NASA Technical Reports Server (NTRS) 20140005403: New Approach to Monitor Transboundary Particulate Pollution over Northeast Asia

0.67 MB·English

by  NASA Technical Reports Server (NTRS)

Atmos.Chem.Phys.,14,659–674,2014 Atmospheric O
p
www.atmos-chem-phys.net/14/659/2014/ e
Chemistry n
doi:10.5194/acp-14-659-2014 A
c
and Physics c
©Author(s)2014.CCAttribution3.0License. e
s
s
New approach to monitor transboundary particulate pollution over
Northeast Asia
M.E.Park1,C.H.Song1,R.S.Park1,J.Lee2,3,4,J.Kim2,S.Lee1,J.-H.Woo5,G.R.Carmichael6,T.F.Eck4,7,
B.N.Holben4,S.-S.Lee8,C.K.Song9,andY.D.Hong9
1SchoolofEnvironmentalScienceandEngineering,GwangjuInstituteofScienceandTechnology(GIST),Gwangju,
500-712,Korea
2DepartmentofAtmosphericSciences,YonseiUniversity,Seoul,120-749,Korea
3EarthSystemScienceInterdisciplinaryCenter,UniversityofMaryland,CollegePark,Maryland,USA
4LaboratoryforTerrestrialPhysics,GoddardSpaceFlightCenter,NationalAeronauticsandSpaceAdministration(NASA),
Greenbelt,Maryland,USA
5DepartmentofAdvancedTechnologyFusion,KonkukUniversity,Seoul,143-701,Korea
6DepartmentofChemicalandBiochemicalEngineering,UniversityofIowa,IowaCity,Iowa,USA
7UniversitiesSpaceResearchAssociation,Columbia,Maryland,USA
8NationalInstituteofMeteorologicalResearch,Seoul,156-720,Korea
9AirQualityResearchDepartment,NationalInstituteofEnvironmentalResearch(NIER),Incheon,404-170,Korea
Correspondenceto:C.H.Song([email protected])
Received:29March2013–PublishedinAtmos.Chem.Phys.Discuss.:14June2013
Revised:29November2013–Accepted:3December2013–Published:22January2014
Abstract. A new approach to more accurately monitor and pared to background average AOD (aerosol optical depth)
evaluate transboundary particulate matter (PM) pollution is at the four AERONET sites in Korea, and average AOD in-
introduced based on aerosol optical products from Korea’s creases of 121% were found when averaged over the entire
Geostationary Ocean Color Imager (GOCI). The area stud- Korean peninsula. This paper demonstrates that the use of
ied is Northeast Asia (including eastern parts of China, the multi-spectral AOD retrievals from geostationary satellites
Korean peninsula and Japan), where GOCI has been mon- canimproveestimatesoftransboundaryPMpollution.Such
itoring since June 2010. The hourly multi-spectral aerosol data will become more widely available later this decade
optical data that were retrieved from GOCI sensor onboard when new sensors such as the GEMS (Geostationary Envi-
geostationary satellite COMS (Communication, Ocean, and ronment Monitoring Spectrometer) and GOCI-2 are sched-
Meteorology Satellite) through the Yonsei aerosol retrieval uledtobelaunched.
algorithm were first presented and used in this study. The
GOCI-retrievedaerosolopticaldataareintegratedwithesti-
mated aerosol distributions from US EPA Models-3/CMAQ
(Community Multi-scale Air Quality) v4.5.1 model simu- 1 Introduction
lations via data assimilation technique, thereby making the
aerosoldataspatiallycontinuousandavailableevenforcloud Transboundary air pollution is an important issue in Asia
contaminationcells.Theassimilatedaerosolopticaldataare (Arndt et al., 1998; Holloway et al., 2002; Yamaji et al.,
utilized to provide quantitative estimates of transboundary 2012). In particular, the long-range transport (LRT) of air
PMpollutionfromChinatotheKoreanpeninsulaandJapan. pollutants from China (source) to the Korean peninsula and
Fortheperiodof1Aprilto31May,2011thisanalysisyields Japan (receptors) due to persistent westerlies over North-
estimates that AOD as a proxy for PM2.5 or PM10 during east Asia has been a central issue in this region (Arndt et
long-range transport events increased by 117–265% com- al.,1998;SongandCarmichael,2001;Yamajietal.,2012).
Transboundary air pollution is a serious issue not only in
PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion.
660 M.E.Parketal.:Transboundaryparticulatepollution
Asia,butalsoinEuropeandNorthAmerica,causinganum- theUSAGOESseries(GOES8,13),theJapaneseGMSand
ber of different international problems such as acid precipi- MTSAT, and European SEVIRI have been meteorological
tation, rising levels of hemispheric ground-level ozone, and sensors with only one to two visible channels. The history
regional visibility impairment (Levy II and Moxim, 1987; ofGEOsatellitesensorsfortheretrievaloftheaerosolopti-
Brankov et al., 2003; Holloway et al., 2003; Tulet et al., calpropertiesissummarizedinTable1.
2003;Parketal.,2004;Wangetal.,2009).Tomanagetrans- This situation is changing as new geostationary satel-
boundaryairpollutionbetweencountries,theConventionon lite based sensors are becoming available and will be ex-
Long-Range Transboundary Air Pollution (CLRTAP) under panded by upcoming planned missions. The first hourly
theUNECE(UnitedNationsEconomicCommissionforEu- multi-spectral aerosol data observed from a geostationary
rope) was launched to lessen air pollution through the LRT satellite sensor, GOCI (Geostationary Ocean Color Imager)
in1979(http://www.unece.org/env/lrtap/lrtap_h1.html).The is used in this paper to demonstrate how this new informa-
EMEP(EuropeanMonitoringandEvaluationProgramme)is tion can be used to improve our estimates of LRT of pol-
ascientificeffortestablishedtosupporttheCLRTAP. lutants. Specifically, the level 1 radiance data from GOCI,
Quantitative evaluation of transboundary air pollution re- onboard COMS (Communication, Ocean, and Meteorology
mains a difficult task and currently cannot be done based Satellite), is converted into level 2 AOD products via the
on observations alone due to a lack of explicit source sig- “Yonsei aerosol retrieval algorithm” (the details of the al-
naturesandlackofanappropriateobservationsystem(Mal- gorithm will be discussed in the Methods section) (Lee et
colm et al., 2000; Brankov et al., 2003; Tulet et al., 2003; al.,2010).COMS/GOCIwaslaunchedinJune2010,andthe
Park et al., 2004; Schaub et al., 2005; Wang et al., 2009). COMS/GOCI data center began to release radiance data in
Most current evaluations of transboundary transport of pol- April2011,afteraninitialoperationtest(IOT)period.This
lutantsarebasedonchemistry-transportmodel(CTM)sim- satellite sensor was designed for monitoring ocean colors
ulations(LevyIIandMoxim,1987;Arndtetal.,1998;Hol- around the Korean peninsula, but it can also provide AOD
lowayetal.,2002;Wangetal.,2009;Yamajietal.,2012).By dataforastrategicallyimportantregion,i.e.,NortheastAsia.
controllingtheemissionsfromasourcecountry(e.g.,China) The domain covered by the GOCI sensor shown in Fig. 1
intheCTMsimulations,transboundaryairpollutionfromthe is 2500km×2500km and includes eastern parts of China,
sourcecountrytoreceptorregionshasbeenevaluated.How- theKoreanpeninsula,andJapan.TheGOCI-retrievedAOD
ever, these model-based evaluations have significant limita- products were integrated with CMAQ-simulated AOD data
tionsbecauseofthelargeuncertaintiesintheemissionrates toproducethemoreaccurateaerosoldatasetusedtoinvesti-
of the air pollutants from the source country (or region), gatetheimpactsoftransboundaryPMpollutionbetweenthe
the highly nonlinear nature of atmospheric chemistry, and sourceandreceptorcountriesinNortheastAsia.
uncertainties related to the CTM and meteorological mod-
els (MM). For example, both atmospheric secondary parti-
2 Methods
cle formation and primary particle (e.g., mineral dust) gen-
eration/transportaredifficulttocorrectlysimulate(Songand
The procedure to produce GOCI-constrained AOD product
Carmichael,2001;Songetal.,2005;Volkameretal.,2006;
combines the GOCI level 2 product with AOD predictions
Robinsonetal.,2007).Theseuncertaintiesareamainreason
fromtheUSEPAModel-3/CMAQ(CommunityMulti-scale
for disagreement in source and receptor estimates made by
Air Quality) v4.5.1 model simulations. The CMAQ model
differentcountries,whichhasledtodistrustoftheoutcomes
simulations were performed over East Asia, with emission
ofmodel-basedassessments.
datapreparedappropriatelyforthemodeldomainandmete-
Current efforts to reduce the uncertainty in estimates of
orological fields produced from the PSU/NCAR MM5 (the
long-rangetransportofpollutantsbyintegratingmodelsand
fifth generation meteorological model) simulations. The fi-
observationsthroughdataassimilationarelimitedbyourcur-
nal AOD product is spatially continuous and available even
rent observational systems. Surface-based observations are
for locations where GOCI retrievals are not available (e.g.,
toosparsetoconstrainthesystem,whilecurrentsatellitesare
cloud-contaminatedpoints,sun-glintareas,desert,andsnow-
unabletoretrieveaerosolconcentrations.Instead,satellitere-
covered land). The use of GOCI-retrievals overcomes the
trievalsofAODareusedtohelpconstrainmodeledAODand
“temporal limitation” typical in analyses with LEO (low
themodeledrelationshipbetweenAODandsurface-levelPM
Earthorbit)aerosoldataandtheassimilationovercomesthe
(particulatematter)canthenbeusedtoestimatesurface-level
“spatial limitation” that has been typical in analyses based
PMconcentrations(WangandChristopher,2003;Liuetal.,
onlyonsatellitedata.TheassimilatedAODdatasetusingthe
2005;vanDonkelaaretal.,2006;Schaapetal.,2009;Tsaiet
GOCIdataisutilizedtomonitorandevaluatetransboundary
al.,2011).Thisapproachiscurrentlylimitedbythefactthat
particulate pollution from China to the Korean peninsula in
AOD from polar-orbiting satellites produce retrievals only
thisstudy.
once per day (Kaufman et al., 1997; Diner et al., 2001) and
currentgeostationarysatelliteslack“multi-spectral”aerosol
data (Knapp et al., 2002; Wang et al., 2003). For example,
Atmos.Chem.Phys.,14,659–674,2014 www.atmos-chem-phys.net/14/659/2014/
M.E.Parketal.:Transboundaryparticulatepollution 661
Table1.Briefdescriptionsofaerosolopticalpropertyretrievalsfromsomegeostationarysatellites.
Satellite& GMS51 MTSAT-1R2 MSG3 GOES4 COMS5 GOES-R GEO-KOMPSAT6
Sensor VISSR7 JAMI8 SEVIRI9 GOES-8 GOCI10 ABI11 GEMS12&ABI&
GOCI-2
Sensortype MET13sensor METsensor METsensor METsensor Oceancolorimager METsensor Environ.&ocean
colorimager
Temporalresolution 1h 30min 15min 15min 1h TBD17 1h(GEMS)
Spatialresolution a1.25km/b5km 1.25km/5km 1km/3km 1km/4km 0.5km/0.5km 0.5km/2km 0.25km/1km
Numberofbands 1VIS14/3IR16 1VIS/4IR 2VIS/2NIR15/ 1VIS/4IR 6VIS/2NIR 2VIS/4NIR/ 13VIS/NIR
8IR 10IR
Coveringregion EastAsia EastAsia Central Eastern Northeast US East,Southeast
&westernPacificOcean &westernPacificOcean Europe US Asia &CentralAsia
Launchyear Mar1995 Feb2005 Aug2002 Apr1994 Jun2010 Scheduledtobe Scheduledtobe
launchedin2015 launchedin2018
References Masudaetal.(2002); Kimetal.(2008) Poppetal.(2007) Christopheretal.(2002); Leeetal.(2010); Laszloetal.(2008)
Wangetal.(2003) Knappetal.(2002,2005) thisstudy
1GMS:GeostationaryMeteorologicalSatellite,2MTSAT-1R:Multi-functionalTransportSatellite,3MSG:MeteosatSecondGeneration,4GOES:GeostationaryOperationalEnvironmentalSatellite,5COMS:Communication,Ocean,andMeteorological
Satellite,6GEO-KOMPSAT:GeostationaryEarthOrbitKOreaMulti-PurposeSATellite,7VISSR:VisibleandInfraRedSpinScanRadiometer,8JAMI:JapaneseAdvancedMeteorologicalImager,9SEVIRI:SpinningEnhancedVisibleandInfraRedImager,
10GOCI:GeostationaryOceanColorImager,11ABI:AdvancedBaselineImager,12GEMS:GeostationaryEnvironmentMonitoringSpectrometer,13MET:Meteorological,14VIS:VISible,15NIR:NearInfraRed,16IR:InfraRed,17TBD:ToBe
Determined,aResolutionforthevisiblechannels,bResolutionfortheinfraredchannels.
2.1 Meteorologyandchemistry-transportmodel
simulations N CMAQ domain
0
5
ThePSU/NCARMM5andUSEPAModel-3/CMAQmodel GOCI domain
v4.5.1 were respectively used as the meteorological model N (A)
(MM) and chemistry-transport model (CTM) in this study 40 (cid:36)(cid:67)(cid:71)(cid:80)(cid:73)(cid:80)(cid:91)(cid:71)(cid:81)(cid:80)(cid:73) (B)
(cid:59)(cid:81)(cid:80)(cid:85)(cid:71)(cid:75)(cid:65)(cid:55)(cid:80)(cid:75)(cid:88)
(cid:59)(cid:71)(cid:78)(cid:78)(cid:81)(cid:89) (C)
(Byun and Schere, 2006; Park et al., 2011). The accuracy (cid:41)(cid:53)(cid:89)(cid:71)(cid:67)(cid:67)(cid:80)(cid:73)(cid:76)(cid:87)(cid:65)(cid:41)(cid:43)(cid:53)(cid:54) (cid:49)(cid:85)(cid:67)(cid:77)(cid:67)
(cid:41)(cid:81)(cid:85)(cid:67)(cid:80)(cid:65)(cid:53)(cid:48)(cid:55)
of the MM5 simulations was enhanced by nudging with 30N (cid:37)(cid:74)(cid:39)(cid:75)(cid:80)(cid:67)(cid:67)(cid:85)(cid:86)(cid:53)(cid:71)(cid:67)
the observed meteorological data (Grell et al., 1995). Ini-
tial conditions (ICs) and boundary conditions (BCs) for the
MM5 simulations were updated continuously with the Na- 0N
2
tionalCentersforEnvironmentalPrediction(NCEP)reanal-
ysis data. The latest remotely sensed surface wind data re- 90E 100E 110E 120E 130E 140E 150E
trievedfromtheAdvancedSCATterometer(ASCAT),which
was launched on 19 October 2006 onboard the Meteoro- N
2
logical Operational satellite A (MetOp-A) was used for 4
(A)
the four-dimensional data assimilation (FDDA), since dur-
(cid:36)(cid:67)(cid:71)(cid:80)(cid:73)(cid:80)(cid:91)(cid:71)(cid:81)(cid:80)(cid:73)
N
ing regional-scale, long-range transport events the signif- 38 (B)
(cid:59)(cid:71)(cid:78)(cid:78)(cid:81)(cid:89) (cid:59)(cid:81)(cid:80)(cid:85)(cid:71)(cid:75)(cid:65)(cid:55)(cid:80)(cid:75)(cid:88)
icant fractions of air pollutants are frequently transported (cid:53)(cid:71)(cid:67) (C)
throughthemarineboundarylayer(MBL)betweencountries N (cid:41)(cid:89)(cid:67)(cid:80)(cid:73)(cid:76)(cid:87)(cid:65)(cid:41)(cid:43)(cid:53)(cid:54)
4
inNortheastAsia(e.g.,throughtheMBLovertheYellowSea 3
(cid:41)(cid:81)(cid:85)(cid:67)(cid:80)(cid:65)(cid:53)(cid:48)(cid:55)
betweenChinaandKoreaandthroughtheMBLovertheEast
124E 128E 132E
ChinaSeabetweenChinaandJapan;regardingthelocations
of the Yellow Sea and the East China Sea, refer to Fig. 1) Fig. 1. Domains of CMAQ model simulations (in blue) and
COMS/GOCIsensor(inred).AlsoshownarefiveAERONETsites
(Daumetal.,1996;Rajeevetal.,2000;Songetal.,2007).
inthereceptorareas(KoreaandJapan).Forthesakeofanalytical
The schemes chosen for CMAQ model simulations were
convenience,threesubregions(A),(B),and(C)intheKoreanpenin-
the same as those used in previous research (Park et al.,
sulaaredefined;theserepresentthenorthern,central,andsouthern
2011). The SAPRC-99 chemical mechanism and AERO4
partsoftheKoreanpeninsula,respectively.
aerosol module were selected for simulating gas-phase
chemistryandatmosphericaerosolthermodynamic/dynamic
processes,respectively.MoredetailsoftheMM5andCMAQ
model simulations were described in a previous publication tionalCenterforAtmosphericResearch)FINN(FireInven-
(Parketal.,2011). tory from NCAR) biomass-burning (BB) emissions (Wied-
TheCMAQmodelsimulationswerecarriedoutusingthe inmyeretal.,2011).TheeffortsaresummarizedinTable2.
best available emission inventories over East Asia that have Fortheconsiderationofanthropogenicpollutantsemissions,
been made more sophisticated with, for example, the appli- suchasNO ,SO ,CO,blackcarbon(BC),NMVOCs,etc.,
x 2
cations of monthly variations of anthropogenic emissions, Intercontinental Chemical Transport Experiment-Phase B
the uses of an operational dust forecast model at the KMA (INTEX-B), Regional Emission inventory in Asia (REAS),
(KoreaMeteorologicalAdministration)andtheNCAR(Na- andCleanAirPolicySupportSystem(CAPSS)wereapplied
www.atmos-chem-phys.net/14/659/2014/ Atmos.Chem.Phys.,14,659–674,2014
662 M.E.Parketal.:Transboundaryparticulatepollution
as the best suitable emission data over China, Japan, and 2.2 Remote-sensingdata
theKoreanpeninsula,respectively.Theseemissioninvento-
ries originally adjusted to the year 2006 were updated for 2.2.1 COMS/GOCIdata
the year 2011 modeling period, using the projection fac-
torsbasedontheGreenhousegasandAirpollutionINterac- SensorsonboardLEOsatellites,suchasMODIS(Moderate
tionsandSynergies(GAINS)modelingemissionsscenarios Resolution Imaging Spectroradiometer) and MISR (Multi-
(http://gains.iiasa.ac.at/index.php/gains-asia). In the mean- angleImagingSpectroradiometer),haveordinarilybeenuti-
while,inordertoconsiderisopreneandmonoterpeneemis- lized to obtain information on aerosol optical properties
sionsoverthemodelingdomain,theisopreneemissionsfrom (AOPs), but their monitoring has taken place at most once
Model of Emissions of Gases and Aerosols from Nature per day. To more accurately monitor the transboundary
(MEGAN) and MOdel for Hydrocarbon emissions by the LRT events of air pollutants, hourly (or highly temporally
CANopy(MOHYCAN)wereutilizedfortheCMAQmodel- resolved) monitoring data from geostationary Earth orbit
inginputdata(Mülleretal.,2008).Monoterpeneemissions (GEO) sensors are needed instead of the temporally lim-
were also estimated from the isoprene emissions, using the ited LEO measured data. As mentioned previously and also
ratiosofmonoterpenetoisoprenefromGlobalEmissionsIn- presented in Table 1, MSG, GOES, and MTSAT-1R sen-
ventoryActivity(GEIA). sors have been partly monitoring the AOPs over Europe,
In particular, this modeling period (a spring season in US, and Japan, respectively (Knapp et al., 2005; Popp et
Northeast Asia)could not be explained without considering al., 2007; Kim et al., 2008), but these meteorological sen-
the effects of BB and mineral dust emissions. FINNv1 is a sorsareequippedwithonlyonevisiblechannel,whichpos-
BBemissioninventoryannuallyprovidedfromtheyear2002 sibly makes the quality of the AOPs from these meteoro-
by NCAR (http://bai.acd.ucar.edu/Data/fire/). This FINNv1 logical sensors less accurate. In contrast, GOCI sensor on-
has produced the BB emission data with a high horizontal board geostationary COMS has six visible bands at 412,
resolutionof∼1km×1kmovertheentireglobe,beingap- 443, 490, 555, 660, 680 nm, which can make advances by
pliedforvariableregionsfromlocaltoglobalscales(Wied- more accurately calculating the AOPs, compared to other
inmyeretal.,2011).Inthisstudy,theFINNv1wasusedfor meteorological GEO satellite sensors. In addition, atmo-
the BB emissions. The CMAQ v4.5.1 model does not in- spheric compensations, such as cloud-screening make the
clude any dust modeling modules related to occurrence and numerical accuracy of the AOPs more improved, because
transportofmineraldustevents.However,withouttakingthe the COMS/GOCI has 2 more bands at 745nm and 865nm
influence of mineral dust events into account, accurate esti- (i.e.,nearIR).TheGOCIsensorcoversNortheastChina,the
mationoftheaerosolseffectsobservedbylong-rangetrans- Korean peninsula, and Japan, and the products can be gen-
portsintheKoreanpeninsulaandJapanwouldnotbepossi- erated with 500m×500m spatial resolution. GOCI level 1
ble. Therefore, the outbreak and transport of Asian mineral dataduringdaytimecanbeobtainedeighttimesfrom09:00
dustwereconsideredbyAsianDustAerosolModel(ADAM) to16:00LST(localstandardtime)(refertohttp://kosc.kordi.
simulations. The ADAM is currently operational model for re.kr/oceansatellite/coms-goci/introduction.kosc).
forecasting Asian mineral dust events at the KMA and the ThehourlyGOCIlevel1dataisreleasedsomewhatslowly
details of the ADAM simulations were reported previously after the IOT and data calibration procedures. The GOCI
bye.g.,ParkandLee(2004),Kimetal.(2011),andParket level 1 radiance is then converted into level 2 AOD data
al.(2011). using a multi-spectral algorithm being developed by Jhoon
ThedomainoftheCMAQmodelsimulationisalsoshown Kim and coworkers at Yonsei University, Korea (Lee et al.,
in Fig. 1. The CMAQ domain (blue color) covered the re- 2010). The level 2 AOD data retrieved with a pixel size of
gionofapproximately95–145◦E,20–50◦N,whichincludes 6km×6km was used in this study. The retrieval algorithm
COMS/GOCI domain (red color) composed of Northeast was originally based on the NASA MODIS (i.e., LEO) al-
China, the Korean peninsula, and Japan. In this study, the gorithm, but it has evolved to date to include the aerosol
CMAQ domain was chosen to be larger than the GOCI modelsoptimizedforEastAsiaandtoconsiderthegeometry
domain, since the air quality inside the GOCI domain is fromGEOsatelliteinstrumentssuchasGOCI,GOCI-2,and
stronglyinfluencedbytheinfluxesfromtheouterareaswith GEMS(GeostationaryEnvironmentMonitoringSpectrome-
persistent westerlies in East Asia. The horizontal grid spac- ter) (both the GOCI-2 and GEMS sensors are scheduled to
ing of the CMAQ modeling was 30km×30km and the belaunchedin2018).TheYonseiaerosolretrievalalgorithm
vertical domain was distributed in 14 layers following σ- includesthecapabilitytoclassifyaerosoltypesbasedonab-
coordinatefrom1000hPato70hPa.TheMM5-CMAQsim- sorptiontestsandsizeinformation(Kimetal.,2007;Leeet
ulation period started at 00:00UTC, 1 April to 23:00UTC, al.,2010).TheevaluationofthecurrentAODdatafromthe
31 May 2011 (with a week-long CMAQ spin-up period), algorithm shows that the slope of the linear regression line
since the first batch of the level1 radiance data from the is close to 1.03 over ocean, but 1.15 over land when com-
COMS/GOCIdatacenterwasreleasedforthisperiod. pared to values from AERONET. Although the detailed ex-
planation and validation procedures of this Yonsei aerosol
Atmos.Chem.Phys.,14,659–674,2014 www.atmos-chem-phys.net/14/659/2014/
M.E.Parketal.:Transboundaryparticulatepollution 663
Table2.EmissioninventoriesusedintheCMAQmodelsimulations.
Region Emissioninventory Species
ChinaandNorthKorea INTEX-B1 SO2,NOx,NMVOCs,CO,CO2,BC,OC
EDGAR2 NH3
SouthKorea CAPSS3 SO2,NOx,NMVOCs,CO,NH3
INTEX-B CO2,BC,OC,ChemicalsplitfactorsofNMVOCs
Japan REAS4 SO2,NOx,NMVOCs,CO,CO2,NH3,BC,OC
Entiredomain GAINS5emissionscenarios aInterannualvariationsofCO,NOx,VOC,SO2,NH3,CO2,CH4,PM2.5
FINN6fireemissions CO,NO,NO2,SO2,NH3,NMVOCs,BC,OC,PM2.5,PM10
MEGAN7+MOHYCAN8 Isoprene
GEIA9 bRatiosofmonoterpenetoisoprene
1INTEX-B:IntercontinentalChemicalTransportExperiment-PhaseB,2EDGAR:EmissionDatabaseforGlobalAtmosphericResearch,3CAPSS:CleanAirPolicySupport
System,4REAS:RegionalEmissioninventoryinASia,5GAINS:GreenhousegasandAirpollutionINteractionsandSynergies,6FINN:FireInventoryfromNCAR,
7MEGAN:ModelofEmissionsofGasesandAerosolsfromNature,8MOHYCAN:MOdelforHydrocarbonemissionsbytheCANopy,9GEIA:GlobalEmissionsInventory
Activity,aEmissionincrease(ordecrease)ratesfrom2006to2011fortheairpollutantswereobtainedfromtheGAINSemissionscenarios,bMonoterpeneemissionsarenot
includedinMEGAN+MOHYCANemissions.Therefore,theratiosofmonoterpenetoisopreneinthebiogenicemissionswereobtainedfromtheGEIAemissioninventory.
retrievalalgorithmweregiveninpreviouspublications(Lee
et al., 2010, 2012), a brief introduction on the algorithm is
madehere.TheretrievalsoftheAOPswereperformedwith
(a) MODIS-retrieved AOD (b) GOCI-retrieved AOD
topofatmospherespectralreflectances.ToobtaintheAOPs
120E 130E 140E 120E 130E 140E
over the regions including clouds, clouds are screened by N N N
spatialvariabilityandthresholdtestsateachpixel.Evenso, 00 45 45 45
:
there are limitations to the complete removal of clouds and 0
1 N N N
theperfectretrievalofsurfacereflectanceduetothelackof 0 35 35 35
1
IRbandsonGOCI.Therefore,theYonseiaerosolretrievalal- 4/
0 N N N
gorithmeliminatescloudsutilizingameteorologicalimager 25 25 25
(MI) that has 4 IR channels. The MI sensor is also onboard N N N
0 5 5 5
COMS with GOCI. In addition, the Yonsei aerosol retrieval 0 4 4 4
:
3
algorithm is composed of two different retrieval logics that 1 N N N
are applied to “continent” including near-coastal turbid wa- 10 35 35 35
terareasand“ocean”,respectively. 4/
0 N N N
Figure 2 presents one example: a comparison of the 25 25 25
Terra/MODIS-retrieved and Aqua/MODIS-retrieved AODs N N N
withGOCI-retrievedAODsatapproximately10:00LSTand 00 45 45 45
:
0
13:00LSTon10Apriland12April,respectively.Asshown 1 N N N
in Fig. 2, there are agreements in the spatial AOD distri- 12 35 35 35
butions from the two MODIS and GOCI sensors. As dis- 04/ N N N
5 5 5
cussedpreviously,atmosttwosnapshotsofinformationper 2 2 2
day from the LEO sensors (i.e., MODIS) can be obtained, N N N
0 5 5 5
whereas eight snapshots per day from the GEO sensor (i.e., 0 4 4 4
:
3
GOCI) are available. This is an obvious “temporal” advan- 1 N N N
tagethatcanbeachievedfromtheGEOsensor.Furthermore, 12 35 35 35
4/
as shown in Fig. 2, the AOD data from the GOCI sensor 0 N N N
5 5 5
showsabetterspatialcoverage,comparedtothosefromthe 2 2 2
120E 130E 140E 120E 130E 140E
MODIS(seewhitebandsalongthenorth–southswathinthe
MODIS data). This lack of the MODIS data was caused by 0.0 0.3 0.6 0.9 1.2 1.5
sun-glintovertheoceanareas(detailsaboutthesun-glintef-
Fig. 2. Snapshots of (a) Terra and Aqua/MODIS-retrieved AODs
fectsarediscussedfurtherinAppendixA).Bycontrast,such
and (b) GOCI-retrieved AOD over Northeast Asia at 10:00LST
lack of the data is not found in the GOCI data. This is an-
(Terra/MODIS) and 13:00 LST (Aqua/MODIS) on 10 April and
other“spatial”advantagethatcanbeachievedwiththeGEO
12April.
sensor.
www.atmos-chem-phys.net/14/659/2014/ Atmos.Chem.Phys.,14,659–674,2014
664 M.E.Parketal.:Transboundaryparticulatepollution
2.2.2 AERONET particulate species i at the size bin or particle mode j, re-
spectively. The terms related to the MEEs (βij) and the hy-
Manystudieshaveusedtheground-basedNASAAERONET groscopicity(fij(RH))havebeendefineddifferentlyinsev-
datatoverifymodelingresultsandsatellitemonitoringdata eralstudies,e.g.,d’Almeidaetal.(1991),Malmetal.(1994),
(e.g., Chin et al., 2002, 2004; Yumimoto and Takemura, Chin et al. (2002), Malm and Hand (2007), and Park et
2011). The AERONET data measured by sun photometers al.(2011).Accordingtothepreviousinvestigations,theway
were utilized in this study to investigate the accuracy of tocomputetheextinctioncoefficients(σ )hasbeenmainly
ext
CMAQmodelsimulationsandGOCIAOPdata. classifiedintotwocategories.Thefirstmethodistouseem-
Inthismanuscript,thedataobtainedfromfiveAERONET pirical equations (e.g., Malm et al., 1994; Malm and Hand,
observation sites (i.e., Baengnyeong, Yonsei Univer- 2007;Songetal.,2008;Parketal.,2011).Theothermethod
sity (Yonsei_Univ), GIST (Gwangju_GIST), Gosan istousetheequationsobtainedbasedontheMietheory(e.g.,
(Gosan_SNU), and Osaka) in the Korean peninsula and d’Almeida et al., 1991; Chin et al., 2002). In this study, the
Japanwereselectedinordertoanalyzelong-rangetransport methodsfromChinetal.(2002)andParketal.(2011)were
effectsoverNortheastAsia.TheAERONETAODlevel2.0 mainlyutilizedandcompared.
data used for comparison was obtained from the official Based on the particulate composition and meteorolog-
NASA web site at http://aeronet.gsfc.nasa.gov. The only ical parameters obtained from the MM5-CMAQ model
exception was the AERONET AOD data for the Yonsei simulations, the AOD was calculated at a wavelength of
University (Seoul) site, which were not calibrated. Before 550nm (AOD ), using the conversion algorithms of Chin
550
use,theywereupgradedbyeliminatingsomesporadiccloud etal.(2002),Parketal.(2011),andothers(d’Almeidaetal.,
contamination of the initial high AOD data (AOD>1.0) 1991;Malmetal.,1994;Hessetal.,1998;MalmandHand,
using the meteorological data from the automatic weather 2007):
stations (AWS) colocated with Cimel sun photometers. (cid:5)
z
The uncertainty and possible anomalies in the Cimel sun AOD = σ (550nm)dz, (4)
550 ext
photometerAODdataattheYonseiUniversitysitewerealso 0
checked by the Laboratory for Terrestrial Physics at NASA where z represents the vertical height. For example, the
Goddard Space Flight Center. AOD uncertainties of 0.01 to CMAQ-simulated AOD products calculated from Park et
0.04wereestimated,mainlyduetocalibration. al. (2011) usually showed underestimations when com-
pared to the AERONET AOD data. However, when Chin
2.3 AODcalculationsfromCMAQmodelsimulations
et al.’s (2002) conversion algorithm was applied, the AOD
values were usually larger than those calculated by Park et
Concentrations of nitrate, sulfate, ammonium, organic
al.’s (2011) method. These results will be discussed further
aerosols (OAs), BC, sea-salt aerosols, and mineral dust are
inalatersection.
provided as results of the CMAQ model simulations. This
aerosolcompositioncanbeconvertedintotheAOPsinclud-
2.4 Dataassimilation
ing AOD by considering the mass extinction efficiencies
(MEEs) depending on the scattering and absorbing charac- In several previous studies, optimal interpolation (OI) tech-
teristicsforeachaerosolcompositionandthehygroscopicity. niquewithaKalmangainmatrixhasbeenemployedtoim-
First, the absorption, scattering, and extinction coefficients provethepredictionsofAODandaerosolmassdistributions
(σabs, σscat, and σext) were calculated by the following for- with the combination of the modeling results and monitor-
mulas: ing data (Collins et al., 2001; Yu et al., 2003; Adhikary et
al., 2008; Chung et al., 2010; Park et al., 2011). For exam-
σ (λ)=σ (λ)+σ (λ), (1)
ext abs scat ple, Adhikary et al. (2008) showed some enhancements in
the estimations of the AOD values over Asia by combin-
ing the STEM-modeled and MODIS-retrieved data through
(cid:2) (cid:3) (cid:4) (cid:4)
σabs λ:Mm−1 = ni mj {(1−ωij(λ))·βij(λ)·fij(RH)·[C]ij}, (2) theOItechnique.TheaccuracyoftheAODvalueswasalso
greatlyimprovedaftertheapplicationoftheOImethodusing
CMAQ-simulated and MODIS-retrieved AOD data (AOD
m
σscat(cid:2)λ:Mm−1(cid:3)=(cid:4)ni(cid:4)mj {ωij(λ)·βij(λ)·fij(RH)·[C]ij}, (3) AanOdDA(cid:4)O=DoAiOnDEq.+5)K(P[AarOkDeta−l.,H2(0A1O1)D. )], (5)
where i and j represent the particulate species i at the size m m o m
bin or particle mode j; fij(RH) the hygroscopic enhance-
ment factor as a function of relative humidity (RH); [C]ij K=BHT(HBHT+O)−1, (6)
theconcentrationsofparticulatespeciessuchas(NH ) SO ,
4 2 4
NH NO ,BC,OAs,mineraldust,andsea-saltaerosols;and whereHdenotesalinearoperatorforinterpolationfromthe
4 3
ωij(λ) and βij(λ) the single scattering albedo and MEE of modelinggridtothesatelliteretrievalgrid.BandOrepresent
Atmos.Chem.Phys.,14,659–674,2014 www.atmos-chem-phys.net/14/659/2014/
M.E.Parketal.:Transboundaryparticulatepollution 665
theerrorcovariancesofthemodelingandsatellite(observa-
(a) CMAQ AOD (b) GOCI AOD (c) Assimilated AOD
tion)fields,respectively. 120E 130E 140E 120E 130E 140E 120E 130E 140E
N N N N
In this study, the OI technique was again utilized for the 00 45 45 45 45
assimilationoftheAODdatafromtheGOCIsensorintothe 09: N N N N
CobMtaAinQedmboydethl.eTchoevaKriaalnmceanofgaminodmelaetdrixan(dKoibnseErvqe.d6)fiewldass 04/10 25N35 25N35 25N35 25N35
(B and O in Eq. 6) which is a function of free parameters. 00 45N 45N 45N 45N
Terhael pyreeavriloyufisxsetdudfireese(pCaorallminesteertsawl.,er2e0m01a;inYluy uetseadl.,in20s0ev3-; 4/10 11: 35N 35N 35N 35N
0 N N N N
Adhikary et al., 2008; Chung et al., 2010), whereas Park et 25 25 25 25
al. (2011) proposed applying seasonal variations of free pa- 00 45N 45N 45N 45N
rameters.Thebestfreeparametersforfourseasonssuggested 3:
1 N N N N
in Park et al. (2011) were used here. Detailed explanations 10 35 35 35 35
4/
of the OI technique including the selection of the free pa- 0 25N 25N 25N 25N
rameters were made in Park et al. (2011). The assimilated N N N N
00 45 45 45 45
AODdataproducedbycombiningtheCMAQ-simulatedand 5:
1 N N N N
theGOCI-retrievedAODswereusedtoinvestigatethelong- 10 35 35 35 35
4/
rangePMtransporteventsinthisstudy. 0 N N N N
25 25 25 25
120E 130E 140E 120E 130E 140E 120E 130E 140E
2.5 Long-rangetransportofparticulatematter 0.0 0.3 0.6 0.9 1.2 1.5
Fig.3.Spatialandtemporaldistributionsof(a)CMAQ-simulated
TransboundarytransportsoftheAODplumesfromtheChina
AOD (calculated via Chin et al.’s (2002) conversion algorithm),
continent to the Korean peninsula and Japan were the fo- (b)GOCI-retrievedAOD(retrievedviatheYonseiaerosolretrieval
cus of this study. However, defining the LRT events over algorithm (Lee et al., 2010)) and (c) assimilated AOD with two-
Northeast Asia is not easy, mainly due to complexities in hour resolution over Northeast Asia at 09:00, 11:00, 13:00, and
airmassmovements.Weconsideredlarge-scaleAODplumes 15:00LSTon10April2011.
thatcausedthepeakAODvalueslargerthan0.4attherecep-
tor sites (“peak AOD filter”). Such LRT AOD plumes were
Engel-Coxetal.,2004;Liuetal.,2005;vanDonkelaaretal.,
further traced back via NOAA HYSPLIT (HYbrid Single-
2006;Choietal.,2009;Schaapetal.,2009;Tsaietal.,2011).
Particle Lagrangian Integrated Trajectory) model analysis
Several studies have estimated two-dimensional PM from
(Draxler and Hess, 1998) from the points of time when the 10
the relationships between ground-based PM and satellite-
peak AOD values occurred. The HYSPLIT model simula- 10
retrieved AOD (Wang and Christopher, 2003; Engel-Cox et
tionswereperformedwithmeteorologicaldata,whichwere
al., 2004; Liu et al., 2005; Schaap et al., 2009; Tsai et al.,
takenfromtheGDAS(GlobalDataAssimilationSystem)of
2011)ortheratiosofCTM-simulatedPM toAOD(Liuet
NCEP. When the backward trajectories from the occurring 10
al.,2004;vanDonkelaaretal.,2006;Choietal.,2009).
points of the peak AOD values had passed through China
within48–72h,theAODplumeswerefinallydeterminedto
be LRT plumes (“backward trajectory filter”). A LRT case
3 Resultsanddiscussions
shown in Fig. 3 was decided by using both of the filters.
Forexample,72hHYSPLITbackwardtrajectoryanalysisat
The hourly assimilated AOD data was used to identify and
three AERONET receptor sites in Korea for the LRT case
analyze the PM LRT events over Northeast Asia. Based on
in Fig. 3 is presented in Fig. 4. As shown, all the backward
this,thevariationsoftheAODvaluesduringLRTeventsin
trajectories were passed through China. By contrast, Fig. 5
comparisonwiththoseduringnon-LRTeventswereanalyzed
showsanexampleofanon-LRTcase.BoththeLRTandnon-
onamorequantitativebasis.
LRTcaseswillbediscussedfurtherinSect.3.1.
In order to investigate the LRT of particulate matter over 3.1 Long-rangetransportofaerosolsfromChinato
Northeast Asia, the AOD values instead of PM concentra- Korea
tionswereused.Inotherwords,theAODhasbeenregarded
asaproxyquantityforPMconcentrationinthisstudy.Inthis A representative snapshot of an LRT event of particulate
context, relationships between PM10 (or PM2.5) and AOD matter from China to the Korean peninsula, Japan, and the
can be important. The relationships are affected by various northwestern Pacific Ocean is presented in Fig. 3. It can
parameters, such as height of the PBL (planetary boundary be seen that a huge AOD plume (AOD>0.6) was gener-
layer), RH, vertical distribution of aerosols, seasonality, di- ated in the greater Beijing area and Shandong province on
urnalvariation, geographical information,particlecomposi- 10Aprilandwasthentransportedoverlongdistancestothe
tions, and size distributions (Wang and Christopher, 2003; Korean peninsula following a low pressure trough. GOCI
www.atmos-chem-phys.net/14/659/2014/ Atmos.Chem.Phys.,14,659–674,2014
666 M.E.Parketal.:Transboundaryparticulatepollution
Backward trajectories (a) CMAQ AOD (b) GOCI AOD (c) Assimilated AOD
120E 130E 140E 120E 130E 140E 120E 130E 140E
120E 130E 140E 00 45N 45N 45N 45N
9:00 45N 04/12 09: 25N35N 25N35N 25N35N 25N35N
0 N N N N N
10 35 1:00 45 45 45 45
04/ 4/12 1 35N 35N 35N 35N
N 0 N N N N
5 25 25 25 25
2
N N N N
5N 13:00 45N 45N 45N 45N
0 4 12 35 35 35 35
0 4/
11: N 0 25N 25N 25N 25N
4/10 35 15:00 45NN 45NN 45NN 45NN
0 12 35 35 35 35
N 4/
25 0 25N 25N 25N 25N
120E 130E 140E 120E 130E 140E 120E 130E 140E
N 0.0 0.3 0.6 0.9 1.2 1.5
5
4
0
0 Fig.5.AsFig.3,foranon-LRTeventon12April2011.Thiscase
:
3 was determined as non-LRT because the main AOD plumes were
1 N
0 5 transportedintoovernorthwesternPacificOcean,notovertheKo-
1 3
4/ reanpeninsulaand/orJapan.
0
N
5
2
N ilated AOD fields are much improved in comparison to the
5
4 predicted (a priori) CMAQ AOD distributions (first column
0
0 ofFig.3).TheaprioriCMAQAODvalueshaveatendency
:
5
1 N tobeunderestimatedcomparedtotheGOCI-retrievedAOD
10 35 values,asshowninFig.3.Thisunderestimationisbelieved
4/ toberelatedtouncertaintiesintheemissionratesofgaseous
0
N andparticulatepollutantsandtheCMAQmodelsimulations,
5
2
as well as the uncertainties in the assumptions made during
120E 130E 140E
theAODestimationfromtheCMAQmodelingresults.
Fig.4.NOAAHYSPLIT72hbackwardtrajectoryanalysisatthree In this analysis, the AOD calculation via Chin et
KoreanAERONETsites(Baengnyeong,Yonsei_Univ(Seoul),and al.’s (2002) conversion method was used. However, sev-
Gwangju_GIST) with 2h resolution at 09:00, 11:00, 13:00, and eral other estimations were also tested (e.g., Malm et al.,
15:00LSTon10April2011. 1994; Chin et al., 2004; Park et al., 2011). One example
is shown in Fig. 6 and Table 3, which were produced by
Parketal.’s(2011)conversionmethod.Itwasfoundthatthe
retrievals with one-hour intervals from 09:00 to 16:00LST AOD values were underestimated because the MEE values
(eightsnapshotsperday)on10Aprilshowhighaerosolload- and hygroscopic enhancement factor (fij(RH)) of Park et
ings(AOD>1.2)overtheShandongpeninsulaandShanghai al.’s (2011) conversion method were reported to be in gen-
regions that extend over Korea. The plumes extend farther eralsmallerthanthoseofChinetal.’s(2002)method.After
outovertheYellowSeaandslowlyshifttolowerlatitudesas assimilationboththemagnitudesandspatialdistributionsare
time progresses. In Fig. 3, only two-hour resolution results significantlyimproved.TheAODLRTeventshowninFig.3
are presented. It is important to note that GOCI AOD val- isonlyanexample.Seventofourteensmall-andlarge-scale
uesarenotretrievedinregionswherecloudsexist.Asshown AODLRTeventsoccurredoverthetwo-monthperiod(from
inFig.3,cloud-contaminatedpixelsappearwhite,whichin- 1Aprilto31May2011).
dicates that no data is available at these pixels. The GOCI Furthermore, the non-LRT events were also identified.
retrievals,whenassimilatedintotheCMAQmodel,produce During the non-LRT periods, air masses were generally
completespatialdistributionswhereAODisestimatedinall stationary.Althoughtheinfluxofairpollutantsemittedfrom
pixels(seethethirdcolumnofFig.3).Therefore,theassim- the highly polluted China regions to the Korean peninsula
Atmos.Chem.Phys.,14,659–674,2014 www.atmos-chem-phys.net/14/659/2014/
M.E.Parketal.:Transboundaryparticulatepollution 667
Table3.ComparisonanalysisoftheassimilatedAODforLRTandnon-LRTeventsatfiveAERONETsitesduringtheentiremodelingperiod
of1Aprilto31May2011.
Non-LRTevents LRTevents Datafractionof AODincrease RatioofAODincrease
LRTevents duringLRT duringLRT
(xLRTinEq.8) events3 events4
N1 σ2 Avg. Max. N σ Avg. Max.
Baengnyeong 849 0.11 0.16 0.95 606 0.36 0.59 2.15 0.42 0.43 2.59
Yonsei_Univ(Seoul) 821 0.15 0.22 0.77 634 0.29 0.49 2.13 0.44 0.28 1.28
Gwangju_GIST 1024 0.13 0.20 1.10 431 0.31 0.47 1.89 0.30 0.27 1.36
Gosan_SNU 967 0.12 0.17 1.09 488 0.24 0.41 1.53 0.34 0.24 1.37
Osaka 1247 0.11 0.16 1.16 208 0.21 0.39 1.23 0.14 0.23 1.41
Total 0.12 0.18 1.16 0.31 0.49 2.15 0.33 0.31 1.68
ThecalculationswereconductedwiththeassimilatedAODproducts.TheAODdataassimilationswerecarriedoutwiththeGOCI-retrievedAODproducts(retrievedviatheYonseialgorithm(Leeet
al.,2010))andtheCMAQ-simulatedAODproducts(calculatedviaParketal.’sconversionalgorithm(Parketal.,2011)).
1Thenumberofdatafornon-LRTorLRTeventsfortheentiremodelingperiod;
2standarddeviation;
3AODincreasedbyLRTeventsabovetheaveragebackgroundAODcalculatedduringnon-LRTperiods;
4theratioswereestimatedby((AODLRT−AODnon-LRT)/AODnon-LRT).Here,AODLRTandAODnon-LRTindicatetheaverageAODcalculatedduringtheLRTandnon-LRTevents,respectively.
Table4.ComparisonanalysisoftheassimilatedAODforLRTandnon-LRTeventsatfiveAERONETsitesduringtheentiremodelingperiod
of1Aprilto31May2011.
Non-LRTevents LRTevents Datafractionof AODincrease RatioofAODincrease
LRTevents duringLRT duringLRT
(xLRTinEq.8) events3 events4
N1 σ2 Avg. Max. N σ Avg. Max.
Baengnyeong 849 0.12 0.22 0.98 606 0.48 0.79 2.79 0.42 0.57 2.65
Yonsei_Univ(Seoul) 821 0.21 0.30 1.13 634 0.37 0.66 2.93 0.44 0.35 1.17
Gwangju_GIST 1024 0.16 0.27 1.22 431 0.40 0.63 2.50 0.30 0.36 1.30
Gosan_SNU 967 0.12 0.22 1.14 488 0.31 0.54 1.95 0.34 0.32 1.47
Osaka 1247 0.11 0.21 0.90 208 0.22 0.48 1.26 0.14 0.27 1.27
Total 0.15 0.24 1.22 0.40 0.65 2.93 0.33 0.40 1.67
ThecalculationswereconductedwiththeassimilatedAODproducts.TheAODdataassimilationswerecarriedoutwiththeGOCI-retrievedAODproducts(retrievedviatheYonseialgorithm(Leeet
al.,2010))andtheCMAQ-simulatedAODproducts(calculatedviaChinetal.’sconversionalgorithm(Chinetal.,2002)).
1Thenumberofdatafornon-LRTorLRTeventsfortheentiremodelingperiod;
2standarddeviation;
3AODincreasedbyLRTeventsabovetheaveragebackgroundAODcalculatedduringnon-LRTperiods;
4theratioswereestimatedby((AODLRT−AODnon-LRT)/AODnon-LRT).Here,AODLRTandAODnon-LRTindicatetheaverageAODcalculatedduringtheLRTandnon-LRTevents,respectively.
and Japan was significantly reduced, the stagnant weather caseswerealsointroducedbyAl-Saadietal.(2005)withthe
conditions could seriously deteriorate air quality. Figure 5 MODISAODdataovertheUnitedStates.Anotherpossibil-
shows a case of the non-LRT events. Noticeable aerosol ityforthisplumeisthataerosolscouldbegeneratedthrough
plumesdirectlytransportedfromChinatotheKoreanpenin- Siberian forest fires. In order to check this possibility, the
sula and Japan could not be identified. Some AOD plumes global MOZART (Model for OZone And Related chemi-
weregeneratedinthesouthernpartsofChina,andwerethen cal Tracers) model simulation data was investigated (http:
transported over the northwestern Pacific Ocean. This case //www.acd.ucar.edu/wrf-chem). But no Siberian fire signif-
wasclassifiedasnon-LRTcasesincetheAODplumeswere icantlyaffectedaerosolplumeson/before12April2011.
notdirectlytransportedtotheKoreanpeninsulaand/orJapan. The AOD LRT events were investigated at the five
In addition, a high AOD plume (in the second column AERONET sites (Holben et al., 1998) in the Korean penin-
in Fig. 5) was transported through southern Russia (be- sula and in Japan shown in Fig. 1. The AERONET AOD
tween09:00and11:00LST).Thisplumeappearstobethin values observed at ground sites are regarded as “ground
clouds (i.e., cloud optical depth (COD)). This can be con- truth”becausethesurfacereflectancedoesnotinterferewith
firmed partly via the Terra/MODIS AOD data at 10:30LST them (Holben et al., 1998; Dubovik et al., 2000; Chin et
on 12 April. The Terra/MODIS AOD data did not show al., 2002, 2004; Park et al., 2011), and due to the high ac-
any AOD plume over the same region (refer to the third curacy (∼0.01–0.02) of the fully calibrated data (Eck et
and the fourth rows in Fig. 2). Since the GOCI AOD algo- al., 1999). Figure 7 shows the comparison of the GOCI-
rithm has limitations in completely eliminating clouds, thin retrieved/assimilatedAODproducts(redhorizontalbarsand
clouds are sometimes considered aerosol plumes. Similar blue solid line in Fig. 7, respectively) with the AERONET
www.atmos-chem-phys.net/14/659/2014/ Atmos.Chem.Phys.,14,659–674,2014
668 M.E.Parketal.:Transboundaryparticulatepollution
timatedfromEq.(7)(greybars)inFig.8.Asshown,thereare
(a) CMAQ AOD (b) GOCI AOD (c) Assimilated AOD
120E 130E 140E 120E 130E 140E 120E 130E 140E reasonableagreementsbetweenthetwoPM concentrations
N N N N 10
00 45 45 45 45 over two-month periods, except for several time windows
09: N N N N such as 9–11 April and 13–17 May. This example (Fig. 8)
10 35 35 35 35
04/ 25N 25N 25N 25N issurpfarecseePntMedcohnerceenitnraotirodnesr.tAoltshhoouwghhroewasoAnOaDbleisagrreeleamtedenttos
N N N N
00 45 45 45 45 betweentheobservedandestimatedPMconcentrationswere
1:
4/10 1 35N 35N 35N 35N sdheovwelnopiendFinigo.r8d,erthteocmoonrveearscicounramteeltyhcoodnsvsehrotuAlOdDbetofuPrMth1e0r
0 25N 25N 25N 25N (or PM2.5) over East Asia, with the consideration of more
00 45N 45N 45N 45N factors,suchasheightofPBL,RH,andpresenceofdustand
13: N N N N BB plumes in the free troposphere (Wang and Christopher,
4/10 35 35 35 35 2003; Engel-Cox et al., 2004; Liu et al., 2005; van Donke-
0 N N N N
25 25 25 25 laaretal.,2006;Choietal.,2009;Schaapetal.,2009;Tsai
00 45N 45N 45N 45N etal.,2011).
5:
10 1 35N 35N 35N 35N 3.2 Estimationoftheimpactsoflong-rangetransport
4/
0 N N N N
25 25 25 25
120E 130E 140E 120E 130E 140E 120E 130E 140E Thegrey-shadedareasinFig.7representtheperiodsduring
0.0 0.3 0.6 0.9 1.2 1.5 which the receptors were affected by the PM LRT events.
ThePMLRTeventsweredeterminedbasedonthemeteoro-
Fig. 6. As Fig. 3, except that the CMAQ-simulated AOD values
logicalfieldsgeneratedfromMM5simulationstogetherwith
werecalculatedviaParketal.(2011)’sconversionalgorithm.
comprehensive backward trajectory analysis. For example,
the Baengnyeong site was affected by the small- and large-
AOD data (black bars in Fig. 7). The GOCI-retrieved AOD scaleLRTeventsfourteentimesoverthetwo-monthperiod.
valuesagreewellwiththeAERONETvalues(e.g.,R=0.84, Mostofthehigh(andwide)AODpeaksareduetothelarge-
see Table 5). However, both AERONET and GOCI data scale LRT events from China, and most of the small/sharp
show large time periods with no retrievals, which points spikes are usually caused by local pollution under stagnant
to the need for models to provide additional information. meteorologicalconditions.Thegrey-shadedperiodsinFig.7
The assimilated AOD product is well correlated with the usuallystartandendwhenthelarge-scaleLRTAODplumes
AERONET and GOCI-retrieved AOD values and is much arriveinanddepartfromtheAERONETsites.
more accurate than the a priori CMAQ AOD product when The background average AOD values (calculated exclud-
comparedwiththeAERONETAODvalues.Thissituationis ing the AOD values during the LRT events) are represented
summarizedinTable5,wheretheerrorsandbiasesafterthe by dashed lines at the five AERONET sites in Fig. 7. The
assimilationbecomemuchsmallerthanthosebeforetheas- average impact of the transboundary PM pollution over the
similation, indicating that the assimilation clearly improves entiretwo-monthperiod(i.e.,over“thenon-LRT+LRTpe-
the quality of the AOD products. The assimilated product riods”) was calculated using the following formula over the
providesacontinuousestimateofthetimeseriesofAODat two-monthperiod:
everylocation,whichtheobservationsalonecannot(thered (cid:8)
horizontalbarsinFig.7onlycoverlimitedtimeperiods). AODavg= Ni=1{AODLRT×xLRT+NAODnon-LRT×(1−xLRT)}, (8)
As discussed previously, the AOD values are related to
surface PM concentrations. Figure 8 shows an example of
how to estimate the surface PM concentrations from the
AOD =(cid:6)AOD=AOD −AOD , (9)
assimilated AOD values. The daily averaged AOD values inc LRT non-LRT
((AOD)assimilated)showninFig.7wereconvertedintoPM10 whereAODavg,AODnon-LRT,andAODLRT representtheav-
((PM10)estimated) at five AERONET sites, using following erage AOD values during the entire period and the average
proportionalrelationship(vanDonkelaaretal.,2006): AOD values calculated during the non-LRT and LRT peri-
(cid:6) (cid:7)
ods,respectively.x isthefractionoftheLRTeventsover
PM LRT
(PM10)estimated= 10 ×(AOD)assimilated. (7) thetwo-monthperiod.N isthetotalnumberofthegridcells
AOD
simulated being considered (for the calculation at an AERONET site,
Twenty-four hour PM was also measured simultaneously N =1).Afterthat,theaverageAODincrease(AOD )dur-
10 inc
at three Korean AERONET locations by the Korean Na- ingthetwo-monthperiodwascalculatedbyEq.(9).There-
tionalInstitute ofEnvironmental Research(NIER). Theob- sultsfortheanalysisfortheAERONETsitesaresummarized
served24hPM concentrations(blackdots)werethenplot- inTable4.Fromthisanalysis,theaverageAODincreaseof
10
tedtogetherwiththedailyaveragedPM concentrationses- 117–265%isfoundatthefiveAERONETsites.
10
Atmos.Chem.Phys.,14,659–674,2014 www.atmos-chem-phys.net/14/659/2014/