This content has been downloaded from IOPscience. Please scroll down to see the full text.

Download details:

IP Address: 210.72.27.67

This content was downloaded on 20/05/2016 at 10:21

Please note that terms and conditions apply.

Did the widespread haze pollution over China increase during the last decade? A satellite

view from space

View the table of contents for this issue, or go to the journal homepage for more

2016 Environ. Res. Lett. 11 054019

(http://iopscience.iop.org/1748-9326/11/5/054019)

Home Search Collections Journals About Contact us My IOPscience

Environ. Res. Lett. 11 (2016) 054019 doi:10.1088/1748-9326/11/5/054019

LETTER

Did the widespread haze pollution over China increase during thelast decade? A satellite view from space

Minghui Tao1,2, LiangfuChen1,2, ZifengWang1,2, JunWang3, JinhuaTao1,2 andXinhuiWang1

1 State Key Laboratory of Remote Sensing Science, Institute of Remote Sensing andDigital Earth, Chinese Academy of Sciences, Beijing100101, People’s Republic of China

2 Joint Center forGlobal Change Studies, Beijing 100875, People’s Republic of China3 Department of Earth&Atmospheric Sciences, University of Nebraska—Lincoln, NE 68588,USA

E-mail: chenlf@radi.ac.cn andwangzf@radi.ac.cn

Keywords: haze, aerosol,MODIS, China

Supplementarymaterial for this article is available online

AbstractWidespread haze layers usually cover China like low clouds, exertingmarked influence on airquality and regional climate.With recent Collection 6MODISDeep Blue aerosol data in2000–2015, we analyzed the trends of regional haze pollution and the corresponding influence ofatmospheric circulation in China. Satellite observations show that regional haze pollution ismainlyconcentrated in northern and central China. The annual frequency of regional haze in northernChina nearly doubles between 2000 and 2006, increasing from 30–50 to 80–90 days. Though thereis amarked decrease in annual frequency during 2007–2009 due to both reduction ofanthropogenic emissions and changes ofmeteorological conditions, regional pollution increasesslowly but steadily after 2009, andmaintains at a high level of 70–90 days except for the suddendecrease in 2015. Generally, there is a large increase in the number of regional-scale haze eventsduring the last decade. Seasonal frequency of regional haze exhibits distinct spatial and temporalvariations. The increasing winter haze events reach a peak in 2014, but decrease strongly in 2015due partly to synoptic conditions that are favorable for dispersion. Trends of summer regional hazepollution aremore sensitive to changes of atmospheric circulation. Our results indicate that thefrequency of regional haze events is associated not only with the strength of atmosphericcirculation, but also with its direction and position, as well as variations in anthropogenicemissions.

1. Introduction

Under the background of large increase in anthropo-genic emissions during the last decades, haze pollutionhas been a common problem over China with thedegradation of visibility and air quality (Zhanget al 2012, Han et al 2014). Satellite observations showthat haze pollutionnot only exists in the urban regions,but also usually appears in the extensive rural areas,which form widespread haze layers like low cloudsover China (Tao et al 2012). Such regional heavyaerosol loading not only threatens public health nearthe surface, but also exerts marked influences onradiation budget and cloud properties (Fan et al 2008,Li et al 2011). In addition, thick haze layers can further

aggravate air pollution near surface by radiative feed-back (Gao et al 2015).

Recently, several extreme haze events over easternChina have led to wide concern due to their extensivespatial coverage and record-breaking particle pollu-tion (Tao et al 2013, Yang et al 2013, Wang et al 2014).Observations at meteorological stations shows a largeincreasing trend in haze days in China since 2000 (Quet al 2015). Moreover, interannual trends of the hazedays have a close relationship with regional circulation(Cao et al 2015). However, visibility mainly reflectsinformation of horizontal extinction near surface,which can be impacted by local emissions andmeteor-ology, especially in urban regions. By contrast, colum-nar extinction of aerosols exhibited different spatial

OPEN ACCESS

RECEIVED

21March 2016

REVISED

5May 2016

ACCEPTED FOR PUBLICATION

9May 2016

PUBLISHED

19May 2016

Original content from thisworkmay be used underthe terms of the CreativeCommonsAttribution 3.0licence.

Any further distribution ofthis workmustmaintainattribution to theauthor(s) and the title ofthework, journal citationandDOI.

© 2016 IOPPublishing Ltd

characteristics (Tao et al 2015, Qu et al 2016). Forma-tion and variation of the thick haze layers is mostlyconnected with regional transport and moist airflows(Tao et al 2014), which can be more sensitive to varia-tions of atmosphere circulation. So far, it is still notclear how these common haze plumes change overspace and time as well as their connection with regio-nal circulation.

The recent Collection 6 (C6)Moderate ResolutionImaging Spectrodadiometer (MODIS) Deep Blue(DB) aerosol retrieval has expanded to entire landareas (Hsu et al 2013), and performs well in the heavyaerosol loading conditions of China (Tao et al 2015).Based on the daily DB aerosol optical depth (AOD) in2000–2015 and meteorological data, we investigatedinterannual variations of the haze clouds over China,and analyzed the potential influence of regional atmo-spheric circulation.

2.Data andmethods

By measuring reflected and emitted radiance of theEarth-atmospheric system in a broad spectral rangewith 36 bands between 0.4–14.4 μm, the MODISsensor can provide near global detection of diverseatmospheric parameters with a wide swath ∼2330 kmand relatively fine resolution at 250–1000 m. MODIShas been flying on Terra satellite since 2000, and onAqua satellite since 2002, with equatorial overpassingtimes around 10:30 am and 13:30 pm, respectively.

The MODIS aerosol retrieval is first realized overdense vegetation regions utilizing the linear

relationship between visible and shortwave bands(Levy et al 2007). The MODIS DB algorithm isdesigned and implemented to obtain aerosol informa-tion over bright-reflecting regions by assuming thatsurface reflectance in near-UV blue channel (0.41 μm)is much lower than those in longer visible bands (Hsuet al 2006). By considering scattering angles and vege-tation density in the estimate of surface reflectance, theC6 DB retrieval has extended to all cloud-free and ice/snow-free land areas (Hsu et al 2013).

MODIS aerosol data is provided at a normal spa-tial resolution of 10 × 10 km2 in Level 2 (L2) data(MOD04 for Terra and MYD04 for Aqua), and with acoarser resolution of 1° × 1° in aggregated L3 pro-ducts (https://ladsweb.nascom.nasa.gov/data/search.html). In addition to this, associated para-meters such as cloudmask and quality assurance (QA)are also included in L2 aerosol data for examining ori-gins of retrieval uncertainties. Ground-based evalua-tion shows that the expected error envelope of C6 DBAOD is generally within ± (0.03 + 0.2AODMODIS)(Sayer et al 2013). Since DB aerosol retrieval is avail-able in both bright and dark surfaces, DB AOD canwell reveal spatial coverage of the regional haze pollu-tion in China (Tao et al 2015). To further interpretspatial and temporal variation of the haze pollution,wind fields from the National Centers for Environ-mental Prediction (NCEP) Reanalysis were used toanalyze variations of meteorological conditions(http://www.esrl.noaa.gov/psd/data/reanalysis/reanalysis.shtml).

Figure 1. (a)AquaMODIS true color image of the extensive haze pollution inChina on 4 January 2014; (b)MODISDeep Blue AODat550 nmon the same day; (c) terrain of themain land of China; (d) annual average AOD in 2015.

2

Environ. Res. Lett. 11 (2016) 054019

3. Results and discussion

3.1. Satellite view of thewidespread haze pollutionoverChinaLarge-scale haze pollution, as shown in figure 1(a),often appears over China. Such dense haze layers notonly exist in urban regions, but also cover extensiverural areas. Satellite observations show that AOD at0.55 μmof the haze layers is far above 1.0 inmost areas(figure 1(b)). A haze day is usually defined as a day withlow visibility <10 km and relative humidity <80%near surface. Here we mainly focus on regionalvariations of heavy aerosol loading rather than theusual haze with, low visibility observed from groundsites. Though there is no fixed connection betweenAOD and visibility, high regional AOD is a goodindicator of serious air pollution (tao et al (2013), Jiaet al 2015) and various past studies have shown thatlow visibility is commonly associated with high AODexcept in cases of high attitude aerosol layer (Kessneret al 2013). Considering that AOD of the large-scalehaze plumes usually far exceed 1.0, we selectAOD > 1.0 as threshold of regional haze pollution inthis study.

The annual average value of AOD is >0.6 in mostareas of eastern China, and exceeds 0.8 in the mainpolluted region of northern China (figure 1(d)). To geta direct view of temporal variations of the haze pollu-tion, we count linear variations of spatial mean valuesof the frequency of AOD > 1.0 in the main pollutedregion of northern China (figure 2). Frequency fromTerra AOD is slightly higher than from Aqua due todiurnal changes of aerosol loading and cloud amount.Ground observations show that aerosol loading in themorning is usually higher than in the afternoon innorthern China (Kuang et al 2015). Meanwhile, theretends to be somewhat more clouds over land in theafternoon (King et al 2013). Considering that AquaMODIS has a higher accuracy in calibration thanTerra, Aqua data is used when it is available after 2002.

Frequency of high AOD in summer is consistently lar-ger than in winter, which can be mostly due to intensehygroscopic growth of aerosol particles under com-mon moist weather conditions in summer (Quet al 2016).

Figure 3 shows the annual frequency of regionalhaze pollution in China during 2000–2015. Occur-rence of large-scale haze events is mainly concentratedin northern and central China, where most anthro-pogenic emissions are located (Lu et al 2011). Highaerosol loading also exists in the cloudy Sichuan Basin,which can be related to the moist and foggy environ-ment. There are generally 30–40 days with AOD> 1.0in the Taklimakan Desert every year. Despite the sametrends in figures 2 and 3, spatial average obviouslydecreases the peak values of frequency due to the inho-mogeneous distribution of haze pollution.Occurrenceof large-scale haze events in the main polluted regionsof northern China nearly doubled during 2000–2006,increasing from 30–50 days to 80–90 days. Althoughthere is a marked decrease during 2007–2009, regionalpollution increases slowly but steadily after 2009, andmaintains at a high level of 70–90 days except suddendecrease in 2015. The annual available days of havingAqua MODIS DB aerosol retrieval in northern Chinaranges around 180 days (Tao et al 2015), indicatingthat the regional heavy haze pollution nearly accountsfor half of the cloud-free days.

Despite the increase of anthropogenic emissionssuch as carbonaceous aerosol, decline in sulfur dioxide(SO2) emissions in China appears from 2006 (Luet al 2011), which may partly explain the decrease ofhaze pollution. The temporary decline in nitrogen oxi-des (NOx) emissions in China during 2009 due to eco-nomic downturn can further promote the reductionof regional haze (Lin and McElroy (2011)). When thereduction of SO2 emissions are offset by the increase ofother pollutants such as NOx and carbonaceous aero-sol, frequency of the haze pollution can rise again.Compared with the decrease of annual frequency in

Figure 2.Average value of (a) annual frequency (number of days per year) ofMODISDeep Blue AOD> 1.0 during 2000–2015 and (b)seasonal frequency inwinter and summer during 2001–2015 (Aqua data used after 2002) in northernChina (113.5E-118.5E, 40.5N-34.5N, the area of the red rectangle in figure 1(d)).

3

Environ. Res. Lett. 11 (2016) 054019

2007, increase in haze pollution during winter andsummer indicates that meteorological conditions canhave also an important contribution.

3.2. Interannual variation of regional haze inwinterand summerTo further examine annual variations of the regionalhaze, we analyzed the seasonal frequency of high AODin winter and summer. Figure 4 shows changes intrends of AOD> 1.0 in winter during 2001–2015. The

winter of one year here denotes months of January,February, and December of last year. Despite of slightdifferences, as shown in figure 2, annual trends of thewinter haze are generally consistent with that of theyearly frequency. By comparison, regional character-istics of seasonal heavy pollution are clearer, withnotable spatial changes in major haze areas. The highfrequency of haze pollution during the year tends to beconcentrated in northwestern part of the northernChina, where the industry belt is located (Lu

Figure 3.Annual frequency (number of days per year) ofMODISDeep Blue AOD> 1.0 during 2000–2015.

2001 2002 2003 2004

2005 2006 2007 2008

2009 2010 2011 2012

2013 2014 2015 <Value>0 - 11 - 55 - 1010 - 1515 - 2020 - 2525 - 30

Figure 4. Frequency ofMODISDeep Blue AOD> 1.0 during thewinter of 2001–2015.

4

Environ. Res. Lett. 11 (2016) 054019

et al 2011). During the winter, haze areas are moreextensive, and major regions move southwardobviously in 2013 and 2014, which can be caused bysouthward transport. Themost polluted period occursin the winter of 2014, with heavily hazy days innorthern China reaching one month. However, regio-nal winter haze pollution declines substantially in2015, due largely to changes in atmospheric circula-tion (seen in section 3.3).

Figure 5 shows the trends of regional haze eventsin summer during 2001–2015. Similar to the annualfrequency, summer haze is mainly concentrated in thenorthern part of northern China. The amount ofregional haze in summer generally increases, but itstrend is different as compared to the annual frequency(figure 2). Comparedwith the low frequency inwinter,there is a sudden peak of summer haze events during2002 due mostly to the variations of the Asian mon-soon. Then summer haze increases from 2004, andreached high values during 2007 and 2008. Despite ofan obvious decrease in 2009, the most summer hazeevents occurred in 2010 and 2011, with the number ofheavy haze pollution exceeding one month. Thoughmajor haze areas in northern China get smaller in2012–2014, the frequency of summer haze is still athigh levels.

The frequency of high AOD in summer is muchlarger than in winter. Severe particle pollution innorthern China usually appears in winter associatedwith low boundary layer and more coal burning forheating. Although strong convection in summer canreduce particle concentration near surface, agri-cultural biomass burning and intense moist airflowsgreatly enhance the columnar extinction of the

elevated aerosol layers. The number of low visibilityevents in meteorological sites exhibit drastic annualchanges in winter (Cao et al 2015), which is consistentwith the large yearly spatial variations of major hazeareas. Different from the slow growth in low visibilityevents near the surface, the annual frequency of regio-nal dense haze layers varies largely in northern Chinain summer, indicating that elevated aerosols are moresensitive to changes of the atmospheric circulation.Despite of the abrupt large decrease in 2015, variationsof regional haze in winter are generally consistent withthat of anthropogenic emissions (Lu et al 2011). Cloudamount can impact comparability of the haze fre-quency, but its annual variation is slight (Maet al 2014).

3.3. Potential influence of atmospheric circulationon regional haze inChinaBesides changes of anthropogenic emissions, themarked interannual variations of regional haze pollu-tion in northern China can be largely associated withchanges in atmospheric circulation. Several studieshave shown that variations of the visibility are closelyrelated to atmospheric circulation such as the El NiñoSouthern Oscillation (ENSO) or the Asian monsoon(Cao et al 2015, Li et al 2016). However, the strength ofsuch circulation in certain years is uncorrelated withthe frequency of hazy days. Different from previousstudies that focus on the general relationship betweenvisibility and atmosphere circulation in seasonal scale,here we examine the role of atmospheric circulationby comparing and analyzing meteorological condi-tions in specific years with large difference in hazefrequency.

2001 2002 2003 2004

2005 2006 2007 2008

2009 2010 2011 2012

2013 2014 2015 <Value>0 - 11 - 55 - 1010 - 1515 - 2020 - 2525 -30>30

Figure 5. Frequency ofMODISDeep Blue AOD> 1.0 during the summer of 2001–2015.

5

Environ. Res. Lett. 11 (2016) 054019

As shown in figure 6(a), dry and cold north-westerly winds prevail in northern China during win-ter, which can blow the anthropogenic pollutantsaway. Recent studies emphasize that the weakening ofthe east Asian winter monsoon favors increase of fogand haze (Qu et al 2013, Li et al 2016). Regional hazepollution in winter reaches its peak of the last decadein 2014, with haze pollution exceeding 25 days inmostareas of northern China (figure 5). By contrast, there isa large decrease in the amount of haze events in 2015.In general, regional circulation gets slower in the win-ter of 2014with notable strengthening of southerly air-flows in January and February (figure 6(b)). Duringthe same period in 2015, northern winds weakenslightly in northern China in January and becomestronger in February. Despite the similar monthlywind fields, there are large differences in the frequencyof regional haze. Enhancement of the southwesterlywinds in January 2014 is accompanied withmore hazeevents than that of the southeastern direction in Feb-ruary. Though northern winds strengthen in Januaryof both 2014 and 2015, much more haze events occurin northern China when the winds originate fromnortheastern China rather than northwestern China.Therefore, the influence of atmospheric circulation onregional air pollution is associated with several com-plex factors besides its strength.

Southerlymoist andwarm airmasses are prevalentin northern China during summer (figure 7(a)).Model simulations suggest that the weakening of theeastern Asian summer monsoon can increase the

aerosol concentration in eastern China (Zhuet al 2012). However, satellite observations show thatregional aerosol loading increases largely in northernChina when southeasterly winds strengthen(figure 7(c)). By comparison, when strong south-western airflows appear in August 2011, there are noobvious changes in the amount of haze events. Severalfactors such as changes in direction and position of themonsoon belt can influence the spatial distributionand transport of anthropogenic pollutants. Whensoutheasterly humid airflows from the ocean prevail innorthern China, dispersion of regional aerosols andgaseous precursors can be obstructed by the moun-tains in the north and west of the North China Plain(figure 1(c)). Similar results also exist in 2008 and 2009(seen in figures 1S and 2S in the supplementary mat-erial), indicating that changes in atmospheric circula-tion have an important contribution in the decrease ofhaze pollution besides reduction of anthropogenicemissions.

Extensive thick haze layers can linger over north-ern China for nearly one week before northern orsouthern winds become strong enough to blow themaway (Tao et al 2014). Satellite observations show thatprevalent dust plumes from deserts in northwesternChina can mix with anthropogenic pollutants andcause regional pollution (Tao et al 2012). Formation ofregional heavy haze can be driven by several factorssuch as southerly moist airflows and northwesterlydust transport, which varies with seasons and loca-tions. Connection between air pollution and

Figure 6. (a)Average ofmonthly winds and geopotential height (m) at 850 hPa during thewinter of 2001–2015; departure betweenmonthly wind and above average at 850 hPa and frequency ofMODISDeep Blue AOD> 1.0 in thewinter of (b) 2014 and (c) 2015.

6

Environ. Res. Lett. 11 (2016) 054019

atmospheric circulation is usually analyzed in yearly orseasonal scales. The factual circulation for one seasonis composed of numerous processes, which can be verydifferent in conditions such as origin, wind direction,locations of the air masses. When the circulationmainly includes air masses from opposite directions,seasonal or monthly mean value of meteorologicaldata may consider it as stagnant weather even com-mon regional transport exist. To further understandthe role of atmospheric circulation in regulating varia-tions of regional haze pollution, more detailed studiesare needed.

4. Conclusion

The extensive haze pollution in China has receivedwide attention due to its marked influence on airquality and public health. By analyzing MODIS DeepBlue AOD in 2000–2015 and associated meteorologi-cal data, we provide the first large-scale insight into thetrends of the common regional haze pollution inChina and their connection with atmospheric circula-tion. Frequent regional haze pollution is found innorthern and central China. During the last decade,regional haze pollution in northern China generallyincreases from 30–50 to 80–90 days, which almostaccounts for half of the cloud-free days. However,there is a decrease in the amount of haze pollutionfrom 2007 to 2009 due to reductions in anthropogenicemissions and changes of meteorological conditions.

In addition, a sudden large decrease appears in 2015,which can be mostly attributed to variations of atmo-spheric circulation.

Compared with the annual frequency, regionalhaze pollution in winter and summer exhibits muchlarger temporal and spatial variations. The frequencyof regional winter haze pollution reaches peak in 2014with a sudden decline in 2015 due partly to thestrengthening of cold and clean northern winds.Despite similar trends with the annual amount, majorhaze areas of the winter haze pollution has notablespatial changes caused by reginal transport. By con-trast, frequency of the summer haze pollution hasnotable temporal variations. Seasonal variations of theamount of regional haze events exhibit sensitiveresponse to changes in atmosphere circulation, espe-cially in summer. Our results demonstrate that direc-tion and position of the atmospheric circulation canalso play an important role in its influence on regionalair quality besides its strengths.

Acknowledgments

This study was supported by National Science Foun-dation of China (Grant No. 41401482) and the Type BLeading Special Project of CAS (XDB05020100). Wethank theMODIS team and NCEP Reanalysis Deriveddata provided by the NOAA/OAR/ESRL PSD for thedata used in ourwork.

Figure 7. (a)Average ofmonthly winds and geopotential height (m) at 850 hPa during the summer of 2001–2015; departure betweenmonthly wind and above average at 850 hPa and frequency ofMODISDeep Blue AOD> 1.0 in the summer of (b) 2004 and (c) 2011.

7

Environ. Res. Lett. 11 (2016) 054019

References

CaoZ, Sheng L, LiuQ, YaoX andWangW2015 Interannualincrease of regional haze-fog inNorthChina Plain in summerby intensified easterly winds and orographic forcingAtmos.Environ. 122 154–62

Fan J, Zhang R, TaoW-K andMohrK I 2008 Effects of aerosoloptical properties on deep convective clouds and radiativeforcing J. Geophys. Res. Atmos. 113D08209

GaoY, ZhangM, Liu Z,Wang L,Wang P, Xia X, TaoMandZhuL2015Modeling the feedback between aerosol andmeteorological variables in the atmospheric boundary layerduring a severe fog–haze event over theNorthChina PlainAtmos. Chem. Phys. 15 4279–95

HanX, ZhangM,Gao J,Wang S andChai F 2014Modeling analysisof the seasonal characteristics of haze formation in BeijingAtmos. Chem. Phys. 14 10231–48

HsuNC, JeongM J, BettenhausenC, Sayer AM,Hansell R,Seftor C S,Huang J andTsay SC 2013 EnhancedDeep Blueaerosol retrieval algorithm: the second generation J. Geophys.Res. Atmos. 118 9296–315

HsuNC, Si-Chee T, KingMDandHerman J R 2006Deep blueretrievals of Asian aerosol properties duringACE-Asia IEEETrans. Geosci. Remote 44 3180–95

Kessner A,Wang J, Levy R andColarco P 2013Remote sensing ofsurface visibility from space: a look at theUnited States EastCoastAtmos. Environ. 81 136–47

KingMD, Platnick S,MenzelWP, Ackerman SA andHubanks PA2013 Spatial andTemporal Distribution of CloudsObservedbyMODISOnboard the Terra andAqua Satellites IEEETrans.Geosci. Remote 51 3826–52

KuangY, ZhaoC, Tao J andMaN2015Diurnal variations of aerosoloptical properties in theNorthChina Plain and theirinfluences on the estimates of direct aerosol radiative effectAtmos. Chem. Phys. 15 5761–72

LevyRC, Remer LA,Mattoo S, Vermote E F andKaufmanY J 2007Second-generation operational algorithm: retrieval of aerosolproperties over land from inversion ofmoderate resolutionimaging spectroradiometer spectral reflectance J. Geophys.Res. Atmos. 112D13211

LiQ, Zhang R andWangY 2016 Interannual variation of thewintertime fog–haze days across central and easternChinaand its relationwith East Asianwintermonsoon Int. J.Climatol. 36 346–54

Lin J T andMcElroyMB2011Detection from space of a reductionin anthropogenic emissions of nitrogen oxides during theChinese economic downturnAtmos. Chem. Phys. 11 8171–88

LuZ, ZhangQ and StreetsDG2011 Sulfur dioxide and primarycarbonaceous aerosol emissions inChina and India,1996–2010Atmos. Chem. Phys. 11 9839–64

Ma J,WuH,WangC, ZhangX, Li Z andWangX 2014Multiyearsatellite and surface observations of cloud fraction over ChinaJ. Geophys. Res. Atmos. 119 7655–66

QuW,Wang J, Gao S andWuT2013 Effect of the strengthenedwestern Pacific subtropical high on summer visibilitydecrease over easternChina J. Geophys. Res. Atmos. 1187142–56

QuW,Wang J, ZhangX, Sheng L andWangW2016Oppositeseasonality of the aerosol optical depth and the surfaceparticulatematter concentration over the northChina PlainAtmos. Environ. 127 90–9

QuW,Wang J, ZhangX, Yang Z andGao S 2015 Effect of cold waveonwinter visibility over eastern China J. Geophys. Res. Atmos.120 2394–406

Sayer AM,HsuNC, BettenhausenC and JeongM J 2013Validation and uncertainty estimates forMODIS Collection6 ‘Deep Blue’ aerosol data J. Geophys. Res. Atmos. 1187864–72

TaoM,Chen L, Su L andTao J 2012 Satellite observation of regionalhaze pollution over theNorthChina Plain J. Geophys. Res.Atmos. 117D12203

TaoM, Chen L, Xiong X, ZhangM,Ma P, Tao J andWang Z 2014Formation process of the widespread extreme hazepollution over northern China in January 2013:implications for regional air quality and climateAtmos.Environ. 98 417–25

TaoM, Chen L,Wang Z, Tao J, CheH,Wang X andWang Y 2015Comparison and evaluation of theMODISCollection 6aerosol data in China J. Geophys. Res. Atmos. 1206992–7005

TaoM,Chen L,Wang Z, Tao J and Su L 2013 Satellite observation ofabnormal yellow haze clouds over East China during summeragricultural burning seasonAtmos. Environ. 79 632–40

WangY, Yao L,Wang L, Liu Z, Ji D, TangG, Zhang J, Sun Y,HuB andXin J 2014Mechanism for the formation of theJanuary 2013 heavy haze pollution episode over central andeasternChina Sci. China Earth Sci. 57 14–25

YangK,DickersonRR, Carn SA, GeC andWang J 2013 Firstobservations of SO2 from the satellite SuomiNPPOMPS:widespread air pollution events over ChinaGeophys. Res. Lett.40 4957–62

ZhangXY,Wang YQ,Niu T, ZhangXC,Gong S L, ZhangYMandSun J Y 2012Atmospheric aerosol compositions inChina:spatial/temporal variability, chemical signature, regionalhaze distribution and comparisons with global aerosolsAtmos. Chem. Phys. 12 779–99

Zhu J, LiaoH and Li J 2012 Increases in aerosol concentrationsover eastern China due to the decadal-scale weakening ofthe East Asian summermonsoonGeophys. Res. Lett. 39L09809

8

Environ. Res. Lett. 11 (2016) 054019