Element analysis
Half of each sample and blank filter was digested at 170 • C for 4 h in high-pressure Teflon digestion vessel with 3 mL concentrated HNO 3 , 1 mL concentrated HCl, and 1ml concentrated HF.After cooling, the solutions were dried, and then diluted to 10 mL with distilled deionized water.Total 24 elements (Al, Fe, Mn, Mg, Mo, Ti, Sc, Na, Ba, Sr, Sb, Ca, Co, Ni, Cu, Ge, Pb, P, K, Zn, Cd, V, S, and As) were measured by using an inductively coupled plasma atomic emission spectroscopy (ICP-OES; SPECTRO, Germany).The detailed analytical procedures are given in Sun et al., 2004;and Zhuang et al., 2001.

Pre-expo pollution: dust invasion
The spring phase of the field campaign included almost one month of the pre-expo period (2-30 April) and the first half of the opening expo month (1-14 May) as shown in Fig. 2. The average PM 2.5 and PM 10 concentrations during this period were 34.5 ± 20.9 and 74.5 ± 56.7 µg m −3 , respectively.
In the newly enacted Chinese Ambient Air Quality Standards (GB3095-2012), the Grade II standards for annual PM 2.5 and PM 10 concentrations were set as 35 and 70 µg m −3 , respectively.Based on this criterion, PM 2.5 concentration was below this standard while PM 10 exceeded.Persistent emission controls during the past five years and additional control measures during the expo had made during the 2010 expo the best air quality period of the past decade in Shanghai (Lin et al., 2012).The PM 2.5 /PM 10 ratio had a moderate value of 50 % in this period; because eastern China is frequently influenced by the floating dust originating from the deserts in northern and western China in spring (Huang et al., 2010(Huang et al., , 2012a;;Wang et al., 2007).
From 2 to 25 April, particle concentrations generally stayed at low levels with several small peaks occurring occasionally.The mean concentrations of PM 2.5 and PM 10 during this period were 28.9 ± 21.4 and 52.8 ± 32.6 µg m −3 .However, a high pollution episode happened after this clean period.Starting from 15:00 LST (local standard time) on April 26, particle concentrations sharply climbed up with hourly peaks of over 300 µg m −3 .Heavy pollution lasted till 28 April.Afterwards, particle concentrations, especially PM 10 , started to decrease quickly.As this pollution occurred right before the opening of the expo, we denoted it as the "pre-expo pollution".During this period, PM 2.5 and PM 10 averaged 58.0 and 191.8 µg m −3 , respectively.Compared to the previous period (i.e., April 2 to 25), PM 2.5 increased about 1 fold while PM 10 increased almost 3 fold with a low PM 2.5 /PM 10 ratio of 0.30.Hence, this pre-expo pollution was evidently caused by coarse particles.Figure 3b shows the daily Al concentrations in the total suspended particles (Al TSP ) along with the hourly PM 10 concentrations during the spring of 2010.Al is a good trace element for mineral aerosol and its abundance could be used to quantify the intensity of the mineral source.Corresponding to the high pollution on 26-28 April, Al TSP also presented very consistent high peaks and temporal variation with that of PM 10 .The average Al TSP during the pre-expo pollution episode reached 16.2 µg m −3 .To quantify the mass concentration of mineral aerosol, it could be estimated by summing the major mineral elements with oxygen for their normal oxides, by using the formula: (Malm et al., 1994).Hence, the mineral aerosol during the pre-expo pollution episode accounted for a dominant mass fraction of 69 % in TSP. Figure S3 illustrates the  Compared to the same study period in 2009 (Fig. 3a), we also observed a floating dust event at the end of April.On April 25 in 2009, the daily Al TSP reached 13.7 µg m −3 and the mineral aerosol accounted for 76.8 % of TSP.This pollution had been investigated and determined to originate from the Gobi Desert (Huang et al., 2012a).Similar meteorological conditions on the dusty days were observed.As shown in Fig. 4a, b and d, e, the two dust events in 2010 and 2009 both occurred in specific meteorological conditions, e.g., prevailing northwesterlies, reduced dew points and low humidity.The prevalence of the Mongolian anticyclone originating from Mongolia and northern China triggered dust events frequently in spring.The intrusion of dust aerosol was not accidental and could be a potential threat to the air quality of the downstream regions.
If the dust events were excluded in both years, the average Al TSP concentration in the spring of 2009 and 2010 were 4.25 ± 1.89 and 3.48 ± 2.11 µg m −3 , respectively.Approximate 20 % reduction of the mineral dust was found in 2010.In megacities of China, mineral dust mainly derived from construction works and re-suspended road dust.In order to alleviate the dust emission, the Shanghai government banned most construction sites, implemented various dustproof measures, and cleaned the roadsides regularly, etc.It was estimated that construction dust dropped by 29 % during the expo (CAI-Asia, 2011).This indicated that the special regulations did help to reduce the emission of local dust and explained the lower mineral aerosol concentrations in 2010 than in 2009.

Response of secondary aerosol to human activities
Figure 4 shows the daily concentrations of major secondary inorganic aerosol (SNA, i.e., SO 2− 4 , NO − 3 and NH + 4 ) in PM 2.5 with meteorological parameters (wind speed/direction, temperature, dew point, precipitation, relative humidity and atmospheric pressure) during the spring of both 2010 and 2009.Comparison between the temporal variations of SNA in these two years illustrated that intensive pollution episodes occurred in completely different time frames.In the spring of 2009, one intensive pollution episode occurred from 4 to 10 April with average SNA concentration of 48.86 ± 5.01 µg m −3 , which had been investigated to be related to the local anthropogenic emission (Huang et al., 2012a).During the same period in 2010, SNA concentration was about 70 % lower.Wind speed could be a major factor responsible for this difference as it was higher in 2010 (mean: 3.0 ms −1 ) than that in 2009 (mean: 2.4 ms −1 ).Additionally, air masses from 4 to 10 April in 2010 predominantly came from the northeast and southeast over the East China Sea, facilitating the dispersion of air pollutants.Apart from the high pollution episode in 2009, SNA stayed at relatively low levels of 10.9 ± 4.7 µg m −3 during the remaining days.However, the situation became the opposite in 2010.Mean SNA concentration from 11 April to 14 May in 2010 was 15.4 ± 8.4 µg m −3 , about 40 % higher than 2009.No significant differences were found among the major meteorological parameters except precipitation in these two years since 11 April (Fig. 4a, b, d, e).Although more precipitation events occurred since April 11 in the spring of 2010 (e.g., on 11, 13-14, and 20-21 April) than in the same period of 2009 (note that no aerosol was sampled on 19 and 24 April when intense precipitation events occurred), the mean SNA level was still found higher than in 2009 as discussed above.This suggested that the anthropogenic emission was enhanced as the expo was approaching.Since 22 April in 2010, rare precipitation events occurred and the role of wet scavenging on cleansing the air pollutants was negligible.The temporal variation of SNA presented an evidently increasing trend from 22 April to 2 May in 2010 (Fig. S4).Although temperature gradually increased during this period and higher temperature did not favor the accumulation of nitrate and ammonium in the particulate phase, nitrate still had significant increase from 4.0 µg m −3 on 22 April to 13.4 µg m −3 on 2 May with a peak value of 18.0 µg m −3 on 29 April.The daily increasing rate of nitrate during this period reached 1.1 µg m −3 d −1 with a linear correlation coefficient of 0.9.The NO −  increasing rate of 0.41 µg m −3 d −1 , which was probably due to the increasing electricity demand.In regard of the high pollution on 29 April, there was a mandatory abatement of power plants emissions afterwards and around 30 % of SO 2 emission was reduced to ensure the recovery of air quality in the next few days (SEMC, 2011).It was observed that the air quality was alleviated to some extent on April 30, indicating the emission reduction was effective.However, SNA concentrations rose again in the first two days of the expo (May 1 and 2) although meteorological conditions were favorable, e.g., southeast winds from the ocean and high wind speeds (Fig. 4a).Figure S5 shows the daily numbers of the expo visitors from 1 to 14 May.The temporal variations of SNA since the opening of the expo corresponded well with that of visitor numbers, suggesting the human activities can be a major factor affecting the pollution level.On 1 and 2 May, visitor numbers both exceeded 200,000, which were at high levels in this study period, and could explain the high SNA level on these two days.Afterwards, the daily visitor numbers from 3 to 7 May decreased about 40 %.It was found that several rainfall events occurred on 5 and 6 May, and this probably partly accounted for the lower attendance number.Both reduced expo visitors and the occurrences of precipitation were beneficial for the reduction of air pollution, which was reflected in the decrease of PM concentrations (Fig. 2) and also the SNA concentrations (Fig. 4c).After that, there was an increasing trend of expo visitors and slightly increasing SNA concentrations were also observed.Overall, we found out human activities dominated the variations of aerosol chemistry in the spring phase of the expo.

Expo in summer: biomass burning pollution
The summer phase of the expo study period lasted from 25 July to 24 August.After almost three months of operation, the daily visitor numbers of the expo became relatively  steady with a daily number of around 42 000.During this period, the major meteorological conditions fluctuated insignificantly.For example, the standard deviation of temperature, dew point, RH (relative humidity), and atmospheric pressure was 3.1 • C, 1.2 • C, 10.7 %, and 3.9 mb, respectively, as compared to their average values of 31.1 • C, 26.1 • C, 75.8 %, and 994.0 mb.Winds predominantly blew from the southeast as shown in Fig. 5a.Precipitation events occurred mainly on four days, i.e., on 2, 4, 15, and 18 August.The wet scavenging would generally reduce the aerosol concentrations, especially the heavy rain on 18 August caused a significant decrease of SNA levels as shown in Fig. 5c.The average PM 2.5 and PM 10 concentrations during this period were 37.5 ± 30.9 and 56.6 ± 37.1 µg m −3 , respectively.Compared to spring, PM 2.5 was higher while PM 10 decreased a lot.Higher humidity and the absence of dust events during summer should be the major reasons for the lower coarse-particle concentrations.Particle concentrations fluctuated significantly, and  two intensive pollution episodes were observed, one from 28 July to 4 August, and the other from 11 to 17 August (Figs.2, 5c).In this period, meteorologic conditions did not vary much (Fig. 5a, b), indicating the fluctuation of air quality was more emission driven.Overall, SNA contributed a moderate fraction of 40 % to PM 2.5 and mineral aerosol contributed less than 10 %, leaving a significant fraction of particle mass unexplained.The Pearson linear correlations between SO 2 , NO 2 , CO and PM 2.5 were plotted for each study period in Fig. S6.In summer, SO 2 showed weak correlation with PM 2.5 (correlation coefficient R = 0.47) and NO 2 showed moderate correlation (R = 0.66), while CO showed the most significant correlation with PM 2.5 (R = 0.76).This meant that anthropogenic emission was not the dominating source of air pollution.The highest correlation between CO and PM 2.5 was typically observed during biomass burning events as CO was a major product from incomplete combustion of biomass (Huang et al., 2012a;Sun et al., 2009).Figure 6 shows the time series of hourly CO and daily particulate K + concentrations.These two species varied very consistently with each other and both increased during the two intensive episodes.K + was a typical tracer for biomass burning and its concurrent temporal variation with CO, PM 2.5 , and PM 10 suggested that biomass burning was indeed a considerable source for the two intensive episodes during summer.MODIS detected intense fires burning over the Yangtze River delta region as shown by the fire spots in Fig. S7.During the first pollution episode (28 July to 4 August), considerable numbers of fire spots were distributed over the majority area of the Yangtze River delta, especially in the central and southern part of Jiangsu Province, and northern part of Zhejiang Province.For the second pollution episode ( 11 to 17 August), although the total numbers of fire spots reduced a lot compared to the first one, the Shanghai metropolitan area was still surrounded by intense biomass burning from neighboring provinces, i.e., northern Zhejiang and part of Jiangsu Province.Facilitated by the prevailing southeast winds (Fig. 5a), air quality of Shanghai could be indeed impacted by biomass burning from the neighboring regions.From a high resolution biomass burning emission inventory FLAMBE (Fire Locating and Modeling of Burning Emissions, (Reid et al., 2009)), we calculated the hourly biomass carbon emission in the domain of 28-32 • N, 119-123 • E and presented its temporal variation in Fig. 6.The intensity of biomass carbon emission co-varied relatively well with CO and K + concentrations, further corroborating the impact of biomass burning on the local air quality.
Summer is the main harvest season in eastern China.Farmers were busy during the harvest, with substantial production of crop residues.Callback or recycling use of these biofuels was costly, while the most convenient way to get rid of the crop residues was to burn them (Yang et al., 2008).Although Shanghai, Jiangsu and Zhejiang governments jointly issued the announcement on prohibiting the open burning of crop residues during the expo (SEPB, 2010), the results in this study indicated that considerable and widespread biomass burning still occurred.On the one hand, unpredictable human activities on biofuel combustion and the sparse distribution of biomass emissions increased the difficulty of preventing the occurrence of biomass burning.On the other hand, regulation enforcement in most Chinese local governments was not good enough, especially in some remote counties.Also, due to the lack of common sense of environmental protection, some farmers did not obey the regulations and just burned the crop residues privately.

Deterioration of air quality on the closing day of the expo
As the expo approached its end, more visitors sought the last chance to visit it, causing October to be the busiest month contributing over 20 % of the total visitor numbers during the whole expo.Air quality was monitored from 20 October to the close of the expo on 31 October with continuous measurement till 2 December.As shown in Fig. 2, a notable change of air quality between the two time frames was observed.From 20 to 30 October before the close of the expo, PM 2.5 and PM 10 averaged 30.6 and 51.3 µg m −3 .Other pollutants, for example, SO 2 , NO 2 and CO averaged 19.1, 41.2, and 821.4 µg m −3 , respectively.The air quality in Shanghai during this period could be regarded as "good" and we believed that stringent control measures must be implemented to keep the air quality "good".In addition, Shanghai experienced strong northerly winds from 20 to 30 October (Fig. 8a) and this also contributed to the better air quality.However, all the air pollutants drastically increased during the day (31 October) when the expo was announced to be closed.
Figure 7 shows the diurnal variation of PM 2.5 , PM 10 , NO 2 , CO, and SO 2 with an hourly wind profile from 30 October to 2 November.No precipitation events were observed during this period.On 30 October, the concentrations of all the air pollutants stayed at a relatively low level.Strong wind was partly responsible for this.From 00:00 LST on 31 October,  PM 2.5 and PM 10 concentrations gradually increased till the early morning of the next day.Although stronger winds appeared from around 09:00 to 20:00 LST on 31 October, the concentrations of air pollutants did not decrease and instead continued to increase, suggesting an increase of local emissions.From 20:00 LST to the next morning, the atmosphere turned to be stagnant as indicated by the absence of winds.PM 2.5 and PM 10 continued to rise and reached the extremely high concentrations of 320.8 and 407.8 µg m −3 at 08:00 LST of 1 November.Afterwards, PM concentrations sharply decreased and reached troughs around noon.The appearance of northeast and north winds evidently was beneficial for cleaning the air pollution.In addition, the elevated mixing layer at noon due to higher temperature could also dilute the emission.However, PM climbed up again and stayed at high levels till 07:00 LST on 2 November.Afterwards, persistent northeast and north winds from the ocean helped cleanse the air of pollutants again.During these four days, insignificant differences of wind pattern, temperature, atmospheric pressure, relative humidity and dew point were observed.However, compared to the reference period of 30 October, completely different diurnal patterns and concentrations of air pollutants in the following days were observed.Enhanced local emission was suggested to be mainly responsible for this tremendous rebound of all the air pollutants.This clearly indicated the lifting of short-term emission control measures (e.g., loose control on the vehicle flows, and allowance of high-duty vehicles into the city) that took place right after the announcement of the closing of the expo.
The temporal variations of pollution gases could probably give us some clues on the pollution source.Figure 7b shows that NO 2 and CO varied very consistently and showed peaks as those of particles around the same time.NO 2 and CO both started to climb very quickly at 17:00 LST on 31 October and reached over 150 and 2500 µg m −3 , respectively, lasting almost 14 h till 09:00 LST on 1 November.Another similar pollution episode started at 16:00 LST on 1 November and ended at 07:00 LST on 2 November.Mean NO 2 and CO concentrations reached 104.5 and 1715.8 µg m −3 on 31 October, and 144.5 and 2398.9 µg m −3 on 1 November, respectively.The NO 2 level even exceeded that of the ever recorded heaviest pollution in Shanghai on January 19, 2007 (Fu et al., 2008).NO 2 and CO both significantly correlated with PM 2.5 with correlation coefficients of 0.81 and 0.88 during this pollution episode, respectively.Fire spots detected from MODIS on Aqua and Terra satellites on 31 October and 1 November are plotted in Fig. S8.Almost no or very few fire spots were observed in Shanghai and areas adjacent to Shanghai.Thus, the possible impact of biomass burning emission on the extremely high CO and NO x concentrations could be neglected.The peak occurrences of NO 2 and CO during the rush hours indicated traffic emission should be one of the main causes for the deterioration of air quality near the end of the expo and after the expo.Although the SO 2 concentration during this period also increased compared to during the expo, its temporal variation did not fluctuate as strongly as that of NO 2 and CO as shown in Fig. 7b.Also, the linear relationship between SO 2 and PM 2.5 only presented a moderate correlation coefficient of 0.42.This indicated that stationary sources (e.g., power plants, industrial emission) were not the dominant contributors to this heavy pollution.
Owing to the significant enhanced pollutant precursors, the corresponding increase of SNA was expected as shown in Fig. 8c.SNA in PM 2.5 reached 42.1 and 68.2 µg m −3 on 31 October and 1 November, respectively.Compared to 20-30 October, SNA increased 3-6 fold.Of which, nitrate increased the most, about 5-8 fold, while sulfate and ammonium increased about 2.5-4.5 fold.Among the total water soluble inorganic ions, nitrate accounted for the largest fraction of 50 %, corroborating the impact from the enhanced vehicle emission.Overall, air quality in Shanghai had plummeted since the expo was announced closed, sliding from "good" to "severely polluted".

Other secondary inorganic pollution episodes
Compared to the spring and summer study periods, much more occurrences of intensive pollution episodes and higher pollution peaks were observed during the post-expo period as shown in Fig. 2. Except for the heavy pollution episode discussed above, there were also several other secondary inorganic pollution episodes as shown in Fig. 8c.Three SNA pollution episodes were sorted out, which occurred on 6-7, 19-21 November, and 1-2 December.The average SNA concentration during these episodes reached 46.9 µg m −3 and accounted for 55 % of PM 2.5 .The three pollutant gases (i.e., SO 2 , NO 2 and CO) all presented significant correlations with PM 2.5 , which were the highest during the three study periods (Fig. S6), indicating the dominant role of emissions.It  was noted that these pollution episodes were all associated with northeast winds from the ocean (Fig. 8a).Thus, this synoptic meteorology precluded the possibility of the longrange/regional transport from inland regions, which meant that local emission was the major source of pollution.Winds speeds during these SNA pollution episodes were relatively lower than those low pollution periods as shown in the figure.Also, relative humidity was usually higher.Those unfavorable meteorological conditions would surely contribute to the deterioration of air quality.
As we compare the same study period in the autumn of 2010 and 2009 (Fig. 8), air quality during the expo and post-expo period was completely different between the two years.During the expo, SNA was 12.2 µg m −3 , about 25 % lower than the same period in 2009.Implementation of strict control measures and favorable meteorological conditions (e.g., higher wind speeds) were the major reasons.During the post-expo period, SNA averaged 28.8 ± 15.8 and 13.5 ± 10.8 µg m −3 in 2010 and 2009, respectively.Over a 100 % increase of SNA in 2010 than in 2009 during the postexpo period clearly suggest the lifting of control measures was the main cause for the frequent occurrence of pollution episodes and poor air quality.

contributed from transported dust and local dust
On 12-13 November, another high pollution episode occurred as shown in Fig. 2. PM 2.5 far exceeded the criterion of 75 µg m −3 with the average concentration of 96.4 µg m −3 , and PM 10 reached the highest concentration of 398.1 µg m −3 during the whole study period.The mean PM 2.5 /PM 10 ratio was as low as 0.24, indicating this pollution was caused by dust, again, similar as the pre-expo dust pollutant discussed in Sect.3.1.1.A view from space helps to visualize the transport pathway of aerosol (Fig. S9) on 12 November.There was an obviously high AOD belt stretching from northern and eastern China out over the East China Sea and  Sea of Japan.Three-day back trajectories indicated that the source of aerosol originated from the Gobi Desert in Mongolia and Inner Mongolia.Compared to the pre-expo pollution event, this post-expo pollution event was evidently much stronger.Figure 9 plots the daily Al TSP and the hourly PM 10 concentrations, showing a very consistent variation of these two parameters.Between 02:00-13:00 LST on 12 November, PM 10 showed its highest peak with an average concentration of 665.1 µg m −3 , corresponding to the highest Al TSP of 43.1 µg m −3 .In this dust event, mineral aerosol accounted for a dominant fraction of 60 % in TSP.During the normal times (i.e., excluding the dust invasion events), mineral aerosol in urban areas mainly derived from construction works and re-suspended dust.Comparison between 2010 and 2009 (Fig. 9) found out the temporal variation of Al TSP varied relatively little in 2009 while it fluctuated much more intensively in 2010.For instance, the average Al TSP from 20-30 October 2010 was at a relatively low level of 2.21 ± 1.05 µg m −3 .Over the next two days, the daily Al TSP increased 3-4 fold on 31 October and 1 November.Due to lifting of the short-term control measures during the expo, construction sites started to re-open as soon as the expo was announced to be closed (SEMC, 2011).Thus, the resumed construction activities also contributed to this pollution episode in addition to the enhanced traffic emission as stated in Sect.3.3.1.
In order to quantitatively assess the impact from construction works, we calculated the concentrations of anthropogenic calcium, which could be used as a tracer for construction works in urban areas.The anthropogenic Ca concentration was calculated by removing its crustal source using Al as the tracer: Ca anthropogenic = Ca total − Al total × (Ca/Al) crust .The value of (Ca/Al) crust is 0.5, which is the ratio of Ca vs. Al in crust.Based on this method, the average Ca anthropogenic concentration was 3.55 ± 1.25 and 3.03 ± 1.92 µg m −3 during the expo period (20-30 October) and the post-expo period (31 October-20 November) in 2009, respectively.Obviously, no distinct difference between the two time frames was observed in 2009.In other words, the daily intensity of construction works was relatively stable.However, the situation was quite different in 2010.The average Ca anthropogenic concentration was 2.88 ± 1.85 and 6.98 ± 3.19 µg m −3 during the Expo and post-expo (excluding the transported dust events), respectively.Anthropogenic Ca in 2010 during the expo was about 20 % lower than the same period in 2009, indicating the control measures on construction emissions were effective during the expo.However, anthropogenic Ca in 2010 during the post-expo increased by 140 % compared to during the expo and 130 % compared to the same period in 2009.More and more construction sites started to re-open after the expo, causing the increased emission of construction dust.Thus, the resumption of construction works after the expo and the easing of pollution controls was also one of the causes contributing to the rebound of poor air quality.

Seasonal comparison of soluble ions between 2010 and 2009
Before we compare the differences of aerosol chemical composition between 2010 and 2009, it is necessary to evaluate if the meteorological conditions between these two years were distinctly different.Figure S10 shows the monthly mean values of the major meteorological parameters, i.e., temperature, dew point, wind speed, atmospheric pressure and monthly accumulated precipitation amount between 2009 and 2010.
As shown in the figure, there were small differences between the two years for the first four parameters.As for temperature, the monthly difference (2010 minus 2009) during the expo was −1.4,−1.8, −0.1, 2.5, 0.6, and −1.4 • C from May to October.As for dew point, wind speed and atmo-spheric pressure, the monthly difference was within the range of −1.7 ∼ 1.8 • C, −0.5 ∼ 0.5 m s −1 , and −1.2 ∼ 2.2 mb, respectively.Compared to their absolute values, these differences could be considered as trivial.Monthly precipitation had the largest divergence between the two years due to its high variability in both spatial and temporal scales.The total rainfall amount during the expo in 2010 was 709.2 mm, about 14 % lower than that of the same period in 2009 (828.8 mm).Overall, we did not find very significant differences in the major meteorological conditions between 2009 and 2010.Hence, in regard of a long-term period (e.g., on a monthly or seasonal basis), the air quality should be mainly determined by the emission.Figure 10 compares the concentration levels of major soluble ions in PM 2.5 and TSP between 2010 and 2009 during the three seasons, respectively.The left panels compare the seasonal concentrations of each species and the right panels present the percentage changes of 2010 relative to 2009.The result of spring is shown in Fig. 10a and b.Significant decreases of SO 2− 4 were observed in both PM 2.5 and TSP.Average SO 2− 4 in PM 2.5 decreased from 7.9 µg m −3 in 2009 to 5.4 µg m −3 in 2010 with a reduction of 32 %.The decrease of SO 2− 4 in TSP mainly came from its reduction in fine particles.Closing dirty and inefficient units of power plants and reducing the coal burning emission (UNEP, 2009) were the main causes for the reduction of particulate SO 2− 4 .Opposite to SO 2− 4 , NO − 3 in PM 2.5 increased from 6.2 µg m −3 in 2009 to 6.9 µg m −3 in 2010, indicating an increase of vehicle emission before and at the beginning of the expo.Of the cations, NH + 4 , K + , and Ca 2+ were found to have the most significant decreases.NH + 4 in PM 2.5 decreased from 3.7 µg m −3 in 2009 to 3.1 µg m −3 in 2010.Since NH + 4 was the major neutralization species for the acids, it was expected to decrease due to the significant decrease of SO 2− 4 .K + , a typical tracer for biomass burning, showed about 30 % decreases in both PM 2.5 and TSP. Figure S11a and b show the spatial distribution of biomass burning carbon emission from FLAMBE (Reid et al., 2009) during spring of 2010 and 2009, respectively.It could be observed that the intensity of biomass burning diminished over the Yangtze River delta region in 2010 as compared to 2009.Banning of open field biomass burning as a special temporary control measure probably took effect during the beginning of the expo, and this could explain the lowered K + level.Ca 2+ had the most significant decrease among all of the ions.Around a 69 % reduction of Ca 2+ in TSP was achieved and should be due to the restriction of construction works and the frequent cleaning of traffic roads.
Figure 10c and d show the comparison results in summer.Similar to spring, control measures on SO 2 emission continued to take effect, resulting in about 15 % reduction of SO 2− 4 compared to 2009.Also, increase of NO − 3 was evident, elevated from 4.5 µg m −3 in 2009 to 5.2 µg m −3 in 2010 in PM 2.5 and from 9.3 to 11.8 µg m −3 in TSP.A notable percentage of 15 and 27 % increases of NO − 3 in PM 2.5 and  TSP during summer suggested that more and more visitors came to the expo caused a substantial increase of vehicle emissions.NH + 4 had 37 and 28 % reductions in PM 2.5 and TSP, respectively, as compared to 2009.On the one hand, the reduction of SO 2− 4 offset the increase of NO − 3 .On the other hand, mean temperature in the summer of 2010 reached 31.1 • C, about 3.5 degrees higher than the same period in 2009.Higher temperature did not favor the formation of ammonium salts (e.g., NH 4 NO 3 and (NH 4 ) 2 SO 4 ) in the particulate phase and probably resulted in stronger depletion of NH + 4 .Na + and Cl − , which mainly derived from the marine source during summer in Shanghai (Huang et al., 2008), were found to decrease 30-50 % in 2010.As shown in Fig. S12a, wind in the summer of 2009 predominantly blew from the east, which is the open ocean of the East China Sea.While in the summer of 2010, winds shifted and mainly blew from the east to the south (counterclockwise) (Fig. S12b), where the impact from the marine source would be less important as continental outflows from Zhejiang Province and other continental regions would also contribute.Thus, the different transport pathways should be an important factor for the difference of marine aerosol levels and other pollutants.K + was the only cation found to increase in PM 2.5 .In Sect.3.2, we have ascribed biomass burning to be a major source of pollution in summer.Comparison of the FLAMBE biomass burning carbon emissions in the summer of 2010 and 2009 further corroborated our measurement results as shown in Fig. S11c  and d.Compared to 2009, biomass burning emission in 2010 was evidently more intense in Shanghai.As for the other parts of the Yangtze River delta, biomass burning in northern Zhejiang Province was most severe and evidently not well controlled.Via the prevailing south and southeast winds (Fig. S12b), Shanghai was probably impacted by the enhanced biomass burning from both local and regional transport.Ca 2+ continued to have significant reductions of about 77 % in both PM 2.5 and TSP.Precipitation during the summer study period of 2010 was 149 mm, only half of that in 2009.Thus, stringent control measures on the construction works and road dust should be the major cause of decreased Ca 2+ levels.
The comparison of results in the autumn study period were distinctly different from those of spring and summer.As shown in Fig. 10e and f, all the ion species unexceptionally increased in 2010 compared to 2009.Among the secondary tributed to two main causes.On the one hand, since 2006, more strict controls were implemented on the coal combustion emission, including reduction of the proportion of coal in the energy mix, close and/or replacement of inefficient and dirty coal-fired power plants, and installation of FGD devices for all the coal-fired stations of over 10 GW capacities in Shanghai.As a result, annual SO 2− 4 concentrations had a significant decreasing trend as shown in the table.In the meanwhile, the annual growth rate of vehicle stocks of Shanghai reached over 12 % since 2000 (UNEP, 2009).NO 2 vertical column density from space over YRD continued to increase in the recent 5 yr (Wang and Tian, 2010).From the ground measurement results in Table 1, PM 2.5 nitrate also showed a slightly increasing trend.Thus, the change of emission sources is reflected in the increased NO − 3 /SO 2− 4 ratios.Due to the expansion of the transportation system, NO x emissions were projected to increase by 60-70 % by 2020 (Chen et al., 2006), suggesting that nitrogen emission had the potential to be the prior pollutant in megacities of China.
We conducted three air quality campaigns before, during, and after the 2010 World Expo in Shanghai.Trace gases, aerosol chemical components, and major meteorological factors were measured.The results showed the response of secondary aerosol components to both the control measures and the human activities during the expo.In spring, the most severe pollution episode was caused by a floating dust originating from northwestern China on 26-28 April, right before the opening of the expo.A comparison to the similar period of 2009 found that floating dust was a common phenomenon impairing the air quality of eastern China in spring.A significant increasing trend of SNA (SO 2− 4 , NO − 3 , and NH + 4 ) concentrations was observed from 22 April to 2 May, which was attributed to the enhanced human activities as the expo was approaching.Nitrate had the most significant daily increasing rate of 1.1 µg m −3 d −1 due to enhanced vehicle emission.In summer, two intensive pollution episodes were found to be a mixed pollution of SNA with biomass burning due to loose control of post-harvest straw burning.In the autumn phase of the expo, before the closing of the expo (20 to 30 October), the air quality over Shanghai was much better than ever before.However, the air quality rapidly plummeted as soon as the expo was announced closed.SNA increased 3-6 fold to be 42.1 and 68.2 µg m −3 on 31 October and 1 November, respectively, as compared to 20-30 October.Of which, nitrate increased most ∼5-8 fold, indicating the serious impact from enhanced vehicle emission.The anthropogenic Ca as a tracer from construction dust increased from 2.88 ± 1.85 µg m −3 during the expo to 6.98 ± 3.19 µg m −3 during the post-expo period, attributed to the resumption of construction works after the expo.No successive control measures and loose regulations after the expo were responsible for this jump to bad quality.
Compared to the spring and summer of 2009, NO − 3 increased 12-15 % while SO 2− 4 showed reductions of 15-30 % in 2010.Continuous desulfurization of SO 2 emission from power plants in recent years was responsible for the lowered SO 2− 4 , while enhanced traffic emission due to tremendous number of expo visitors was the major contributor to the increased NO − 3 .Compared to the autumn in 2009, all the ion components increased in 2010 owing to the lifting of emission control measures after the expo.SO 2− 4 was found to have a lower increase while NO − 3 had a significant increase of 150 %.For the first time, we found out the mass concentration of NO − 3 exceeded that of SO 2− 4 in Shanghai during certain periods of the expo.Reducing NO x emission will be China's priority in the future in order to improve the air quality over the megacity.In spring and summer, Ca 2+ had the most significant reduction among all the ions.Prohibition of construction works and frequent cleaning of the traffic roads were testified as effective in lowering the mineral aerosol levels.The neutralization ability of Ca 2+ on acids was estimated to decrease by 7-17 %.However, the resumption of construction works and enhanced traffic flows after the expo made Ca 2 + gain a tremendous increase of 320 %.It was suggested that controlling the emission of mineral aerosol was also beneficial for the alleviation of air pollution in China.
During the 2010 expo, apparent improvement of air quality was achieved, which was attributed to that the Shanghai municipal government has implemented a series of control strategies including stepwise, long-term, region-wide and emergency measures (Table S1).The growth of total energy consumption for the industrial, transport and building sectors was controlled and the energy efficiency had improved a lot.Use of natural gas, installation of wind power facilities and imported electricity from neighboring provinces (e.g., Anhui Province, and electricity generated by the Three Gorges Hydropower Station) resulted in less coal combustion (UNEP, 2009).Some strict emission control measures, e.g., emission control from power plants, vehicle flows, and ban of burning straws, did take effect during the expo.However, this kind of strict emission control measures did not last long, and it was only given to a specific event in China, such as the Beijing Olympic Games (Zhang et al., 2009), the Shanghai Expo, and the Guangzhou Asian Games (Liu et al., 2012).We suggest that the government should abandon the short-sighted attitude of only considering some specific events.Instead, a long-term effort is required for the improvement and sustainability of air quality in megacities of China.

Figure. 1 .Fig. 1 .
Figure.1.Map of Shanghai with district borders.The area of the Expo campus is shown by t polygon, and the monitoring sites at Fudan and Weifang are denoted by the black stars (This fig modified based on the map from http://en.wikipedia.org/wiki/Shanghai). 841 Fig. 1.Map of Shanghai with district borders.The area of the expo campus is shown by the red polygon, and the monitoring sites at Fudan and Weifang are denoted by the black stars (This figure is modified based on the map from http://en.wikipedia.org/wiki/Shanghai).

Figure 2 .
Figure 2. Time-series of hourly PM2.5 and PM10 concentrations (μgm -3 ) measured in Shanghai during 2010.The study period spans from spring (April 2 -May 14), summer (July 25 -August 24) to autumn (October 20 -November 29) as marked by the blue lines in the figure.Pre-Expo (before May 1) and post-Expo (after November 30) periods are separated from the Expo period in the figure marked by the black lines.

Fig. 2 .
Fig. 2. Time series of hourly PM 2.5 and PM 10 concentrations (µg m −3 ) measured in Shanghai during 2010.The study period spans from spring (2 April-14 May) to summer (25 July-24 August) and autumn (20 October-29 November) as marked by the blue lines in the figure.Pre-expo (before 1 May) and post-expo (after 30 November) periods are separated from the expo period in the figure marked by the black lines.

Figure 3 .
Figure 3. Time-series of hourly PM10 concentrations and daily Al concentrations in TSP (AlTSP) during the spring study period from April 2 to May 12 in 2009 (a) and 2010 (b), respectively.

Fig. 3 .
Fig. 3. Time series of hourly PM 10 concentrations and daily Al concentrations in TSP (Al TSP ) during the spring study period from 2 April to 12 May in 2009 (a) and 2010 (b), respectively.

Figure 5 .
Figure 5. Same as Figure 4 but for the summer of 2010.

Figure 6 .
Figure 6.Hourly CO concentration (μgm -3 ), hourly biomass carbon emission (tons/hr) from the FLAMBE biomass burning emission inventory in the domain of 28 -32 °N, 119 -123°E in Eastern China with daily K + concentration in PM2.5 during the summer of 2010.

Fig. 6 .
Fig. 6.Hourly CO concentration (µgm −3 ), hourly biomass carbon emission (tons h −1 ) from the FLAMBE biomass burning emission inventory in the domain of 28-32 • N, 119-123 • E in eastern China with daily K + concentration in PM 2.5 during the summer of 2010.

Figure 8 .
Figure 8. Same as Figure 4 but for the autumn of 2010 and 2009, respectively.

Figure 9 .
Figure 9. Time-series of hourly PM10 concentrations and daily Al concentrations in TSP (AlTSP) during the autumn study period from October 20 to November 19 in 2009 (a) and 2010 (b), respectively.

Fig. 9 .
Fig. 9. Time series of hourly PM 10 concentrations and daily Al concentrations in TSP (Al TSP ) during the autumn study period from 20 October to 19 November in 2009 (a) and 2010 (b), respectively.

Figure 10 .
Figure 10.(a,c,e) The seasonal average concentrations of Cl -, NO 3 -, SO 4 2-, Na + , NH 4 + , K + , and Ca 2+ in PM 2.5 and TSP during 2009 and 2010.(b,d,f) The seasonal average percentage changes of the ions referred above in 2010 relative to 2009.

Table 1 .
Historical (and present study) concentrations of SO 2− 4 and NO − 3 with the ratio of NO −